**Meet the editors**

Prof. Dr. Stéphane Molotchnikoff following graduation from State University of New York (Buffalo), he joined the faculty as professor of physiology at the Département de Sciences Biologiques de l'Université de Montréal. Currently, he is also adjunct professor at the Engineering School of University of Sherbrooke. For the past several decades Professor Molotchnikoff taught courses

in physiology from comparative physiology to neurosciences. Professor Molotchnikoff maintains an active laboratory and graduate research team that focuses on studying the mechanisms of the brain's visual system and more generally investigating the processes related to sensory functions in various vertebrate species. Presently two main themes are explored: adaptation-induced plasticity, and the modifications of connectomes induced by changes in the properties of visual targets. He was awarded the Purkynĕ medal from Charles University (Prague) and Chevalier de l'Ordre de la Francophonie et du Dialogue des Cultures de l'Assemblée Parlementaire de la Francophonie. His research has been continuously funded by Canadian research agencies.

Prof. Jean Rouat holds a master degree in Physics from Université de Bretagne, France (1981), an E. & E. master degree in speech coding and speech recognition from Université de Sherbrooke (1984) and an E. & E. Ph.D. in cognitive and statistical speech recognition jointly with Université de Sherbrooke and McGill University (1988). His post-doc has been with the Medical Research

Council, Applied Psychological Unit, Cambridge, UK and the Institute of Physiology, Lausanne, Switzerland. He is currently with Université de Sherbrooke where he founded the Computational Neuroscience and Intelligent Signal Processing Research group. He is also adjunct professor in the biological sciences department from Université de Montréal. His laboratory interests are in neurocomputational signal processing. He is an active member of scientific associations (Acoustical Society of America, Int. Speech Communication, IEEE, Int. Neural Networks Society, Association for Research in Otolaryngology, Society for Neuroscience and several others). He is a senior member of the IEEE and participates in many scientific committees.

Contents

**Preface IX** 

Chapter 1 **Projections, Partaken Circuits and Axon** 

José L. Bueno-López, Juan C. Chara,

Alyssa A. Brewer and Brian Barton

Chapter 3 **On the Specific Role of the Occipital Cortex in Scene Perception 61**  Carole Peyrin and Benoit Musel

Chapter 4 **Neural Mechanisms for Binocular Depth, Rivalry and Multistability 83** 

Chapter 5 **Visual Motion: From Cortex to Percept 111** 

Chapter 7 **Linking Neural Activity to Visual Perception:** 

Jackson E.T. Smith, Nicolas Y. Masse, Chang'an A. Zhan and Erik P. Cook

**Non-Natural Facial Expressions 185** 

Chapter 9 **Vision as a Fundamentally Statistical Machine 201** 

and Pedro Luis Sánchez Orellana

Zhiyong Yang

Athena Buckthought and Janine D. Mendola

Craig Aaen-Stockdale and Benjamin Thompson

Chapter 8 **Bio-Inspired Architecture for Clustering into Natural and** 

**Separating Sensory and Attentional Contributions 161** 

Claudio Castellanos Sánchez, Manuel Hernández Hernández

Chapter 6 **Visual Processing in the Action-Oriented Brain 139**  Brendan D. Cameron and Gordon Binsted

**Initial Segments of Cortical Principal Neurons 1** 

Juan L. Mendizabal-Zubiaga and Concepción Reblet

Chapter 2 **Visual Field Map Organization in Human Visual Cortex 29** 

## Contents

#### **Preface** XI


X Contents



## Preface

The neurosciences have experienced tremendous and wonderful progress in many areas, and the spectrum encompassing the neurosciences is expansive. Suffice it to mention a few classical fields: electrophysiology, genetics, physics, computer sciences, and more recently, social and marketing neurosciences. Of course, this large growth resulted in the production of many books. Perhaps the visual system and the visual cortex were in the vanguard because most animals do not produce their own light and offer thus the invaluable advantage of allowing investigators to conduct experiments in full control of the stimulus. In addition, the fascinating evolution of scientific techniques, the immense productivity of recent research, and the ensuing literature make it virtually impossible to publish in a single volume all worthwhile work accomplished throughout the scientific world. The days when a single individual, as Diderot, could undertake the production of an encyclopedia are gone forever. Indeed most approaches to studying the nervous system are valid and neuroscientists produce an almost astronomical number of interesting data accompanied by extremely worthy hypotheses which in turn generate new ventures in search of brain functions.

Yet, it is fully justified to make an encore and to publish a book dedicated to visual cortex and beyond. Many reasons validate a book assembling chapters written by active researchers. Each has the opportunity to bind together data and explore original ideas whose fate will not fall into the hands of uncompromising reviewers of traditional journals. This book focuses on the cerebral cortex with a large emphasis on vision. Yet it offers the reader diverse approaches employed to investigate the brain, for instance, computer simulation, cellular responses, or rivalry between various targets and goal directed actions.

This volume thus covers a large spectrum of research even though it is impossible to include all topics in the extremely diverse field of neurosciences.

For a long time, university students have been taught that after the critical period following birth the brain remains relatively unchanged except for a gradual neuronal loss accompanying aging. As recent research demonstrates this latter view was a false perception of brain organization because indeed the adult brain is incredibly plastic. Thus it should not come as a surprise that this book contains several chapters dealing with adult brain modifications. Furthermore, thanks to recent advances in histological techniques, cellular membranes of neurons exhibit an extraordinary complexity, even though they constitute the structural unit membrane of the neuron. Bueno-Lopez and Chara et al. review subtypes of cerebral cortical principal cells and provide a detailed description of cellular partitions such as axon initial segments. Brewer and Barton expand our current understanding of the visual field map in human visual cortex organization and reciprocal connections as they relate to specific functions. Yet mapping comes with intrinsic problems, e.g., the coupling or uncoupling between neurovasculature and neurometabolic requirements. Ling and Gao et al., describe this challenge, adding an interesting historical note.

Preface XI

evoked responses in the primary visual cortex are modulated by surrounding stimuli. The role of extra-classical receptive field is highlighted in the figure-ground organization of an image. Peyrin and Musel aim to clarify the different attributes of the occipital cortex during scene perception. This suggests that the occipital cortex might serve as an"active blackboard"integrating the rapid analysis of low spatial frequency (LSF) carried out by higher order cortical areas and sent back via feedback connections to occipital areas to influence the subsequent analysis of high spatial frequencies (HSF) features. Concomitantly, the right occipital cortex is predominantly involved in the categorization LSF processing, while the left occipital cortex is chiefly involved in the categorization of HSF scenes. Buckthought and Mendola explain the methodological issues involved in perception investigations, showing that perceptual processing modalities can be conceptualized as a series of visual perceptual processing stages in occipital areas, as well as higher-level cognitive functions in parietal and frontal areas, involving stimulus selection, decision making, motor planning, memory, attention and conscious awareness. Stockdale and Thompson review comprehensively key aspects of visual motion perception with a particular emphasis on the cortical areas thought to be involved and the ability of the visual cortex to bind multiple features together into a coherent, stable visual percept. In addition, the effects of lesions and abnormal development affecting the cortical areas responsible for motion perception are described. Cameron and Binsted examine the relationship between vision and action, and survey evidence that regions of posterior parietal cortex are designed for directly transforming vision into action; interestingly, studies from patients and non-patients suggest that action processing can access visual information that perception does not. Smith and Masse et al. pose a challenging question: how does visual perception arise from the activity of neurons in visual cortex? The authors introduce the important concept of'behavioral sensitivity'neurons to answer this question that requires an understanding of how a visual cortical neuron's activity

becomes correlated with the visually guided behavior of a subject.

methods (i.e., specific models).

The book ends with topics that are likely to blossom in the near future, i.e., modeling the complexity of brain functions. Castellanos-Sanchez and Hernandez et al. introduce the reader to bio-inspired models. Yang considers the brain as a statistical machine. The argument rests on the proposition that the statistics of the natural visual environment must be incorporated into the visual circuitry by successful behavior in the world over evolutionary and developmental time. Indeed, there is an almost infinite number of ways to model any given system, and this is particularly true if we are interested in something as complex as the human brain, including the visual cortex. By contrast Ladurantaye and Rouat et al. attempt in depth to regroup the different types of models into meaningful categories, to give the reader a fair overview of what is possible to achieve in terms of modeling biological vision. Each section of the chapter focuses on one of those categories and includes different implementation

Plasticity in the adult brain constitutes the key theme of the majority of chapters since this particular field has seen remarkable development at all functional levels. Maya-Ventencourt and Caleo compellingly discuss processes of sensory deprivation in the visual system at intracellular signal transduction pathways which regulate changes of chromatin structure and gene expression patterns supporting these plastic phenomena. For many years plasticity was studied by deprivation, that is, eliminating or limiting visual inputs. Fortunately, recent experiments reversed this approach by introducing a novel experimental design that enriches the visual environment or by imposing particular stimuli. Sale and Berardi et al.'s chapter aims to review recent studies, mostly focusing on the effects of enriching the visual environment in promoting visual system development and in reopening neural plasticity windows in the adult brain. Special emphasis is given to excitation / inhibition balance, and to promoting functional recovery from pathological states of severe brain disability. Along this line Bachatene and Bharmauria et al., compellingly describe the change of functional connections between involved neurons following adaptation-induced plasticity. After forcefully adapting the same neurons to a desired stimulus feature, and using the cross correlogram approach, functional connections were established before and after adaptation between the same neurons recorded simultaneously in the primary visual cortex. The data show the network between them changed, reflecting on the ability of visual cortex for plastic modifications of its relationships between neurons following adaptation. Using brain optical imaging Tanaka is of the same opinion that the short-term single-orientation exposure can dramatically alter preferred orientations until postnatal 6 weeks, which is against the current consensus and reveals further complexity in orientation plasticity in cases where animals are exposed to a single orientation for a long time. Gerrikagoitial and Rienda et al. stress convincingly the importance of inhibitory factors regulating connections and sprouting capacity within the cortico-collicular network during plasticity processes.

Yet vision as it is processed in visual cortex is not by all means an end in itself as it initiates image perception that steers our behaviour. This book offers several chapters dealing with the topic.

Contextual influences should not be underestimated when investigating neuronal responses to visual stimuli. Arall and Romeo et al. describe findings showing that evoked responses in the primary visual cortex are modulated by surrounding stimuli. The role of extra-classical receptive field is highlighted in the figure-ground organization of an image. Peyrin and Musel aim to clarify the different attributes of the occipital cortex during scene perception. This suggests that the occipital cortex might serve as an"active blackboard"integrating the rapid analysis of low spatial frequency (LSF) carried out by higher order cortical areas and sent back via feedback connections to occipital areas to influence the subsequent analysis of high spatial frequencies (HSF) features. Concomitantly, the right occipital cortex is predominantly involved in the categorization LSF processing, while the left occipital cortex is chiefly involved in the categorization of HSF scenes. Buckthought and Mendola explain the methodological issues involved in perception investigations, showing that perceptual processing modalities can be conceptualized as a series of visual perceptual processing stages in occipital areas, as well as higher-level cognitive functions in parietal and frontal areas, involving stimulus selection, decision making, motor planning, memory, attention and conscious awareness. Stockdale and Thompson review comprehensively key aspects of visual motion perception with a particular emphasis on the cortical areas thought to be involved and the ability of the visual cortex to bind multiple features together into a coherent, stable visual percept. In addition, the effects of lesions and abnormal development affecting the cortical areas responsible for motion perception are described. Cameron and Binsted examine the relationship between vision and action, and survey evidence that regions of posterior parietal cortex are designed for directly transforming vision into action; interestingly, studies from patients and non-patients suggest that action processing can access visual information that perception does not. Smith and Masse et al. pose a challenging question: how does visual perception arise from the activity of neurons in visual cortex? The authors introduce the important concept of'behavioral sensitivity'neurons to answer this question that requires an understanding of how a visual cortical neuron's activity becomes correlated with the visually guided behavior of a subject.

X Preface

techniques, cellular membranes of neurons exhibit an extraordinary complexity, even though they constitute the structural unit membrane of the neuron. Bueno-Lopez and Chara et al. review subtypes of cerebral cortical principal cells and provide a detailed description of cellular partitions such as axon initial segments. Brewer and Barton expand our current understanding of the visual field map in human visual cortex organization and reciprocal connections as they relate to specific functions. Yet mapping comes with intrinsic problems, e.g., the coupling or uncoupling between neurovasculature and neurometabolic requirements. Ling and Gao et al., describe this

Plasticity in the adult brain constitutes the key theme of the majority of chapters since this particular field has seen remarkable development at all functional levels. Maya-Ventencourt and Caleo compellingly discuss processes of sensory deprivation in the visual system at intracellular signal transduction pathways which regulate changes of chromatin structure and gene expression patterns supporting these plastic phenomena. For many years plasticity was studied by deprivation, that is, eliminating or limiting visual inputs. Fortunately, recent experiments reversed this approach by introducing a novel experimental design that enriches the visual environment or by imposing particular stimuli. Sale and Berardi et al.'s chapter aims to review recent studies, mostly focusing on the effects of enriching the visual environment in promoting visual system development and in reopening neural plasticity windows in the adult brain. Special emphasis is given to excitation / inhibition balance, and to promoting functional recovery from pathological states of severe brain disability. Along this line Bachatene and Bharmauria et al., compellingly describe the change of functional connections between involved neurons following adaptation-induced plasticity. After forcefully adapting the same neurons to a desired stimulus feature, and using the cross correlogram approach, functional connections were established before and after adaptation between the same neurons recorded simultaneously in the primary visual cortex. The data show the network between them changed, reflecting on the ability of visual cortex for plastic modifications of its relationships between neurons following adaptation. Using brain optical imaging Tanaka is of the same opinion that the short-term single-orientation exposure can dramatically alter preferred orientations until postnatal 6 weeks, which is against the current consensus and reveals further complexity in orientation plasticity in cases where animals are exposed to a single orientation for a long time. Gerrikagoitial and Rienda et al. stress convincingly the importance of inhibitory factors regulating connections and sprouting capacity within the cortico-collicular network during plasticity processes.

Yet vision as it is processed in visual cortex is not by all means an end in itself as it initiates image perception that steers our behaviour. This book offers several chapters

Contextual influences should not be underestimated when investigating neuronal responses to visual stimuli. Arall and Romeo et al. describe findings showing that

challenge, adding an interesting historical note.

dealing with the topic.

The book ends with topics that are likely to blossom in the near future, i.e., modeling the complexity of brain functions. Castellanos-Sanchez and Hernandez et al. introduce the reader to bio-inspired models. Yang considers the brain as a statistical machine. The argument rests on the proposition that the statistics of the natural visual environment must be incorporated into the visual circuitry by successful behavior in the world over evolutionary and developmental time. Indeed, there is an almost infinite number of ways to model any given system, and this is particularly true if we are interested in something as complex as the human brain, including the visual cortex. By contrast Ladurantaye and Rouat et al. attempt in depth to regroup the different types of models into meaningful categories, to give the reader a fair overview of what is possible to achieve in terms of modeling biological vision. Each section of the chapter focuses on one of those categories and includes different implementation methods (i.e., specific models).

#### XIV Preface

In closing it is our agreeable duty to express our gratitude to many anonymous reviewers who diligently revised the chapters. Finally, Mrs. Bakic is unreservedly thanked for her help during the tedious editing processes.

### **Prof. Dr. Stephane Molotchnikoff**

Department of Biological Sciences, Faculty of Arts and Sciences, University of Montreal, Canada

#### **Prof. Jean Rouat**

Department of Electrical and Computer Engineering, Faculty of Engineering, University of Sherbrooke, Canada

## **Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons**

José L. Bueno-López, Juan C. Chara, Juan L. Mendizabal-Zubiaga and Concepción Reblet

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/50191

## **1. Introduction**

The axon initial segment (AIS) is the portion of the neuron immediately distal to the axon hillock. The AIS has a specialized membrane that works for a manifold function. The first portion of the AIS membrane possess a collection of ion channels that allows for the modulation of the membrane potential of the parent cell whilst blocking the back propagation of the axon potential. The last portion of the AIS membrane possesses in addition voltagedependent ion channels that are responsible for the ultimate display of the cell, that is, the generation of the axon potential. Much has been investigated recently on the ion channels that are embedded in the AIS membrane of nerve cells. Yet basic parameters such as the length and diameter of the AIS and, of no lesser importance, the number and distribution of boutons synapsing the AIS membrane remains largely unknown for distinct subpopulations of principal cells of cerebral cortex. Principal cells are heterogeneous in many anatomical, molecular and functional aspects but, in agreement with their distinctive possession of combinations of these aspects, they can be classified in different subpopulations. Taking as core features for this grouping the cell laminar address and the pattern of axon projection, this paper reviews subtypes of cerebral cortical principal cells and their AIS features. In doing so, this paper also presents an account of our past and present research on the AIS of principal cells in visual and other areas of cerebral cortex. Yet, aiming to furnish a background to the function of the AIS of cerebral cortical principal cells, we shall begin by reviewing the cellular types and circuits in cortex, and the axon projections arising from the cortex.

## **2. Cerebral cortical neurons**

Spiny neurons are of an excitatory nature and most have extrinsic axons, that is, axons that project outside the cortical area where their somata lie. These cells employ glutamate or

© 2012 Bueno-López et al., licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

predominantly aspartate as a neurotransmitter (for percentages of aspartergic and glutamatergic cells out of projection cell subpopulations to several telencephalic and extratelencephalic targets, see [1]). By contrast, aspiny neurons are inhibitory, and most of their axons are intrinsic, i.e. they use gamma-amino-butyric acid (GABA) as a neurotransmitter and their axons remain within the cortical area in which the parent cell soma lies. They are usually named interneurons.

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 3

neocortex, but also in several areas of the mesocortex, paleocortex and archicortex, which include the piriform cortex, entorhinal cortex, subiculum, hippocampus, fascia dentata, cingulate cortex, claustrum and amygdala [5,11]. Chandelier cells are present in all cortical layers (except layer 1), most abundantly in layer 2/3 [5,12.13]. A single chandelier cell can innervate between 250 (in neocortex) and 1,200 (in hippocampus) principal cells, indicating the potential to synchronize many principal cells [5,14]. Thus, chandelier cells appear ideally suited to shut off entire groups of pyramidal cells, making them the ultimate cortical switches. Each chandelier-terminal innervates single AIS. Moreover, each AIS may be innervated by one or a few chandelier terminals (five or less), which may originate from the same or different chandelier cells (for a review, see [14]). Chandelier cells are parvalbumin-

*Basket cells.* Basket cells are most difficult to classify for their somatodendritic variety and sizes. The term 'basket' comes from the basket-like appearance of their axonal terminals around pyramidal cell somata that results from convergent innervation by many basket cells. About 50% of all inhibitory interneurons are basket cells [12]. Basket cells target the somata (15%-30% of their synapses are on somata) and proximal dendrites of principal cells; yet basket cells also target somata and dendrites of interneurons, particularly other basket cells [5,15,16]. There are *small basket cells* with a dense local axonal arborization that targets more dendrites than somata of principal cells; these small basket cells predominate in infragranular layers. *Large basket cells* are the typical basket cells. They have an extensive horizontal axonal branching with few vertical types of collateral. They are therefore the primary source of lateral inhibition across the cortical columns. Large basket cell predominate in layer 2/3. There are intermediate sizes of basket cells too. Basket cells can be parvalbumin-immunoreactive and fast spiking cells [6,16,17]. Other subgroup of basket cells are regular spiking; large- and medium-size basket cells are cholecystokinin immunoreactive neurons; small basket cells are immunoreactive for other calcium binding protein, or the vasointestinal neuropeptide; all this depends on the species studied [6,12,17].

The dendritic targeting interneurons are more suited to modify and gate incoming excitatory inputs. The dendritic targeting interneurons are the bipolar cells, Martinotti cells, neurogliaform cells and double-bouquet cells. The latter cells target dendrites and spines.

*Double bouquet cells.* The axon of a double bouquet cell forms a tight fascicular axonal cylinder that can extends across all layers, innervating distal dendrites and spines. The highly varicose collaterals that form these columnar bundles are unusually thicker than the axonal main stem. While the morphology and distribution of double bouquet cells are similar in the human and macaque neocortex, these cells are modified or less numerous in the neocortex of other species (e.g. the cat), and may even be absent (e.g. the mouse and rat). Thus, differences in the morphology, number and distribution of double bouquet cells may represent fundamental differences in cortical micro organization between primates and other species [3]. Double bouquet cells usually are calbindin-immunoreactive. Together with being calbindin-immunoreactive, they can be also calretinin-immunoreactive. Other double

reactive and fast-spiking cells.

*2.1.2. Dendritic targeting interneurons* 

### **2.1. Interneurons**

Interneurons are present in all cortical areas and layers and represent approximately 10%–20% of cortical neurons in rats [2] or 15%-30% of the total population in other species [3]. Interestingly, while in the occipital, parietal and frontal cortex of the rat the same proportion of GABAergic neurons among all neurons was found (15%, in [2]), the numerical density of all neurons in the frontal cortex (34,000 per cubic millimetre) was significantly lower than those in the occipital and parietal regions (52,000 per cubic millimetre and 48,000 per cubic millimetre, respectively) [2]. The fixed proportion of interneurons, irrespective of the number of neurons, is in keeping with the idea of the uniformity of cortical circuits. However, this does not exclude the possibility that the proportion of the different types of interneurons might vary in each cortical area, resulting in regional specialization of inhibitory circuits.

Interneurons show great morphological, biochemical and physiological diversity. However, interneurons with the same morphology may have different biochemical characteristics and connectivity [3]. Taken into account this consideration certain interneurons can be recognized by their unique morphological characteristics or they can be more generally divided in subgroups on the bases of their pattern of axonal arborization, synaptic connections (both with pyramidal cells or between themselves) and physiological and biochemical characteristics. One most accepted classification of interneurons is based on the domain of pyramidal neurons their axons target combined with the immunoreactivity for different calcium binding proteins and neuropeptides [4,5].

#### *2.1.1. Axosomatic targeting interneurons*

The axosomatic targeting interneurons are chandelier cells and basket cells. Chandelier cells target on the AIS of principal cells and basket cells target the somata and proximal portions of dendrites of principal cells. Both interneurons are likely to exhibit a greater impact on the direct output of postsynaptic neurons. An important number of them are fast-spiking interneurons, most of which being immunoreactive for the calcium binding protein parvalbumin. In turn, most parvalbumin-immunoreactive cells are fast spiking large basket cells and chandelier cells. Parvalbumin-immunoreactive cells in rodents account for 40%–50% of GABAergic neurons [6]. A much higher percentage (74%) was found in macaque visual cortex [7].

*Chandelier cells.* These cells are the only interneuron that shows clearly recognizable terminal axonal specializations, which form short vertical rows of terminal buttons, resembling candlesticks. These cells only synapse with the AIS of principal cells [8-10]. For this reason they were named axo-axonic cells. Chandelier cells have been found not only in the neocortex, but also in several areas of the mesocortex, paleocortex and archicortex, which include the piriform cortex, entorhinal cortex, subiculum, hippocampus, fascia dentata, cingulate cortex, claustrum and amygdala [5,11]. Chandelier cells are present in all cortical layers (except layer 1), most abundantly in layer 2/3 [5,12.13]. A single chandelier cell can innervate between 250 (in neocortex) and 1,200 (in hippocampus) principal cells, indicating the potential to synchronize many principal cells [5,14]. Thus, chandelier cells appear ideally suited to shut off entire groups of pyramidal cells, making them the ultimate cortical switches. Each chandelier-terminal innervates single AIS. Moreover, each AIS may be innervated by one or a few chandelier terminals (five or less), which may originate from the same or different chandelier cells (for a review, see [14]). Chandelier cells are parvalbuminreactive and fast-spiking cells.

*Basket cells.* Basket cells are most difficult to classify for their somatodendritic variety and sizes. The term 'basket' comes from the basket-like appearance of their axonal terminals around pyramidal cell somata that results from convergent innervation by many basket cells. About 50% of all inhibitory interneurons are basket cells [12]. Basket cells target the somata (15%-30% of their synapses are on somata) and proximal dendrites of principal cells; yet basket cells also target somata and dendrites of interneurons, particularly other basket cells [5,15,16]. There are *small basket cells* with a dense local axonal arborization that targets more dendrites than somata of principal cells; these small basket cells predominate in infragranular layers. *Large basket cells* are the typical basket cells. They have an extensive horizontal axonal branching with few vertical types of collateral. They are therefore the primary source of lateral inhibition across the cortical columns. Large basket cell predominate in layer 2/3. There are intermediate sizes of basket cells too. Basket cells can be parvalbumin-immunoreactive and fast spiking cells [6,16,17]. Other subgroup of basket cells are regular spiking; large- and medium-size basket cells are cholecystokinin immunoreactive neurons; small basket cells are immunoreactive for other calcium binding protein, or the vasointestinal neuropeptide; all this depends on the species studied [6,12,17].

#### *2.1.2. Dendritic targeting interneurons*

2 Visual Cortex – Current Status and Perspectives

**2.1. Interneurons** 

soma lies. They are usually named interneurons.

cortical area, resulting in regional specialization of inhibitory circuits.

different calcium binding proteins and neuropeptides [4,5].

A much higher percentage (74%) was found in macaque visual cortex [7].

*2.1.1. Axosomatic targeting interneurons* 

predominantly aspartate as a neurotransmitter (for percentages of aspartergic and glutamatergic cells out of projection cell subpopulations to several telencephalic and extratelencephalic targets, see [1]). By contrast, aspiny neurons are inhibitory, and most of their axons are intrinsic, i.e. they use gamma-amino-butyric acid (GABA) as a neurotransmitter and their axons remain within the cortical area in which the parent cell

Interneurons are present in all cortical areas and layers and represent approximately 10%–20% of cortical neurons in rats [2] or 15%-30% of the total population in other species [3]. Interestingly, while in the occipital, parietal and frontal cortex of the rat the same proportion of GABAergic neurons among all neurons was found (15%, in [2]), the numerical density of all neurons in the frontal cortex (34,000 per cubic millimetre) was significantly lower than those in the occipital and parietal regions (52,000 per cubic millimetre and 48,000 per cubic millimetre, respectively) [2]. The fixed proportion of interneurons, irrespective of the number of neurons, is in keeping with the idea of the uniformity of cortical circuits. However, this does not exclude the possibility that the proportion of the different types of interneurons might vary in each

Interneurons show great morphological, biochemical and physiological diversity. However, interneurons with the same morphology may have different biochemical characteristics and connectivity [3]. Taken into account this consideration certain interneurons can be recognized by their unique morphological characteristics or they can be more generally divided in subgroups on the bases of their pattern of axonal arborization, synaptic connections (both with pyramidal cells or between themselves) and physiological and biochemical characteristics. One most accepted classification of interneurons is based on the domain of pyramidal neurons their axons target combined with the immunoreactivity for

The axosomatic targeting interneurons are chandelier cells and basket cells. Chandelier cells target on the AIS of principal cells and basket cells target the somata and proximal portions of dendrites of principal cells. Both interneurons are likely to exhibit a greater impact on the direct output of postsynaptic neurons. An important number of them are fast-spiking interneurons, most of which being immunoreactive for the calcium binding protein parvalbumin. In turn, most parvalbumin-immunoreactive cells are fast spiking large basket cells and chandelier cells. Parvalbumin-immunoreactive cells in rodents account for 40%–50% of GABAergic neurons [6].

*Chandelier cells.* These cells are the only interneuron that shows clearly recognizable terminal axonal specializations, which form short vertical rows of terminal buttons, resembling candlesticks. These cells only synapse with the AIS of principal cells [8-10]. For this reason they were named axo-axonic cells. Chandelier cells have been found not only in the The dendritic targeting interneurons are more suited to modify and gate incoming excitatory inputs. The dendritic targeting interneurons are the bipolar cells, Martinotti cells, neurogliaform cells and double-bouquet cells. The latter cells target dendrites and spines.

*Double bouquet cells.* The axon of a double bouquet cell forms a tight fascicular axonal cylinder that can extends across all layers, innervating distal dendrites and spines. The highly varicose collaterals that form these columnar bundles are unusually thicker than the axonal main stem. While the morphology and distribution of double bouquet cells are similar in the human and macaque neocortex, these cells are modified or less numerous in the neocortex of other species (e.g. the cat), and may even be absent (e.g. the mouse and rat). Thus, differences in the morphology, number and distribution of double bouquet cells may represent fundamental differences in cortical micro organization between primates and other species [3]. Double bouquet cells usually are calbindin-immunoreactive. Together with being calbindin-immunoreactive, they can be also calretinin-immunoreactive. Other double

bouquet cells are vasointestinal-peptide-immunoreactive and still others, cholecystokininimmunoreactive [3,5,12].

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 5

Concerning the axonal target, there are associative neurons of intracortical projection in the same hemisphere and commissural neurons that project to homotopic and heterotopic sites of the opposite hemisphere by way of the corpus callosum and anterior commissure. Principal cells of intratelencephalic projection encompass also those that can extend their axon to the

Associative and commissural neurons exist in all layers but with differences concerning projection and species. A limited number of associative and commissural axons arise from spiny stellate cells of layer 4 and low layer 3 [26]. Most associative and commissural axons emerge from pyramidal cells sited in layers 2/3 and 5-6, and also from spiny inverted pyramidal cells and other polymorphic cells of layers 5-6 [23,27,28]. The axons of all these cells extend forward to innervate the layer 4 of cortical areas of higher-order hierarchy [23,27-30]. This is the so-called associative forward projection [31]. In turn, the associative backward projection arise from cells sited in layers 2/3 and 5-6 of higher-order areas to innervate layers 1 and, to a lesser extent, 3, 5 and 6 of areas of lesser-order hierarchy [23,27,32]. This layer-segregation of forward and backward projections indicates that they

Spiny inverted pyramidal cells are morphologically and numerically conspicuous among the cells of layers 5-6 furnishing the associative backward projection in rabbits and cats; these cells also originate an important proportion of commissural, cortico-claustral and cortico-striatal projections; they neglegeably project to extratelencephalic centres such as the geniculate nuclei, colliculi and pons [23,27,33]. See Table 1 for a summary and [28] for a

Commissural axons result from a broad and anatomically diverse population of principal cells that are located primarily in layers 2/3, 5 and 6 of restricted areas; there are interspecies differences in the laminar address of commissural neurons; in ferrets, rabbits and rats the commissural neurons predominate in infragranular layers [23,26,27,34,35]. Commissural cells can be further defined based on patterns of collateral projections to the ipsi- and contralateral striata and cortices, as well as by the expression of combinations of molecular markers [36].

Despite having common morphologies and similar laminar distributions, the commissural and associative neurons have been reported to differ from each other in the rat, cat, and monkey neocortex; there, at least in adult animals, they constitute two separate populations of neurons that rarely have dual projections [26,34]. The expression of the orphan nuclear receptor Nurr1 is associated to layer 6 neurons projecting to the ipsilateral cortex, but not to those cells projecting to contralateral cortical regions [37]. Recent studies in mice motor cortex show that 4% of the commissural cells of layer 2/3 and 34% of layer 5 extends dichotomous axons to ipsilateral prefrontal cortex and contralateral motor cortex [38]. Also, in the rat sensory-motor cortex, there are bifurcated projections to associative

Moreover, certain principal cells of layer 5 of rodents project to the striatum in addition to the ipsilateral and contralateral cortex [40-44]. Cells sited in layer 5 can also project to the ipsi- and

striatum, claustrum, amygdala and other sites that are not out of the telencephalon.

target very different neuronal elements in recipient regions [18].

and contralateral areas from cells of layers 5-6 [39].

review; see also Section 4 in this Chapter for more data on AIS of these cells.

*Martinotti cells* have a prominent axonal projection to layer 1. They have many fine unmyelinated ascending axon collaterals, which fan out as they ascend, bearing 'en passant' boutons in intermediate layers; some of these collaterals reach and branch as well in layer 1. Many other interneurons have ascending axon collaterals, but the prominent axonal arbour in layer 1 distinguishes Martinotti cells. Martinotti cells lie in particular in the deep cortical layers but also in layer 3 [18]. The majority of Martinotti cells are somatostatin-immunoreactive.

*Bipolar cells.* These are small cells with spindle or ovoid somata and narrow bipolar (most often) or bitufted dendrites that extend vertically. These cells can be found in all layers but predominate in layer 2/3 and layer 6. Bipolar neurons can be excitatory by releasing only vasointestinal-peptide, or inhibitory by releasing mainly GABA (though inhibitory bipolar cells also express vasointestinal-peptide). Bipolar cells commonly express calretinin too. Their axon forms a narrow band that crosses all layers leaving a little proportion of terminals, mainly on the basal dendrites of principal cells [12].

*Neurogliaform cells.* These are small-sized 'button-type' cells with many fine, radiating dendrites that are short and aspiny, finely beaded and rarely branched. They form a highly symmetrical and spherical dendritic field. The axon can arise from any part of the soma or from the base of a dendrite, and shortly after its origin, it breaks up into a dense, intertwined arborization of ultra-thin axons with as many as ten orders of branching. Fine boutons are distributed on the axonal collaterals to form GABA synapses onto the dendrites of target cells. The molecular characteristics of neurogliaform cells are still not well understood [3,12].

## **2.2. Spiny neurons**

In turn, the spiny group of cortical cell consists of several subgroups. Pyramidal neurons with upright somatodendritic orientation are by far the largest subgroup within the group of spiny cells. Upright pyramidal neurons are projecting cells of all cortical layers other than layer 1. Upright pyramidal neurons can be further subdivided (see below). In addition to upright pyramidal neurons, in layers 5-6 there is another collection of projection cells, of spiny nature too, the polymorphic-cell subgroup [19-23]. The spiny stellate neurons of layer 4 are an exception to the spiny-cell/extrinsic-projection correlation, because they are implicated in the canonical thalamo-cortical reciprocal circuit. They are directly innervated by thalamic axons and almost exclusively establish synaptic contact with the neighbouring layer 3. In this study we consider as a cerebral cortical principal cell any spiny neuron that extends an axon branch outside the cortical area where its soma is located.

## *2.2.1. Principal cells of intratelencephalic projection (inclusive of the type I cell, of striatal projection)*

Principal neurons have been sub classified by the laminar position of the cell body, somatodendritic morphology, electrophysiology and axonal target [24]. For a review, see [25]. Concerning the axonal target, there are associative neurons of intracortical projection in the same hemisphere and commissural neurons that project to homotopic and heterotopic sites of the opposite hemisphere by way of the corpus callosum and anterior commissure. Principal cells of intratelencephalic projection encompass also those that can extend their axon to the striatum, claustrum, amygdala and other sites that are not out of the telencephalon.

4 Visual Cortex – Current Status and Perspectives

immunoreactive [3,5,12].

**2.2. Spiny neurons** 

*projection)* 

bouquet cells are vasointestinal-peptide-immunoreactive and still others, cholecystokinin-

*Martinotti cells* have a prominent axonal projection to layer 1. They have many fine unmyelinated ascending axon collaterals, which fan out as they ascend, bearing 'en passant' boutons in intermediate layers; some of these collaterals reach and branch as well in layer 1. Many other interneurons have ascending axon collaterals, but the prominent axonal arbour in layer 1 distinguishes Martinotti cells. Martinotti cells lie in particular in the deep cortical layers but also in layer 3 [18]. The majority of Martinotti cells are somatostatin-immunoreactive.

*Bipolar cells.* These are small cells with spindle or ovoid somata and narrow bipolar (most often) or bitufted dendrites that extend vertically. These cells can be found in all layers but predominate in layer 2/3 and layer 6. Bipolar neurons can be excitatory by releasing only vasointestinal-peptide, or inhibitory by releasing mainly GABA (though inhibitory bipolar cells also express vasointestinal-peptide). Bipolar cells commonly express calretinin too. Their axon forms a narrow band that crosses all layers leaving a little proportion of

*Neurogliaform cells.* These are small-sized 'button-type' cells with many fine, radiating dendrites that are short and aspiny, finely beaded and rarely branched. They form a highly symmetrical and spherical dendritic field. The axon can arise from any part of the soma or from the base of a dendrite, and shortly after its origin, it breaks up into a dense, intertwined arborization of ultra-thin axons with as many as ten orders of branching. Fine boutons are distributed on the axonal collaterals to form GABA synapses onto the dendrites of target cells. The molecular characteristics of neurogliaform cells are still not well understood [3,12].

In turn, the spiny group of cortical cell consists of several subgroups. Pyramidal neurons with upright somatodendritic orientation are by far the largest subgroup within the group of spiny cells. Upright pyramidal neurons are projecting cells of all cortical layers other than layer 1. Upright pyramidal neurons can be further subdivided (see below). In addition to upright pyramidal neurons, in layers 5-6 there is another collection of projection cells, of spiny nature too, the polymorphic-cell subgroup [19-23]. The spiny stellate neurons of layer 4 are an exception to the spiny-cell/extrinsic-projection correlation, because they are implicated in the canonical thalamo-cortical reciprocal circuit. They are directly innervated by thalamic axons and almost exclusively establish synaptic contact with the neighbouring layer 3. In this study we consider as a cerebral cortical principal cell any spiny neuron that extends an axon branch

*2.2.1. Principal cells of intratelencephalic projection (inclusive of the type I cell, of striatal* 

Principal neurons have been sub classified by the laminar position of the cell body, somatodendritic morphology, electrophysiology and axonal target [24]. For a review, see [25].

terminals, mainly on the basal dendrites of principal cells [12].

outside the cortical area where its soma is located.

Associative and commissural neurons exist in all layers but with differences concerning projection and species. A limited number of associative and commissural axons arise from spiny stellate cells of layer 4 and low layer 3 [26]. Most associative and commissural axons emerge from pyramidal cells sited in layers 2/3 and 5-6, and also from spiny inverted pyramidal cells and other polymorphic cells of layers 5-6 [23,27,28]. The axons of all these cells extend forward to innervate the layer 4 of cortical areas of higher-order hierarchy [23,27-30]. This is the so-called associative forward projection [31]. In turn, the associative backward projection arise from cells sited in layers 2/3 and 5-6 of higher-order areas to innervate layers 1 and, to a lesser extent, 3, 5 and 6 of areas of lesser-order hierarchy [23,27,32]. This layer-segregation of forward and backward projections indicates that they target very different neuronal elements in recipient regions [18].

Spiny inverted pyramidal cells are morphologically and numerically conspicuous among the cells of layers 5-6 furnishing the associative backward projection in rabbits and cats; these cells also originate an important proportion of commissural, cortico-claustral and cortico-striatal projections; they neglegeably project to extratelencephalic centres such as the geniculate nuclei, colliculi and pons [23,27,33]. See Table 1 for a summary and [28] for a review; see also Section 4 in this Chapter for more data on AIS of these cells.

Commissural axons result from a broad and anatomically diverse population of principal cells that are located primarily in layers 2/3, 5 and 6 of restricted areas; there are interspecies differences in the laminar address of commissural neurons; in ferrets, rabbits and rats the commissural neurons predominate in infragranular layers [23,26,27,34,35]. Commissural cells can be further defined based on patterns of collateral projections to the ipsi- and contralateral striata and cortices, as well as by the expression of combinations of molecular markers [36].

Despite having common morphologies and similar laminar distributions, the commissural and associative neurons have been reported to differ from each other in the rat, cat, and monkey neocortex; there, at least in adult animals, they constitute two separate populations of neurons that rarely have dual projections [26,34]. The expression of the orphan nuclear receptor Nurr1 is associated to layer 6 neurons projecting to the ipsilateral cortex, but not to those cells projecting to contralateral cortical regions [37]. Recent studies in mice motor cortex show that 4% of the commissural cells of layer 2/3 and 34% of layer 5 extends dichotomous axons to ipsilateral prefrontal cortex and contralateral motor cortex [38]. Also, in the rat sensory-motor cortex, there are bifurcated projections to associative and contralateral areas from cells of layers 5-6 [39].

Moreover, certain principal cells of layer 5 of rodents project to the striatum in addition to the ipsilateral and contralateral cortex [40-44]. Cells sited in layer 5 can also project to the ipsi- and

contralateral claustra in rodents and rabbits [23,33,43]. These cortico-claustral cells may branch to the ipsilateral and contralateral cortical areas and striata, in addition to the claustra, but not to the thalamus [43]. By describing genes that identify molecularly distinct subpopulations of commissural neurons, a recent work has cast more light on the heterogeneity of these cells [45]. Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 7

All of these neurons of layer 2/3 and 5 that extend axons to telencephalic centres may be grouped in the type I principal cells upon the condition that they have branches to the striatum [44]. It should be noted, nevertheless, that this name (type I) had been chosen for the neurons of projection to extratelencephalic (or sub-cerebral centres) [1,46], which introduces confusion

The principal cells of extratelencephalic projection typically lie in layers 5-6. They include the projection cells to the thalamus and multiple sites in the brain stem and the spinal cord.

Cortico-striatal axons of layer 5 cells can be collaterals not only of the type I [44] projection but also of the type II, extratelencephalic projection to regions such as the spinal cord, brain stem nuclei, pontine nuclei, colliculi, substantia nigra, zona incerta and subthalamic nucleus [20,40-44,47]. The latter projection arises from type II cells that in their origin may belong to the same kind of principal neuron with extensive sub-cortical projections that during development looses particularly some of them depending on the cortical area in which the parent soma lies, as it is the instance for the visual cortex [1]. As well, the type II projection leaves driver collaterals to the core cells of *higher-order nuclei* of the thalamus, which are parvalbumin-positive cells and in turn project to the cortical layer 4 [41,44,48]. Such a branching pattern has been demonstrated for the axonal pathways of visual, somatosensory, pre-limbic and motor cortex of rats, cats and monkeys [41,42,47-51]. The presynaptic boutons of the axon of type II cells of layer 5 are usually of a big size, typically bigger than

Although principal cells share numerous common features within layer 5, they are heterogeneous in their somatodendritic morphology [1,25,52]. Type II cells of layer 5 have a thick apical dendrite extending into cortical layer 1 with a prominent terminal tuft; these cells produce distinctive initial bursts of tonic firing in response to current injection [46,53- 55]. Depending on the cell-body position in layer 5, type I cells are characterized by having an apical dendrite that can tuft in cortical layer 1 or ascend to this layer without tufting; type I cells tend to fire phasically [44,46,54-57]. In addition, cells of layer 5b of short apical dendrite may project to the thalamus and superior colliculus but not to the striatum [18].

Thus, at least three subtypes of principal cells of layer 5 can be tentatively classified in agreement to their projection: the type I cell of intratelencephalic projection; the type II cell of sub-cerebral projection with collaterals to the striatum and thalamus; finally, the corticothalamic cell of layer 5b, which can project to the superior colliculus but not to the striatum.

It is well known that some cells of cortical layer 6 innervate the thalamus leaving collaterals to cortical layer 4. This cortico-thalamic projection is originated in pyramidal cells but not in spiny inverted or other polymorphic cells of layer 6 (Table 1) [23,28]. This projection is

*2.2.2.1. Principal cells of extratelencephalic projection sited in layer 5 (type II multiple target* 

in the terminology.

*2.2.2. Principal cells of extratelencephalic projection* 

those leaved by axons of type I intratelencephalic cells [44,51].

*2.2.2.2. Principal cells of extratelencephalic projection sited in layer 6* 

*principal cells and cortico-thalamic cells)* 


**Table 1.** Percentages of spiny inverted cells of layers 5-6 out of the total number of principal cells of layer 5-6 of identified projection in cerebral cortex of rabbits [23,27,33] (The commissural projection is not included in this Table, but see [23,27].) All but one\* of these percentages were estimated taking into account all infragranular neurons within clear-cut columns of retrograde labelling that extended along the radial dimension of the cortex. (\*) Spiny inverted neurons make up the majority of cells within a horizontal band of cells located at the border between layers 5 and 6 of primary and secondary visual cortex of projection to the ipsilateral primary visual cortex. This band of retrogradely labelled cells extends for millimetres in the secondary visual cortex from the sites of injection of retrograde tracer in the primary visual cortex. This band is particularly cell-populated in brains after multiple injections of tracer; with single injections, labelled cells lie scattered for millimetres in the border between layer 5 and layer 6 [27,28]. These findings have shown that there is a highly convergent yet diffuse projection from the layer 5/6-border principal cells of secondary visual cortex to discrete points of the primary visual cortex. Spiny inverted neurons are the principal source of this type of projection. This widespread projection is distinct from the backward cortico-cortical projection from secondary to primary visual cortex that originates in discrete columnar patches of cells.

All of these neurons of layer 2/3 and 5 that extend axons to telencephalic centres may be grouped in the type I principal cells upon the condition that they have branches to the striatum [44]. It should be noted, nevertheless, that this name (type I) had been chosen for the neurons of projection to extratelencephalic (or sub-cerebral centres) [1,46], which introduces confusion in the terminology.

## *2.2.2. Principal cells of extratelencephalic projection*

6 Visual Cortex – Current Status and Perspectives

Cortico-cortical intrahemispheric backward projection to primary cortex from columns across

Cortico-cortical intrahemispheric backward projection to primary cortex from a horizontal, extended cell band along the layer 5/6 border of

Cortico-cortical intrahemispheric forward projection to secondary cortex from columns across cortical layers of primary cortex

Cortico-cortical intrahemispheric lateral projection from columns across cortical layers of associative

Cortico-thalamic projection to lateral and medial

cortex that originates in discrete columnar patches of cells.

cortical layers of secondary cortex

secondary and primary cortex\*

cortex

contralateral claustra in rodents and rabbits [23,33,43]. These cortico-claustral cells may branch to the ipsilateral and contralateral cortical areas and striata, in addition to the claustra, but not to the thalamus [43]. By describing genes that identify molecularly distinct subpopulations of commissural neurons, a recent work has cast more light on the heterogeneity of these cells [45].

Cortico-cortical intrahemispheric projection, total 25% — Not studied

Cortico-claustral projection from primary cortex 83% Not studied Not studied Cortico-claustral projection from secondary cortex 23% 24% Not studied Cortico-claustral projection — — 10% Cortico-striatal projection < 20% < 20% Not studied

geniculate nuclei Null Null Not studied Cortico-collicular projection Null Null Not studied Cortico-pontine projection < 1% Not studied Not studied **Table 1.** Percentages of spiny inverted cells of layers 5-6 out of the total number of principal cells of layer 5-6 of identified projection in cerebral cortex of rabbits [23,27,33] (The commissural projection is not included in this Table, but see [23,27].) All but one\* of these percentages were estimated taking into account all infragranular neurons within clear-cut columns of retrograde labelling that extended along the radial dimension of the cortex. (\*) Spiny inverted neurons make up the majority of cells within a horizontal band of cells located at the border between layers 5 and 6 of primary and secondary visual cortex of projection to the ipsilateral primary visual cortex. This band of retrogradely labelled cells extends for millimetres in the secondary visual cortex from the sites of injection of retrograde tracer in the primary visual cortex. This band is particularly cell-populated in brains after multiple injections of tracer; with single injections, labelled cells lie scattered for millimetres in the border between layer 5 and layer 6 [27,28]. These findings have shown that there is a highly convergent yet diffuse projection from the layer 5/6-border principal cells of secondary visual cortex to discrete points of the primary visual cortex. Spiny inverted neurons are the principal source of this type of projection. This widespread projection is distinct from the backward cortico-cortical projection from secondary to primary visual

**cortex** 

**Auditory cortex** 

26% Not studied Not studied

82% Not studied Not studied

7.5% 42 % Not studied

31% 30% Not studied

**Retrosplenial cortex** 

**Projection type Visual** 

The principal cells of extratelencephalic projection typically lie in layers 5-6. They include the projection cells to the thalamus and multiple sites in the brain stem and the spinal cord.

### *2.2.2.1. Principal cells of extratelencephalic projection sited in layer 5 (type II multiple target principal cells and cortico-thalamic cells)*

Cortico-striatal axons of layer 5 cells can be collaterals not only of the type I [44] projection but also of the type II, extratelencephalic projection to regions such as the spinal cord, brain stem nuclei, pontine nuclei, colliculi, substantia nigra, zona incerta and subthalamic nucleus [20,40-44,47]. The latter projection arises from type II cells that in their origin may belong to the same kind of principal neuron with extensive sub-cortical projections that during development looses particularly some of them depending on the cortical area in which the parent soma lies, as it is the instance for the visual cortex [1]. As well, the type II projection leaves driver collaterals to the core cells of *higher-order nuclei* of the thalamus, which are parvalbumin-positive cells and in turn project to the cortical layer 4 [41,44,48]. Such a branching pattern has been demonstrated for the axonal pathways of visual, somatosensory, pre-limbic and motor cortex of rats, cats and monkeys [41,42,47-51]. The presynaptic boutons of the axon of type II cells of layer 5 are usually of a big size, typically bigger than those leaved by axons of type I intratelencephalic cells [44,51].

Although principal cells share numerous common features within layer 5, they are heterogeneous in their somatodendritic morphology [1,25,52]. Type II cells of layer 5 have a thick apical dendrite extending into cortical layer 1 with a prominent terminal tuft; these cells produce distinctive initial bursts of tonic firing in response to current injection [46,53- 55]. Depending on the cell-body position in layer 5, type I cells are characterized by having an apical dendrite that can tuft in cortical layer 1 or ascend to this layer without tufting; type I cells tend to fire phasically [44,46,54-57]. In addition, cells of layer 5b of short apical dendrite may project to the thalamus and superior colliculus but not to the striatum [18].

Thus, at least three subtypes of principal cells of layer 5 can be tentatively classified in agreement to their projection: the type I cell of intratelencephalic projection; the type II cell of sub-cerebral projection with collaterals to the striatum and thalamus; finally, the corticothalamic cell of layer 5b, which can project to the superior colliculus but not to the striatum.

#### *2.2.2.2. Principal cells of extratelencephalic projection sited in layer 6*

It is well known that some cells of cortical layer 6 innervate the thalamus leaving collaterals to cortical layer 4. This cortico-thalamic projection is originated in pyramidal cells but not in spiny inverted or other polymorphic cells of layer 6 (Table 1) [23,28]. This projection is

specific of the cortico-thalamo-cortical loop [5,51,58]. Remarkably, the cortico-thalamic projection of layer 6 cells can innervate both types of thalamic relay neurons, i.e. the firstorder (core, parvalbumin-positive) and the higher-order (matrix, calbindin-positive) cells. Both types of relay neurons not only lie in *first-order relay nuclei* but also *higher-order or associative nuclei*—the relative number of first-order relay cells being predominant in firstorder relay nuclei and vice-versa. Regardless their thalamic address, the first-order relay cells are driven by ascending axons to the thalamus and successively innervate the spiny stellate cells and small pyramidal cells of layers 4 and 3b of precise sites of corresponding cortical areas. The higher-order relay cells widely innervate principal cells of layers 1, 2 and 3a of several cortical areas [48,51]. The first-order relay neurons are driver cells to their innervated cortical cells whereas higher-order relay neurons would be modulator cells to their innervated cortical cells. In turn, cells of layer 6a of cortico-thalamic projection are modulatory cells to the core cells of *first-* and *higher order nuclei* of the thalamus whilst leaving collaterals to the thalamic reticular nucleus and cortical layer 4 cells. (It should be remembered here that the type II cells of layer 5 are driving cells to the core cells of the *higher-order nuclei* of the thalamus and do not leave collaterals to the thalamic reticular nucleus.) Layer 6b cells of cortico-thalamic projection innervate the core and matrix cells of *first-* and *higher order nuclei* of the thalamus and leave collaterals to cortical layer 5 and 6 layer. By this way, cortical output will influence the synchronous ascending first-order and high order thalamo-cortical pathways. In Ray W. Guillery's words, this is "*how anatomical pathways link perception and action"* [59]. At least in the visual primary area, the corticothalamic cells of layer 6a dominate numerically over cortico-thalamic cells of layer 6b. Cortico-thalamic cells of layer 6 have been estimated to be between 30%-50% of all cells of layer 6 neurons (see [60] for review).

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 9

between projection subtypes of all of these intratelencephalic cells is similar [57]. In layer 5 there are several types of projection cells. Synapses occur reciprocally between cells of similar type, e.g., intratelencephalic with intratelencephalic cells, type II cells with type II cells, etc. There are connections between type I and type II cells of layer 5 too, but they are hierarchical, that is, projections from type I cells to type II cells are more probable than projections in opposite way [44,54-56]. Principal cells of layer 6 of intratelencephalic projection have a higher probability of extending reciprocal connections between one another than with other type of projection cells. Next, these cortico-cortical cells have a higher probability of extending collaterals to cortico-thalamic cells than vice-versa [60]. Cells of layer 6 receive collaterals from cells of any other layer except layers 2 and 4 and extend collaterals to cells of layers 5-6. The exception being the layer 6a cell of cortico-thalamic projection to the core cells of thalamic nuclei that also extend collaterals to the thalamic reticular nucleus. This cell originates cortico-

Considering all cortical layers, the weight of the local intrinsic circuit in cat and monkey neocortex is estimated to be 80%-85% of excitatory inputs on principal neurons [58,66] whereas the weight of short- and long-range cortico-cortical circuits are 15% and 3%; on the same cells, the weight of the cortico-thalamic and intriguing cortico-claustral circuits is 1.3% and 3% respectively [66]. Finally, although cells of layer 2/3 probably connect more with cells of the same layer, some cells of layer 2/3 send collaterals to both, type I and type II cells

Principal cells are approximately 80% of the total of cortical neurons; interneurons make up the remaining ≈20%. Over 71% of the synapses in the cortex are derived from principal cells. However, this number does not truly reflect the relative balance between excitation and inhibition in cortex. When the larger efficacy of inhibitory synapses is taken into account, the dominance of the principal cells is reduced to 24%. Thus, the spiny cells provide the basic framework of long-distance excitation in both the vertical and lateral dimensions of the

*Interneurons.* Cortical interneurons innervate mainly principal cells, but they also innervate interneurons [16]. Interneurons form distinct intralaminar and interlaminar networks [16,53]. The probability of reciprocal synaptic connections between principal cells and interneurons varies with the type of interneuron. The intralaminar reciprocal connections between fast-spiking interneurons (chandelier cells and basket parvalbumin-positive cells) and principal cells are significantly higher than the probability of reciprocal connections between non fast-spiking interneurons and principal cells [53]. On the contrary, the interlaminar reciprocal connection between principal cells and interneurons is more

Another distinctive feature of the network of interneurons is its coupling through gap junctions. Electrical coupling of neocortical interneurons is firmly established by anatomical studies and electrophysiological experiments since [68]. The coupling is generally between interneurons of the same type [69]. Chandelier cells are electrically coupled too, as disclosed recently [62]. However, neurogliaform cells are electrically coupled to other neurogliaform

cortical collaterals only to cortical layer 4 cells [60].

cortex, which is moulded by local inhibitory neurons [67].

frequent with non fast-spiking interneurons [53].

of layers 5 and 6.

## **3. Cortical intrinsic circuits**

*Principal cells.* A solitary action potential in a single principal cell of layer 2/3 can trigger polysynaptic chains of activity, detected as excitatory postsynaptic potentials and inhibitory postsynaptic potentials in recorded neurons [61]. This reveals an extremely efficacious means of activity propagation in the cortical network. In human brain slices, a relatively high proportion of basket (20%) and chandelier (33%) neurons could be driven to threshold by a single principal cell spike, in stark contrast to an estimated 1% likelihood of finding polysynaptic events in rats [61,62]. The fact is that the intralaminar circuit of layer 2/3 is highly recurrent and dominates its own cells. Principal cells of layer 2/3 extend collaterals for several millimetres to form patchy connections in the layer in cats and monkeys [63,64]. On the contrary, the intralaminar circuit of layer 5 depends little of the input from its own layer [58]. Intralaminar circuits exist basically on all species studied. However, the weight of the layer 2/3 circuit could be lower in rodents and rabbits because they do not have a patchy pattern connection in layer 2/3 but in layer 5b-6a [27,65].

Only the type I of principal cells occurs in layer 2/3. These cells extend their axon to the cortex, striatum, claustrum and other telencephalic centres. The probability of reciprocal connections between projection subtypes of all of these intratelencephalic cells is similar [57]. In layer 5 there are several types of projection cells. Synapses occur reciprocally between cells of similar type, e.g., intratelencephalic with intratelencephalic cells, type II cells with type II cells, etc. There are connections between type I and type II cells of layer 5 too, but they are hierarchical, that is, projections from type I cells to type II cells are more probable than projections in opposite way [44,54-56]. Principal cells of layer 6 of intratelencephalic projection have a higher probability of extending reciprocal connections between one another than with other type of projection cells. Next, these cortico-cortical cells have a higher probability of extending collaterals to cortico-thalamic cells than vice-versa [60]. Cells of layer 6 receive collaterals from cells of any other layer except layers 2 and 4 and extend collaterals to cells of layers 5-6. The exception being the layer 6a cell of cortico-thalamic projection to the core cells of thalamic nuclei that also extend collaterals to the thalamic reticular nucleus. This cell originates corticocortical collaterals only to cortical layer 4 cells [60].

8 Visual Cortex – Current Status and Perspectives

layer 6 neurons (see [60] for review).

**3. Cortical intrinsic circuits** 

pattern connection in layer 2/3 but in layer 5b-6a [27,65].

specific of the cortico-thalamo-cortical loop [5,51,58]. Remarkably, the cortico-thalamic projection of layer 6 cells can innervate both types of thalamic relay neurons, i.e. the firstorder (core, parvalbumin-positive) and the higher-order (matrix, calbindin-positive) cells. Both types of relay neurons not only lie in *first-order relay nuclei* but also *higher-order or associative nuclei*—the relative number of first-order relay cells being predominant in firstorder relay nuclei and vice-versa. Regardless their thalamic address, the first-order relay cells are driven by ascending axons to the thalamus and successively innervate the spiny stellate cells and small pyramidal cells of layers 4 and 3b of precise sites of corresponding cortical areas. The higher-order relay cells widely innervate principal cells of layers 1, 2 and 3a of several cortical areas [48,51]. The first-order relay neurons are driver cells to their innervated cortical cells whereas higher-order relay neurons would be modulator cells to their innervated cortical cells. In turn, cells of layer 6a of cortico-thalamic projection are modulatory cells to the core cells of *first-* and *higher order nuclei* of the thalamus whilst leaving collaterals to the thalamic reticular nucleus and cortical layer 4 cells. (It should be remembered here that the type II cells of layer 5 are driving cells to the core cells of the *higher-order nuclei* of the thalamus and do not leave collaterals to the thalamic reticular nucleus.) Layer 6b cells of cortico-thalamic projection innervate the core and matrix cells of *first-* and *higher order nuclei* of the thalamus and leave collaterals to cortical layer 5 and 6 layer. By this way, cortical output will influence the synchronous ascending first-order and high order thalamo-cortical pathways. In Ray W. Guillery's words, this is "*how anatomical pathways link perception and action"* [59]. At least in the visual primary area, the corticothalamic cells of layer 6a dominate numerically over cortico-thalamic cells of layer 6b. Cortico-thalamic cells of layer 6 have been estimated to be between 30%-50% of all cells of

*Principal cells.* A solitary action potential in a single principal cell of layer 2/3 can trigger polysynaptic chains of activity, detected as excitatory postsynaptic potentials and inhibitory postsynaptic potentials in recorded neurons [61]. This reveals an extremely efficacious means of activity propagation in the cortical network. In human brain slices, a relatively high proportion of basket (20%) and chandelier (33%) neurons could be driven to threshold by a single principal cell spike, in stark contrast to an estimated 1% likelihood of finding polysynaptic events in rats [61,62]. The fact is that the intralaminar circuit of layer 2/3 is highly recurrent and dominates its own cells. Principal cells of layer 2/3 extend collaterals for several millimetres to form patchy connections in the layer in cats and monkeys [63,64]. On the contrary, the intralaminar circuit of layer 5 depends little of the input from its own layer [58]. Intralaminar circuits exist basically on all species studied. However, the weight of the layer 2/3 circuit could be lower in rodents and rabbits because they do not have a patchy

Only the type I of principal cells occurs in layer 2/3. These cells extend their axon to the cortex, striatum, claustrum and other telencephalic centres. The probability of reciprocal connections Considering all cortical layers, the weight of the local intrinsic circuit in cat and monkey neocortex is estimated to be 80%-85% of excitatory inputs on principal neurons [58,66] whereas the weight of short- and long-range cortico-cortical circuits are 15% and 3%; on the same cells, the weight of the cortico-thalamic and intriguing cortico-claustral circuits is 1.3% and 3% respectively [66]. Finally, although cells of layer 2/3 probably connect more with cells of the same layer, some cells of layer 2/3 send collaterals to both, type I and type II cells of layers 5 and 6.

Principal cells are approximately 80% of the total of cortical neurons; interneurons make up the remaining ≈20%. Over 71% of the synapses in the cortex are derived from principal cells. However, this number does not truly reflect the relative balance between excitation and inhibition in cortex. When the larger efficacy of inhibitory synapses is taken into account, the dominance of the principal cells is reduced to 24%. Thus, the spiny cells provide the basic framework of long-distance excitation in both the vertical and lateral dimensions of the cortex, which is moulded by local inhibitory neurons [67].

*Interneurons.* Cortical interneurons innervate mainly principal cells, but they also innervate interneurons [16]. Interneurons form distinct intralaminar and interlaminar networks [16,53]. The probability of reciprocal synaptic connections between principal cells and interneurons varies with the type of interneuron. The intralaminar reciprocal connections between fast-spiking interneurons (chandelier cells and basket parvalbumin-positive cells) and principal cells are significantly higher than the probability of reciprocal connections between non fast-spiking interneurons and principal cells [53]. On the contrary, the interlaminar reciprocal connection between principal cells and interneurons is more frequent with non fast-spiking interneurons [53].

Another distinctive feature of the network of interneurons is its coupling through gap junctions. Electrical coupling of neocortical interneurons is firmly established by anatomical studies and electrophysiological experiments since [68]. The coupling is generally between interneurons of the same type [69]. Chandelier cells are electrically coupled too, as disclosed recently [62]. However, neurogliaform cells are electrically coupled to other neurogliaform

cells but also to basket cells, regular-spiking interneurons and chandelier cells, among others interneurons. Thus, the neurogliaform cell links networks of particular classes of interneurons, each network being in turn electrotonically coupled itself [70].

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 11

to the action potential initiation zone (see [73] for review). Voltage-threshold for action potential initiation seems to be largely determined by the density of Nav1.6, whereas Nav1.2 channels in the proximal AIS support action potential back-propagation into the somatodendritic compartment. Indeed, axonal action propagation fails to back propagate when the somatic membrane potential is hyperpolarized [77]. Spike-evoked Ca2+ influx in the AIS of different neurons in the SNC has been revealed by fluorescence imaging techniques [73]. These cells fire spike bursts typified by the recruitment of sub threshold Ca2+ influx through low voltage-activated Ca2+ channels, which include T- and R-type isoforms [80]. Traditionally, burst evoked Ca2+ influx was thought to be restricted to the dendrites. Indeed, these same low voltage-activated channels mediate Ca2+ influx in the AIS. Since low voltageactivated channels activate at hyperpolarized potentials, they contribute to sub threshold depolarization of the AIS. Thus, their activity can determine when and if a given stimulus evokes an action potential. Ca2+ influx through cartwheel cell T-type channels of the AIS, is down regulated by a type 3 dopamine receptor (D3R)-dependent pathway [81]. This neuromodulatory pathway is remarkably specific for just AIS T-type channels; neither dendritic T-type channels nor AIS Na+ or K+ channels were affected by D3R signalling [73]. Serotonin1A receptor mediates neuronal hyperpolarization by activating potassium channels in the AIS. In human and monkey neocortex the serotonin1A receptor have been reported to

be highly concentrated in the membrane of the AIS of principal neurons [82,83].

*Associated anchoring proteins and cytoskeletal components.* The structure of the AIS of multipolar neurons possesses a dense layer of finely granular material undercoating the plasma membrane, scattered clusters of ribosomes and fascicles of microtubules. The fascicles of microtubules occur only in the axon hillock and initial segment. An undercoating of the plasma membrane of the axon occurs in the node of Ranvier [84-86]. Interestingly, the plasma membrane of AIS and nodes is surrounded by a large extracellular space containing dense material; this similarity between nodes and AIS is coincident with the presence of voltage gated channels and specialization of the cytoskeleton present therein [73,76]. In addition, the dense material undercoating the plasma membrane of the AIS is separated 15- 25 microns from the internal surface of the AIS membrane [84]. The neuronal cytoskeleton, consisting of interacting spectrins and actins, forms the neuronal structural scaffold and is a spatial delimiter for neuronal membrane proteins; the membrane undercoating is a specialized cytoskeletal element, found only in the AIS. The ßIV isoform of spectrin (an actin-binding protein) and ankyrin G (a spectrin-binding protein) mutually confine each other to the AIS [14,76]. Ankyrin G provides a specific anchor for many AIS-specific proteins, including the Na+ and K+ channel subunits KCNQ2 and 3. PSD-93, other scaffold protein, binds to the Kv1 channels found at the AIS. In addition Kv1 channels are associated to the adhesion molecule Casppr2 in the layer 2/3 of the human cerebral cortex [76]. Silencing of PSD-93 expression in cultured hippocampal neurons blocks the recruitment of Kv1 channels to the AIS but not Na+ channels [87]. The AIS is also enriched in the cell adhesion molecules Nr-CAM and Neurofascin-186, and the cytoskeletal linker ßIV spectrin [76]. Recruitment of these proteins to the AIS also depends on ankyrin G [73,88]. Together, these results point to ankyrin G as the master regulator of AIS assembly. Silencing of AIS proteins in mature neurons in culture reveals that as for development, ankyrin G is required

Fast-spiking interneurons target the somatic and perisomatic domains of the principal cell; thus, these interneurons strongly regulate the output of the principal cell [4,18]. Reciprocal connections between pyramidal cells and fast-spiking interneurons act as a recurrent feedback inhibition that can regulate the timings of pyramidal cell firing [5]. In turn, spike timings of fast-spiking interneurons are correlated with the local field potential in the network of principal neurons during gamma oscillations that depend on the electrical and chemical coupling between fast-spiking, parvalbumin-positive interneurons [17,71]. Parvalbumin-positive basket cells fire counter-phase with principal cells and in the same phase but slightly delayed to chandelier cells [17,72].

## **4. The axon initial segment**

Electrically, the AIS bridges dendritic and axonal compartments, converting graded dendritic inputs into all-or-none action potentials. Molecularly, the AIS maintain neuronal polarity preserving the molecular distribution between the axonal and somatodendritic domains [73]. Recent studies have revealed an ever-expanding complexity in the molecular components and in the types and distribution of ion channels embedded in the AIS. This complexity underlies what is now being recognized as a highly dynamic structure [73]. AIS structure and composition vary considerably across, and even within, neuronal classes [74,75], seemingly tuned to the computational demands of the cell. A recent activity of the cell can affect AIS ion channel kinetics and availability, thus altering action potential waveform, timing, and probability. Over long timescales, even the location and size of the AIS can change to compensate for alterations in neuronal activity.

## **4.1. Ion channels, anchoring proteins and cytoskeletal components of the axon initial segment**

*Ion channels*. Although multiple neuronal sites can support action potential generation, the high density of Na+ channels inherent to the AIS makes it the lowest threshold site for it. Immunostaining, imaging of spike-dependent Na+ flux and electrophysiological studies suggest a similar density of Na+ channels throughout the AIS [73,76]. However, a recent study revealed that Na+ channels at the distal AIS and the adjacent axon have a much lower half-activation voltage (up to 14 mV) than those at the proximal AIS and the soma [77]. Accordingly, the use of newly developed voltage imaging techniques combined with careful analysis of the site of initiation and propagation of the action potentials, show that they preferentially initiate at the distal end of the AIS [77-79]. Consistent with these electrophysiological data, immunostaining results revealed a segregation of two Na+ channel subtypes at the AIS: high-threshold Nav1.2 channels and low-threshold Nav1.6 channels, targeted preferentially to the proximal and the distal AIS, respectively [77]. Immunostaining intensity of NaV1.6 peaked at the distal end of the AIS, corresponding well to the action potential initiation zone (see [73] for review). Voltage-threshold for action potential initiation seems to be largely determined by the density of Nav1.6, whereas Nav1.2 channels in the proximal AIS support action potential back-propagation into the somatodendritic compartment. Indeed, axonal action propagation fails to back propagate when the somatic membrane potential is hyperpolarized [77]. Spike-evoked Ca2+ influx in the AIS of different neurons in the SNC has been revealed by fluorescence imaging techniques [73]. These cells fire spike bursts typified by the recruitment of sub threshold Ca2+ influx through low voltage-activated Ca2+ channels, which include T- and R-type isoforms [80]. Traditionally, burst evoked Ca2+ influx was thought to be restricted to the dendrites. Indeed, these same low voltage-activated channels mediate Ca2+ influx in the AIS. Since low voltageactivated channels activate at hyperpolarized potentials, they contribute to sub threshold depolarization of the AIS. Thus, their activity can determine when and if a given stimulus evokes an action potential. Ca2+ influx through cartwheel cell T-type channels of the AIS, is down regulated by a type 3 dopamine receptor (D3R)-dependent pathway [81]. This neuromodulatory pathway is remarkably specific for just AIS T-type channels; neither dendritic T-type channels nor AIS Na+ or K+ channels were affected by D3R signalling [73]. Serotonin1A receptor mediates neuronal hyperpolarization by activating potassium channels in the AIS. In human and monkey neocortex the serotonin1A receptor have been reported to be highly concentrated in the membrane of the AIS of principal neurons [82,83].

10 Visual Cortex – Current Status and Perspectives

**4. The axon initial segment** 

**initial segment** 

phase but slightly delayed to chandelier cells [17,72].

AIS can change to compensate for alterations in neuronal activity.

cells but also to basket cells, regular-spiking interneurons and chandelier cells, among others interneurons. Thus, the neurogliaform cell links networks of particular classes of

Fast-spiking interneurons target the somatic and perisomatic domains of the principal cell; thus, these interneurons strongly regulate the output of the principal cell [4,18]. Reciprocal connections between pyramidal cells and fast-spiking interneurons act as a recurrent feedback inhibition that can regulate the timings of pyramidal cell firing [5]. In turn, spike timings of fast-spiking interneurons are correlated with the local field potential in the network of principal neurons during gamma oscillations that depend on the electrical and chemical coupling between fast-spiking, parvalbumin-positive interneurons [17,71]. Parvalbumin-positive basket cells fire counter-phase with principal cells and in the same

Electrically, the AIS bridges dendritic and axonal compartments, converting graded dendritic inputs into all-or-none action potentials. Molecularly, the AIS maintain neuronal polarity preserving the molecular distribution between the axonal and somatodendritic domains [73]. Recent studies have revealed an ever-expanding complexity in the molecular components and in the types and distribution of ion channels embedded in the AIS. This complexity underlies what is now being recognized as a highly dynamic structure [73]. AIS structure and composition vary considerably across, and even within, neuronal classes [74,75], seemingly tuned to the computational demands of the cell. A recent activity of the cell can affect AIS ion channel kinetics and availability, thus altering action potential waveform, timing, and probability. Over long timescales, even the location and size of the

**4.1. Ion channels, anchoring proteins and cytoskeletal components of the axon** 

*Ion channels*. Although multiple neuronal sites can support action potential generation, the high density of Na+ channels inherent to the AIS makes it the lowest threshold site for it. Immunostaining, imaging of spike-dependent Na+ flux and electrophysiological studies suggest a similar density of Na+ channels throughout the AIS [73,76]. However, a recent study revealed that Na+ channels at the distal AIS and the adjacent axon have a much lower half-activation voltage (up to 14 mV) than those at the proximal AIS and the soma [77]. Accordingly, the use of newly developed voltage imaging techniques combined with careful analysis of the site of initiation and propagation of the action potentials, show that they preferentially initiate at the distal end of the AIS [77-79]. Consistent with these electrophysiological data, immunostaining results revealed a segregation of two Na+ channel subtypes at the AIS: high-threshold Nav1.2 channels and low-threshold Nav1.6 channels, targeted preferentially to the proximal and the distal AIS, respectively [77]. Immunostaining intensity of NaV1.6 peaked at the distal end of the AIS, corresponding well

interneurons, each network being in turn electrotonically coupled itself [70].

*Associated anchoring proteins and cytoskeletal components.* The structure of the AIS of multipolar neurons possesses a dense layer of finely granular material undercoating the plasma membrane, scattered clusters of ribosomes and fascicles of microtubules. The fascicles of microtubules occur only in the axon hillock and initial segment. An undercoating of the plasma membrane of the axon occurs in the node of Ranvier [84-86]. Interestingly, the plasma membrane of AIS and nodes is surrounded by a large extracellular space containing dense material; this similarity between nodes and AIS is coincident with the presence of voltage gated channels and specialization of the cytoskeleton present therein [73,76]. In addition, the dense material undercoating the plasma membrane of the AIS is separated 15- 25 microns from the internal surface of the AIS membrane [84]. The neuronal cytoskeleton, consisting of interacting spectrins and actins, forms the neuronal structural scaffold and is a spatial delimiter for neuronal membrane proteins; the membrane undercoating is a specialized cytoskeletal element, found only in the AIS. The ßIV isoform of spectrin (an actin-binding protein) and ankyrin G (a spectrin-binding protein) mutually confine each other to the AIS [14,76]. Ankyrin G provides a specific anchor for many AIS-specific proteins, including the Na+ and K+ channel subunits KCNQ2 and 3. PSD-93, other scaffold protein, binds to the Kv1 channels found at the AIS. In addition Kv1 channels are associated to the adhesion molecule Casppr2 in the layer 2/3 of the human cerebral cortex [76]. Silencing of PSD-93 expression in cultured hippocampal neurons blocks the recruitment of Kv1 channels to the AIS but not Na+ channels [87]. The AIS is also enriched in the cell adhesion molecules Nr-CAM and Neurofascin-186, and the cytoskeletal linker ßIV spectrin [76]. Recruitment of these proteins to the AIS also depends on ankyrin G [73,88]. Together, these results point to ankyrin G as the master regulator of AIS assembly. Silencing of AIS proteins in mature neurons in culture reveals that as for development, ankyrin G is required to maintain ion channels at the AIS [89]. Intriguingly, these and other experiments also showed that ankyrin G functions not only to cluster and maintain ion channels, but also to maintain neuronal polarity [89].

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 13

the somatic robust inhibitory barrier; in turn, the back-propagation of a generated axon potential can more effectively result in changes of the membrane potential of the apical dendrite and its branches. Thus, two physically segregated input integrations may simultaneously operate in the cells in which the axon arises from the apical dendrite: one at the immediate apical-tree membrane domain, the other at the somatic and remaining dendritic membrane domain. In the cat visual cortex, the orientation of apical dendrites of a spiny inverted neuron depends on the position of the parent cell within the gyri [28]. Accordingly, the incidence of truly reversed spiny inverted neurons decreased from the top of the gyri to the bottom of the sulci, while that of almost horizontally orientated pyramidal neurons increased. Most axons were found to arise from the apical dendrite or from one of its first

We reconstructed by means of electron-microscopy serial photography the AIS of eleven spiny inverted neurons of unknown projection (though they probably were cortico-cortical. corticoclaustral, or cortico-striatal, see Table 1). The length of the AIS of these neurons ranged between 32.0 and 101.2 microns, and its thickness varied between 0.58 and 1.04 microns. The range of the AIS length described by others shows some variability, indicating that cells of different subtypes of projection neurons, or cells within the subtypes themselves, may have different AIS length; e.g. the commissural and cortico-thalamic cells show more variability in the length of the AIS than the cortico-cortical associative ones [86]. Importantly, the number of synaptic boutons received by the AIS of spiny inverted neurons was found to be 24.4 on the average, but the range was also wide (11-37 boutons). Analysis of the spiny inverted neurons as classified in terms of the emergence site of the axon revealed that AIS which proceed from the apical dendrite were the shortest, thinnest and less innervated, whereas AIS arising from the somatic flank were the longest, widest and most innervated. Thus, AIS originated on apical dendrites averaged a length of 38.8 microns (range 37.1—45.4 microns), a diameter of 6.0 microns (range 0.58–0.62 microns) and a number of 15 apposed boutons (range 11—18 boutons). AIS of lateral somatic emergence averaged a length of 95.5 microns (range 93.2–97.8), a diameter of 1.04 microns (range 1.04—1.04) and 34 boutons (range 31—37). In turn, six AISs of somatic basal emergence averaged 60.9 microns in length, 0.83 microns in diameter and 26 boutons (ranges being 32.0–101.2 microns, 0.71–0.96 microns and 21—29 boutons, respectively) [28]. These features of the AIS may support an output generation that is peculiar to subclasses of spiny inverted neurons. Work is currently in progress in our laboratory to refine to which extent differences in the number of synapses received by the AIS are related to the type of axon

**6. Study with electron microscopy of the axon initial segment of the upright variety of cerebral cortical principal cell with experimentally identified backward ipsilateral and commissural projections in visual** 

We present in the following a preliminary report on AIS of twelve pyramidal neurons of cortico-cortical projection to primary visual cortex; these cells, amid other projection cells,

branches in cat spiny inverted neurons too.

projection of spiny inverted cells.

**cortex** 

## **4.2. Length of the axon initial segment**

AIS-length of cortical principal cells ranges between 17-40 microns as counted from the axon hillock [76,84,86]. This AIS length approximately coincides with the electrophysiological zone of initiation of the action potential, estimated to be ≈35-50 microns from the soma [78,79,90,91]. Interneurons have a shorter AIS than principal cells [15]. Granule cells of the fascia dentata give rise to the smallest unmyelinated fibres of the CNS. In these cells the action potentials are initiated at the distal axon but at ≈5-15 microns from soma with Na+ channel density specialized for robust action potential initiation and propagation with minimal current flow [91,92]. The length of the AIS of granule cells of the hippocampus resembled that of the large basket cells of the neocortex, which on average are 6 microns length although the axons of large basket cells are myelinated [15]. Hippocampal granule cells may be an exception because they are not connected between themselves as it occurs in all cortical neurons [5]. It would be of interest to explore if differences in the length of AIS are functionally relevant. In the AIS of hippocampal granule cells the Nav1.6 channel is the predominant alfa-subunit whilst the Nav1.2 cannot be detected [92]; this suggests that Nav1.6 channels are enough to generate the axon potential, and therefore that the mechanisms governing this generation are different in the thin non-myelinated axon of the granule cell than in other myelinated or unmyelinated neurons having also a short AIS length.

## **5. Study with electron microscopy of the axon initial segment of the inverted variety of cerebral cortical principal cell (of intratelencephalic projections)**

Our studies following the use of retrograde tracers in cerebral cortex of rats, rabbits and cats showed that spiny inverted neurons of layers 5 and 6 originate intra-telencephalic projections [23,27,28,33]. Spiny inverted neurons are odd because of having not only circumscribed projections and an inverse somatodendritic orientation but also for the sites from which their axon arises. As seen with Golgi-method impregnation, retrograde tracers and intracellular filling (for a review, see [28]), these cell sites may be (1) the basal surface of the soma, or even the basal dendrite portion next to the soma, (2) the (lateral) surface of the soma and (3) the apical dendrite, sometimes from a sector more remote from the soma than the emergence site of the first dendrite branching [23]. Of the 127 Golgi-impregnated spiny inverted neurons in the occipital and temporal cortices of rabbits which we examined, 29% of the axons arises from (1), 9.5% from (2) and 61.5% from (3) [23]. This distribution is similar to that in rat visual and sensory-motor cortices [28]: of 28 Golgi-impregnated spiny inverted neurons, 32% of the axons emerges from (1), 21.5% from (2) and 46.5% from (3). Hence, the probability of axon emergence from the apical dendrite is higher both in rabbits and rats. There, inputs on the apical dendrite and its branches can be more operational for the generation of cell outputs, as they do not pass the somatic robust inhibitory barrier; in turn, the back-propagation of a generated axon potential can more effectively result in changes of the membrane potential of the apical dendrite and its branches. Thus, two physically segregated input integrations may simultaneously operate in the cells in which the axon arises from the apical dendrite: one at the immediate apical-tree membrane domain, the other at the somatic and remaining dendritic membrane domain. In the cat visual cortex, the orientation of apical dendrites of a spiny inverted neuron depends on the position of the parent cell within the gyri [28]. Accordingly, the incidence of truly reversed spiny inverted neurons decreased from the top of the gyri to the bottom of the sulci, while that of almost horizontally orientated pyramidal neurons increased. Most axons were found to arise from the apical dendrite or from one of its first branches in cat spiny inverted neurons too.

12 Visual Cortex – Current Status and Perspectives

maintain neuronal polarity [89].

**4.2. Length of the axon initial segment** 

neurons having also a short AIS length.

**projections)** 

to maintain ion channels at the AIS [89]. Intriguingly, these and other experiments also showed that ankyrin G functions not only to cluster and maintain ion channels, but also to

AIS-length of cortical principal cells ranges between 17-40 microns as counted from the axon hillock [76,84,86]. This AIS length approximately coincides with the electrophysiological zone of initiation of the action potential, estimated to be ≈35-50 microns from the soma [78,79,90,91]. Interneurons have a shorter AIS than principal cells [15]. Granule cells of the fascia dentata give rise to the smallest unmyelinated fibres of the CNS. In these cells the action potentials are initiated at the distal axon but at ≈5-15 microns from soma with Na+ channel density specialized for robust action potential initiation and propagation with minimal current flow [91,92]. The length of the AIS of granule cells of the hippocampus resembled that of the large basket cells of the neocortex, which on average are 6 microns length although the axons of large basket cells are myelinated [15]. Hippocampal granule cells may be an exception because they are not connected between themselves as it occurs in all cortical neurons [5]. It would be of interest to explore if differences in the length of AIS are functionally relevant. In the AIS of hippocampal granule cells the Nav1.6 channel is the predominant alfa-subunit whilst the Nav1.2 cannot be detected [92]; this suggests that Nav1.6 channels are enough to generate the axon potential, and therefore that the mechanisms governing this generation are different in the thin non-myelinated axon of the granule cell than in other myelinated or unmyelinated

**5. Study with electron microscopy of the axon initial segment of the inverted variety of cerebral cortical principal cell (of intratelencephalic** 

Our studies following the use of retrograde tracers in cerebral cortex of rats, rabbits and cats showed that spiny inverted neurons of layers 5 and 6 originate intra-telencephalic projections [23,27,28,33]. Spiny inverted neurons are odd because of having not only circumscribed projections and an inverse somatodendritic orientation but also for the sites from which their axon arises. As seen with Golgi-method impregnation, retrograde tracers and intracellular filling (for a review, see [28]), these cell sites may be (1) the basal surface of the soma, or even the basal dendrite portion next to the soma, (2) the (lateral) surface of the soma and (3) the apical dendrite, sometimes from a sector more remote from the soma than the emergence site of the first dendrite branching [23]. Of the 127 Golgi-impregnated spiny inverted neurons in the occipital and temporal cortices of rabbits which we examined, 29% of the axons arises from (1), 9.5% from (2) and 61.5% from (3) [23]. This distribution is similar to that in rat visual and sensory-motor cortices [28]: of 28 Golgi-impregnated spiny inverted neurons, 32% of the axons emerges from (1), 21.5% from (2) and 46.5% from (3). Hence, the probability of axon emergence from the apical dendrite is higher both in rabbits and rats. There, inputs on the apical dendrite and its branches can be more operational for the generation of cell outputs, as they do not pass We reconstructed by means of electron-microscopy serial photography the AIS of eleven spiny inverted neurons of unknown projection (though they probably were cortico-cortical. corticoclaustral, or cortico-striatal, see Table 1). The length of the AIS of these neurons ranged between 32.0 and 101.2 microns, and its thickness varied between 0.58 and 1.04 microns. The range of the AIS length described by others shows some variability, indicating that cells of different subtypes of projection neurons, or cells within the subtypes themselves, may have different AIS length; e.g. the commissural and cortico-thalamic cells show more variability in the length of the AIS than the cortico-cortical associative ones [86]. Importantly, the number of synaptic boutons received by the AIS of spiny inverted neurons was found to be 24.4 on the average, but the range was also wide (11-37 boutons). Analysis of the spiny inverted neurons as classified in terms of the emergence site of the axon revealed that AIS which proceed from the apical dendrite were the shortest, thinnest and less innervated, whereas AIS arising from the somatic flank were the longest, widest and most innervated. Thus, AIS originated on apical dendrites averaged a length of 38.8 microns (range 37.1—45.4 microns), a diameter of 6.0 microns (range 0.58–0.62 microns) and a number of 15 apposed boutons (range 11—18 boutons). AIS of lateral somatic emergence averaged a length of 95.5 microns (range 93.2–97.8), a diameter of 1.04 microns (range 1.04—1.04) and 34 boutons (range 31—37). In turn, six AISs of somatic basal emergence averaged 60.9 microns in length, 0.83 microns in diameter and 26 boutons (ranges being 32.0–101.2 microns, 0.71–0.96 microns and 21—29 boutons, respectively) [28]. These features of the AIS may support an output generation that is peculiar to subclasses of spiny inverted neurons. Work is currently in progress in our laboratory to refine to which extent differences in the number of synapses received by the AIS are related to the type of axon projection of spiny inverted cells.

## **6. Study with electron microscopy of the axon initial segment of the upright variety of cerebral cortical principal cell with experimentally identified backward ipsilateral and commissural projections in visual cortex**

We present in the following a preliminary report on AIS of twelve pyramidal neurons of cortico-cortical projection to primary visual cortex; these cells, amid other projection cells,

were recently studied in our laboratory [93]. All these twelve cells had a typical upright somatodendritic orientation; they sited in the lateral partition of secondary visual cortex of rats (Figure 1). The axons of six of these neurons projected to the ipsilateral primary visual cortex that bears the cortical representation of the vertical central meridian of binocular visual field. The axons of the remaining six neurons projected to the same region of primary visual cortex but contralaterally through the callosum. In each of these two groups, three neurons lay in layer 3 and the remaining three in layer 5. Thus, the neurons reported here were distributed in four subgroups in agreement with their layer address and corticocortical projection.

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 15

Contralateral

!

**Labelling in Oc1 / Oc2L border**

**Figure 2.** Microphotographs of paired coronal sections of left (A) and right (B) hemispheres of rat visual cerebral cortex. See columns of labelling in ipsilateral Oc2L (lateral partition of secondary visual area)

**A B**

microinjection (blue arrow) in Oc1 (primary visual area). Projection cells were selected for the present study among BDA-labelled cells of these or similar columns following comparable injections of BDA in

**Figure 3.** Microphotograph of a cell sited in layer 3 of Oc2L (lateral partition of secondary visual area); the cell was labelled after a biotinylated dextran-amide injection to ipsilateral Oc1 (primary visual area) in the cerebral cortex of a rat. The cell was flat embedded and the resultant slice was glued on top of a resin capsule. The axon initial segment (arrow) underwent serial ultrathin cutting and then it was

examined under electronic microscopy. Scale bar, 15 microns.

and contralateral border between Oc2L and Oc1 after a biotinylated dextran-amide (BDA)

other rats. Scale bars, 500 microns.

**Labelling in Oc2L**

**BDA injec on in lateral Oc1**

Ipsilateral

**Figure 1.** Schematic drawings of the dorsal view (A) and lateral view (B) of the left hemisphere of rat cerebral cortex. Oc1 corresponds with the primary visual area; Oc2M and Oc2L do so with the medial and lateral parts of the secondary visual area, respectively. The border between Oc1 and Oc2L is the cortical representation of the vertical central meridian of binocular visual field. Biotinylated dextranamide (BDA) microinjections were placed in the Oc1 side of this border. Cells labelled with BDA by retrograde axon transport were taken from the ipsilateral Oc2L and homotopic contralateral border between Oc1 and Oc2L.

## **6.1. Material and methods**

To study all these twelve neurons we combined axon track-tracing methods and serial electron microscopy. We injected biotinylated dextran-amide (BDA) in primary visual cortex, in order to identify projection pyramidal neurons under light microscopy (Figures 1, 2); then, we performed serial ultrathin-cutting and -photography to study anatomical parameters of labelled AISs under electron microscopy (Figures 3-5). Studied AIS parameters were length, thickness, and number of apposed synaptic boutons and distribution of these boutons along the AIS membrane. Student's t tests (p-value ≤ 0.05) and linear regressions statistics were used to compare measured AIS parameters (simple linear regression and non-parametric multiple tests; p-value ≤ 0.05).

cortical projection.

between Oc1 and Oc2L.

**6.1. Material and methods** 

were recently studied in our laboratory [93]. All these twelve cells had a typical upright somatodendritic orientation; they sited in the lateral partition of secondary visual cortex of rats (Figure 1). The axons of six of these neurons projected to the ipsilateral primary visual cortex that bears the cortical representation of the vertical central meridian of binocular visual field. The axons of the remaining six neurons projected to the same region of primary visual cortex but contralaterally through the callosum. In each of these two groups, three neurons lay in layer 3 and the remaining three in layer 5. Thus, the neurons reported here were distributed in four subgroups in agreement with their layer address and cortico-

**Figure 1.** Schematic drawings of the dorsal view (A) and lateral view (B) of the left hemisphere of rat cerebral cortex. Oc1 corresponds with the primary visual area; Oc2M and Oc2L do so with the medial and lateral parts of the secondary visual area, respectively. The border between Oc1 and Oc2L is the cortical representation of the vertical central meridian of binocular visual field. Biotinylated dextranamide (BDA) microinjections were placed in the Oc1 side of this border. Cells labelled with BDA by retrograde axon transport were taken from the ipsilateral Oc2L and homotopic contralateral border

Dorsal view Lateral view

To study all these twelve neurons we combined axon track-tracing methods and serial electron microscopy. We injected biotinylated dextran-amide (BDA) in primary visual cortex, in order to identify projection pyramidal neurons under light microscopy (Figures 1, 2); then, we performed serial ultrathin-cutting and -photography to study anatomical parameters of labelled AISs under electron microscopy (Figures 3-5). Studied AIS parameters were length, thickness, and number of apposed synaptic boutons and distribution of these boutons along the AIS membrane. Student's t tests (p-value ≤ 0.05) and linear regressions statistics were used to compare measured AIS parameters (simple linear

regression and non-parametric multiple tests; p-value ≤ 0.05).

**Figure 2.** Microphotographs of paired coronal sections of left (A) and right (B) hemispheres of rat visual cerebral cortex. See columns of labelling in ipsilateral Oc2L (lateral partition of secondary visual area) and contralateral border between Oc2L and Oc1 after a biotinylated dextran-amide (BDA) microinjection (blue arrow) in Oc1 (primary visual area). Projection cells were selected for the present study among BDA-labelled cells of these or similar columns following comparable injections of BDA in other rats. Scale bars, 500 microns.

**Figure 3.** Microphotograph of a cell sited in layer 3 of Oc2L (lateral partition of secondary visual area); the cell was labelled after a biotinylated dextran-amide injection to ipsilateral Oc1 (primary visual area) in the cerebral cortex of a rat. The cell was flat embedded and the resultant slice was glued on top of a resin capsule. The axon initial segment (arrow) underwent serial ultrathin cutting and then it was examined under electronic microscopy. Scale bar, 15 microns.

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 17

As a group, upright principal cells furnishing the associative backward projection from ipsilateral lateral secondary visual area to most-lateral primary visual area in the rat cerebral cortex had, on the average, shorter and thicker an AIS than similar cells of projection to the same area but through the callosum; these differences were however non-significant (Box 1). Importantly, AIS of cells of the associative backward projection averaged more synaptic boutons than AIS of cells of projection through the callosum; thus, the synaptic-bouton density per AIS length-unit was higher for AIS of cells of the associative backward

Results also showed that the neurons sited in layer 3 had, on the average, shorter and thinner an AIS than the neurons sited in layer 5, these differences being non-significant except for the AIS-length comparison. Importantly, the AIS of layer 3 cells averaged more synaptic inputs than that of layer 5 cells; thus, the bouton density per AIS length-unit was

**Box 1.** Averaged values of AIS parameters and their comparison as grouped by projection and layer

In human cortex, the length of AIS of layer 3 neurons ranges between 11-47 microns, as calculated by immunocytochemistry to ßIV spectrin, GAT-1 and Na+ channels [76]. However the length of the AIS is not correlated with the size of the perykarion [84,86, but see 94]. By means of regression tests we compared (a) the lengths versus the diameters, and (b) the lengths versus the numbers of received synaptic boutons for the twelve separate AIS presented here. The first comparison revealed a linear and significant relation (p-value = 0.009), by which the AIS diameter variation explained the 52% the AIS length variation. The second comparison, i.e. the lengths versus the numbers of received synaptic boutons, revealed an inverse, non-significant relation (p-value = 0.136). With a non-parametric multiple regression test we compared for the twelve AIS, as taken separately, the numbers of received synaptic boutons versus the layer addresses of the parent cell somata and the projection of the on-going axons. Direct and highly significant correlations emerged from these comparisons (p-values being 9.3x10-5 and 0.00025, respectively). Thus, AIS with a high

projection; these differences were significant (Box 1).

higher for layer 3 cells. These differences were significant (Box 1).

**6.2. Results** 

address.

**Figure 4.** Electron-microscopy photograph of an axonal bouton (between red and yellow arrows) synapsing on the axon initial segment (AIS) of the cell shown in Figure 3. Most other boutons synapsing on this and other AIS contained flat vesicles. AIS was occupied by biotinylated dextran-amide, which gave it its black, grainy aspect. Notwithstanding this filling, see the pre- and postsynaptic densities, which are pointed at by the yellow arrowhead. Scale bar, 0.5 micron.

**Figure 5.** Electron-microscopy serial reconstruction of the axon initial segment (AIS) of the cell shown in Figures 3-4. The cell sited in layer 3 and furnished the associative backward projection to most-lateral (next to the border with lateral secondary cortex) primary visual cortex. The AIS was 22.27-micron-long, averaged a diameter of 0.93 microns and received 28 synaptic boutons (yellow arrows). Note the uneven distribution of these boutons along AIS: boutons were more abundant in the central tier and then in the distal one. Not all studied neurons in [93] and reported here had this bouton distribution along their AIS. The green arrowhead points to the site at which the myelin sheet begun. Scale bar, 5 microns.

## **6.2. Results**

16 Visual Cortex – Current Status and Perspectives

**Figure 4.** Electron-microscopy photograph of an axonal bouton (between red and yellow arrows) synapsing on the axon initial segment (AIS) of the cell shown in Figure 3. Most other boutons synapsing on this and other AIS contained flat vesicles. AIS was occupied by biotinylated dextran-amide, which gave it its black, grainy aspect. Notwithstanding this filling, see the pre- and postsynaptic densities,

16

AIS

**Figure 5.** Electron-microscopy serial reconstruction of the axon initial segment (AIS) of the cell shown in Figures 3-4. The cell sited in layer 3 and furnished the associative backward projection to most-lateral (next to the border with lateral secondary cortex) primary visual cortex. The AIS was 22.27-micron-long, averaged a diameter of 0.93 microns and received 28 synaptic boutons (yellow arrows). Note the uneven distribution of these boutons along AIS: boutons were more abundant in the central tier and then in the distal one. Not all studied neurons in [93] and reported here had this bouton distribution along their AIS. The green arrowhead points to the site at which the myelin sheet begun. Scale bar, 5

which are pointed at by the yellow arrowhead. Scale bar, 0.5 micron.

microns.

As a group, upright principal cells furnishing the associative backward projection from ipsilateral lateral secondary visual area to most-lateral primary visual area in the rat cerebral cortex had, on the average, shorter and thicker an AIS than similar cells of projection to the same area but through the callosum; these differences were however non-significant (Box 1). Importantly, AIS of cells of the associative backward projection averaged more synaptic boutons than AIS of cells of projection through the callosum; thus, the synaptic-bouton density per AIS length-unit was higher for AIS of cells of the associative backward projection; these differences were significant (Box 1).

Results also showed that the neurons sited in layer 3 had, on the average, shorter and thinner an AIS than the neurons sited in layer 5, these differences being non-significant except for the AIS-length comparison. Importantly, the AIS of layer 3 cells averaged more synaptic inputs than that of layer 5 cells; thus, the bouton density per AIS length-unit was higher for layer 3 cells. These differences were significant (Box 1).


**Box 1.** Averaged values of AIS parameters and their comparison as grouped by projection and layer address.

In human cortex, the length of AIS of layer 3 neurons ranges between 11-47 microns, as calculated by immunocytochemistry to ßIV spectrin, GAT-1 and Na+ channels [76]. However the length of the AIS is not correlated with the size of the perykarion [84,86, but see 94]. By means of regression tests we compared (a) the lengths versus the diameters, and (b) the lengths versus the numbers of received synaptic boutons for the twelve separate AIS presented here. The first comparison revealed a linear and significant relation (p-value = 0.009), by which the AIS diameter variation explained the 52% the AIS length variation. The second comparison, i.e. the lengths versus the numbers of received synaptic boutons, revealed an inverse, non-significant relation (p-value = 0.136). With a non-parametric multiple regression test we compared for the twelve AIS, as taken separately, the numbers of received synaptic boutons versus the layer addresses of the parent cell somata and the projection of the on-going axons. Direct and highly significant correlations emerged from these comparisons (p-values being 9.3x10-5 and 0.00025, respectively). Thus, AIS with a high

number of apposed synaptic boutons has up to 90% of probabilities of being part of a cell sited in layer 3, on the one hand, or projecting backwards to the primary visual cortex, on the other hand.

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 19

**Figure 6.** Distribution of synaptic boutons along each of twelve AIS of the present study [93]. Cells sited either in layer 3 or layer 5 and furnished either the associative backward (ipsilateral) projection or the callosal (contralateral) projection, both to primary visual cortex (3 neurons per layer and projection). Each AIS was divided into three segments of equal length (blue bars, closest tiers to the cell somata; green bars, closest tiers to the axon). X-axes show the bouton distribution for each AIS. Y-axes show the number of boutons counted per tier per AIS. Notice that, whilst all six AIS of layer 5 cells had a comparable decreasing pattern, AIS of layer 3 cells had uneven distributions even among cells of the

In order to know on the distribution of synaptic boutons along AIS, we divided each of the twelve AIS reported here in three equal longitudinal tiers and then we assigned boutons to tiers. As averaged for layer addresses and axon projection cell-subgroups, the distribution was decreasing towards the axon but for the subgroup of layer 3 cells of associative backward (ipsilateral) projection; there, the intermediate tier had more boutons (not shown). Most importantly, we compared one another the AIS of the twelve neurons (Figure 6). Then different patterns emerged not only for cells sited in layer 3 of ipsilateral projection but also for cells sited in layer 3 of commissural projection. Still, each cell sited in layer 5 had the decreasing pattern, notwithstanding if the cell was of ipsilateral or contralateral projection. Cells of layer 3 had uneven distributions, which can be tentatively grouped in two presumptive patterns: a 'reel' one, with more boutons apposed to the proximal and distal tiers of the AIS, and a 'barrel' type, with less boutons on the intermediate tier (Figure 6).

The variety in the distribution observed for the principal cells of layer 3 suggests either the presence in this cell group of examples of minor cell subpopulations occurring within those

same axon projection.

Though the number of measured cells in the present study is admittedly low, we believe it is satisfactory for an electron microscopy study. The numbers of synaptic boutons on AIS of upright principal cells of cortico-cortical projection reported here are in general agreement with others of different cortical areas and mammals, if differences between animals, area addresses and axon projections have to be taken into account. Major discrepancies are with cat visual cortex [85] and monkey sensory-motor cortex [95] (Table 2). It has to be said however that the number of boutons revealed for the cat visual cortex in [85] exceeds by a large amount the number of boutons found for the cat visual cortex in [96] and in [86] though in [86] the layer address of the cells must be considered too. More quantitative studies on AIS innervation of subtypes of principal cells of identified axon projection in different animals and cerebral cortical areas are clearly needed.


**Table 2.** Comparison between numbers of synaptic boutons apposed to the AIS membrane as found in [93] and other studies. Note that the boutons seen on AIS of axons projecting to the thalamus and the colliculi are far fewer than those seen on AIS of axons projecting to intratelencephalic projections. Differences concerning types of projection: (\*) cortico-cortical. (a) Contralateral projection from primary visual area to secondary visual area, 16-23 boutons; ipsilateral projection from primary visual area to secondary visual area, 22-28 boutons. (b) Contralateral projection from primary visual area to the border between primary visual area and lateral secondary visual area, 19-22 boutons; ipsilateral projection from lateral secondary visual area to the border between primary visual area and lateral secondary visual area, 28-34 boutons. (c) Contralateral projection from primary visual area to the border between primary visual area and lateral secondary visual area, 15-16 boutons; ipsilateral projection from lateral secondary visual area to the border between primary visual area and lateral secondary visual area, 20- 23 boutons. \*\* Cortico-collicular projection. \*\*\* Cortico-thalamic projection. Difference concerning anatomical subtypes of principal cells: **+** Spiny inverted neuron AIS.

the other hand.

number of apposed synaptic boutons has up to 90% of probabilities of being part of a cell sited in layer 3, on the one hand, or projecting backwards to the primary visual cortex, on

Though the number of measured cells in the present study is admittedly low, we believe it is satisfactory for an electron microscopy study. The numbers of synaptic boutons on AIS of upright principal cells of cortico-cortical projection reported here are in general agreement with others of different cortical areas and mammals, if differences between animals, area addresses and axon projections have to be taken into account. Major discrepancies are with cat visual cortex [85] and monkey sensory-motor cortex [95] (Table 2). It has to be said however that the number of boutons revealed for the cat visual cortex in [85] exceeds by a large amount the number of boutons found for the cat visual cortex in [96] and in [86] though in [86] the layer address of the cells must be considered too. More quantitative studies on AIS innervation of subtypes of principal cells of identified axon projection in

[96] Cat Visual 2/3 —unknown 18 24 (average) [85] Cat Visual 2/3 — unknown 3 42-44 [86] Cat Area 17 2/3 — \*a 18 16-28\*a

[93] Rat Visual 2/3 — \*b 6 19-34\*b [95] Monkey Sensory-motor 5 — unknown 8 2-26 [28] Rat Visual 5/6 — <sup>+</sup> 11 11-37+

[97] Rabbit Visual 5 — \*\* 1 10\*\* [86] Cat Visual 5 — \*\*\* 10 1-5\*\*\*

**Table 2.** Comparison between numbers of synaptic boutons apposed to the AIS membrane as found in [93] and other studies. Note that the boutons seen on AIS of axons projecting to the thalamus and the colliculi are far fewer than those seen on AIS of axons projecting to intratelencephalic projections. Differences concerning types of projection: (\*) cortico-cortical. (a) Contralateral projection from primary visual area to secondary visual area, 16-23 boutons; ipsilateral projection from primary visual area to secondary visual area, 22-28 boutons. (b) Contralateral projection from primary visual area to the border between primary visual area and lateral secondary visual area, 19-22 boutons; ipsilateral projection from lateral secondary visual area to the border between primary visual area and lateral secondary visual area, 28-34 boutons. (c) Contralateral projection from primary visual area to the border between primary visual area and lateral secondary visual area, 15-16 boutons; ipsilateral projection from lateral secondary visual area to the border between primary visual area and lateral secondary visual area, 20- 23 boutons. \*\* Cortico-collicular projection. \*\*\* Cortico-thalamic projection. Difference concerning

**projection Cells Boutons** 

8 2-52

2/3 — callosal (2 cells) and unknown

& primary 5/6 — \*c 6 15-23\*c

different animals and cerebral cortical areas are clearly needed.

**Reference Animal Type of cortex Layer —**

[95] Monkey Sensory-motor

[93] Rat Visual, secondary

anatomical subtypes of principal cells: **+** Spiny inverted neuron AIS.

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 19

**Figure 6.** Distribution of synaptic boutons along each of twelve AIS of the present study [93]. Cells sited either in layer 3 or layer 5 and furnished either the associative backward (ipsilateral) projection or the callosal (contralateral) projection, both to primary visual cortex (3 neurons per layer and projection). Each AIS was divided into three segments of equal length (blue bars, closest tiers to the cell somata; green bars, closest tiers to the axon). X-axes show the bouton distribution for each AIS. Y-axes show the number of boutons counted per tier per AIS. Notice that, whilst all six AIS of layer 5 cells had a comparable decreasing pattern, AIS of layer 3 cells had uneven distributions even among cells of the same axon projection.

In order to know on the distribution of synaptic boutons along AIS, we divided each of the twelve AIS reported here in three equal longitudinal tiers and then we assigned boutons to tiers. As averaged for layer addresses and axon projection cell-subgroups, the distribution was decreasing towards the axon but for the subgroup of layer 3 cells of associative backward (ipsilateral) projection; there, the intermediate tier had more boutons (not shown). Most importantly, we compared one another the AIS of the twelve neurons (Figure 6). Then different patterns emerged not only for cells sited in layer 3 of ipsilateral projection but also for cells sited in layer 3 of commissural projection. Still, each cell sited in layer 5 had the decreasing pattern, notwithstanding if the cell was of ipsilateral or contralateral projection. Cells of layer 3 had uneven distributions, which can be tentatively grouped in two presumptive patterns: a 'reel' one, with more boutons apposed to the proximal and distal tiers of the AIS, and a 'barrel' type, with less boutons on the intermediate tier (Figure 6).

The variety in the distribution observed for the principal cells of layer 3 suggests either the presence in this cell group of examples of minor cell subpopulations occurring within those

studied in [93], or activity-dependent AIS plasticity, or both. Ion channels composition at the AIS can vary considerably across different neuronal types [74] and diversity in the AIS length and position in the axon may also underlie interneuron variation in firing properties. For example, classes of retinal ganglion cells with different visual properties have initial segments that differ in length, and in their position within the axon [98]. In the avian magnocellularis and laminaris nuclei, which are respectively the second- and third-order nuclei in the auditory pathway [99], the length and the location of the AIS vary with the tuning frequency of neurons [100]. Activity-dependent AIS plasticity has been observed in cultured hippocampal neurons. Increasing chronically neuronal activity over 48 hours resulted in a significant shift in AIS position with the entire structure (ankyrin G, ßIV spectrin, and Na+ channels) moving distally away from the soma [75]. Importantly, hippocampal AIS relocation is a bidirectional phenomenon because the AIS can shift proximally after neurons are returned to baseline activity conditions. In this way, on-going neuronal activity can fine-tune AIS position. Moreover, a lack of auditory input causes a change in AIS length of neurons of chick auditory nucleus magnocellularis [101]. All these studies showed that the AIS plasticity is coupled with changes in neuronal excitability. The distinct length of the AIS of hippocampal pyramidal neurons (≈30 microns) and the neurons of the nucleus magnocellularis (≈10 microns) has been suggested to be the cause of these different plastic changes in the AIS. Indeed, cell type-based variability in AIS plasticity is seen within hippocampal cultures, where GAD 65-expressing interneurons display little or no shift in AIS location upon high potassium stimulation [75]. Dopaminergic neurons in dissociated cultures of rat olfactory bulb show inverse AIS plasticity: their initial segments move proximally after 48 hours depolarization.

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 21

, Juan C. Chara, Juan L. Mendizabal-Zubiaga and Concepción Reblet

not of layer 3. It might be that neurons of layer 3 can be subdivided in subgroups within the groups of cortico-cortical ipsilateral or contralateral projection we have considered in the study. These subgroups might possess undisclosed structural and functional cell features

J.L.B.-L. and C.R. wish to lovingly dedicate this work to Laína H.B, Julia B.R. and Juan H.S. for leaving their mark on us ("*Todo pasa y todo queda, / pero lo nuestro es pasar, / pasar haciendo caminos, / caminos sobre el mar // … // Caminante, son tus huellas / el camino y nada más; / caminante, no hay camino, / se hace camino al andar. // Al andar se hace camino / y al volver la vista atrás / se ve la senda que nunca / se ha de volver a pisar. // Caminante no hay camino / sino estelas en* 

[1] Molnár Z, Cheung, AFP. Classification of subpopulations of layer V pyramidal projection neurons. Neuroscience Research 2006; 55(2) 105–115,

[2] Beaulieu C. Numerical data on neocortical neurons in adult rat, with special reference to the GABA population. Brain Research 1993; 609(1-2) 284–292, doi:10.1016/0006-

[3] DeFelipe J, Alonso-Nanclares L, Arellano JI. Microstructure of the neocortex: comparative aspects. Journal of Neurocytology 2002; 31(3-5) 299-316,

[4] DeFelipe J. Cortical interneurons: from Cajal to 2001. Progress in Brain Research 2002;

[5] Somogyi P, Tamás G, Luján R, Buhl EH. Salient features of synaptic organisation in the cerebral cortex. Brain Research Reviews 1998; 26(2-3)113–35, doi:10.1016/S0165-

[6] Burkhalter A. Many specialists for suppressing cortical excitation. Frontiers in

[7] DeFelipe J, González-Albo MC, Del Río MR, Elston GN. Distribution and patterns of connectivity of interneurons containing calbindin, calretinin and parvalbumin in visual

associated to permanent or transitory synaptic-bouton distributions along the AIS.

*The University of the Basque Country (UPV/EHU), Campus of Leioa (Bizkaia), Spain* 

This work was funded by grants GIU07/14 and UFI11/41 of UPV/EHU.

*Department of Neurosciences, School of Medicine and Dentistry,* 

*la mar...* ", Antonio Machado, Proverbios y Cantares, 1909).

136(Chapter7) 215–238, doi:10.1016/S0079-6123(02)36019-9

Neuroscience 2008; 2(2) 155-167, doi:10.3389/neuro.01.026.2008

doi:10.1016/j.neures.2006.02.008

doi:10.1023/A:1024130211265

**Author details** 

José L. Bueno-López\*

**Acknowledgement** 

**8. References** 

8993(93)90884-P

0173(97)00061-1

Corresponding Author

 \*

Although these plastic changes have not been shown for neocortical neurons to the best of our knowledge, nor is clear to which extent are associated with shifts on the distribution of presynaptic boutons apposed to the membrane of the AIS, it will be fascinating in the future to see how different types of neuron of diverse brain regions use different forms of AIS plasticity in response to perturbations in their electrical activity. Nonetheless, it follows from our study that at least for associative and commissural projections to primary visual cortex in the rat, the innervation of AIS of principal cells of typical somatodendritic orientation is steadier in layer 5 than in layer 3.

## **7. Conclusion**

By presenting data on parameters such as the length and diameter of the AIS and, of no lesser importance, the number and distribution of presynaptic boutons apposed to the AIS membrane, this study advances the knowledge on the control of membrane potential and initiation of axon potential of visual cortex neurons, and particularly of the principal cells with cortico-cortical axon projection from layers 3 and 5 to primary visual area. It unveils that the bouton density on the AIS of principal cells would correlate significantly with the layer address and axon projection of the parent neuron rather than with the length and thickness of AIS proper. The present study additionally shows that there was a repeated decrease in the number of synaptic boutons apposed along the AIS of neurons of layer 5 but not of layer 3. It might be that neurons of layer 3 can be subdivided in subgroups within the groups of cortico-cortical ipsilateral or contralateral projection we have considered in the study. These subgroups might possess undisclosed structural and functional cell features associated to permanent or transitory synaptic-bouton distributions along the AIS.

## **Author details**

20 Visual Cortex – Current Status and Perspectives

move proximally after 48 hours depolarization.

steadier in layer 5 than in layer 3.

**7. Conclusion** 

studied in [93], or activity-dependent AIS plasticity, or both. Ion channels composition at the AIS can vary considerably across different neuronal types [74] and diversity in the AIS length and position in the axon may also underlie interneuron variation in firing properties. For example, classes of retinal ganglion cells with different visual properties have initial segments that differ in length, and in their position within the axon [98]. In the avian magnocellularis and laminaris nuclei, which are respectively the second- and third-order nuclei in the auditory pathway [99], the length and the location of the AIS vary with the tuning frequency of neurons [100]. Activity-dependent AIS plasticity has been observed in cultured hippocampal neurons. Increasing chronically neuronal activity over 48 hours resulted in a significant shift in AIS position with the entire structure (ankyrin G, ßIV spectrin, and Na+ channels) moving distally away from the soma [75]. Importantly, hippocampal AIS relocation is a bidirectional phenomenon because the AIS can shift proximally after neurons are returned to baseline activity conditions. In this way, on-going neuronal activity can fine-tune AIS position. Moreover, a lack of auditory input causes a change in AIS length of neurons of chick auditory nucleus magnocellularis [101]. All these studies showed that the AIS plasticity is coupled with changes in neuronal excitability. The distinct length of the AIS of hippocampal pyramidal neurons (≈30 microns) and the neurons of the nucleus magnocellularis (≈10 microns) has been suggested to be the cause of these different plastic changes in the AIS. Indeed, cell type-based variability in AIS plasticity is seen within hippocampal cultures, where GAD 65-expressing interneurons display little or no shift in AIS location upon high potassium stimulation [75]. Dopaminergic neurons in dissociated cultures of rat olfactory bulb show inverse AIS plasticity: their initial segments

Although these plastic changes have not been shown for neocortical neurons to the best of our knowledge, nor is clear to which extent are associated with shifts on the distribution of presynaptic boutons apposed to the membrane of the AIS, it will be fascinating in the future to see how different types of neuron of diverse brain regions use different forms of AIS plasticity in response to perturbations in their electrical activity. Nonetheless, it follows from our study that at least for associative and commissural projections to primary visual cortex in the rat, the innervation of AIS of principal cells of typical somatodendritic orientation is

By presenting data on parameters such as the length and diameter of the AIS and, of no lesser importance, the number and distribution of presynaptic boutons apposed to the AIS membrane, this study advances the knowledge on the control of membrane potential and initiation of axon potential of visual cortex neurons, and particularly of the principal cells with cortico-cortical axon projection from layers 3 and 5 to primary visual area. It unveils that the bouton density on the AIS of principal cells would correlate significantly with the layer address and axon projection of the parent neuron rather than with the length and thickness of AIS proper. The present study additionally shows that there was a repeated decrease in the number of synaptic boutons apposed along the AIS of neurons of layer 5 but José L. Bueno-López\* , Juan C. Chara, Juan L. Mendizabal-Zubiaga and Concepción Reblet *Department of Neurosciences, School of Medicine and Dentistry, The University of the Basque Country (UPV/EHU), Campus of Leioa (Bizkaia), Spain* 

## **Acknowledgement**

This work was funded by grants GIU07/14 and UFI11/41 of UPV/EHU.

J.L.B.-L. and C.R. wish to lovingly dedicate this work to Laína H.B, Julia B.R. and Juan H.S. for leaving their mark on us ("*Todo pasa y todo queda, / pero lo nuestro es pasar, / pasar haciendo caminos, / caminos sobre el mar // … // Caminante, son tus huellas / el camino y nada más; / caminante, no hay camino, / se hace camino al andar. // Al andar se hace camino / y al volver la vista atrás / se ve la senda que nunca / se ha de volver a pisar. // Caminante no hay camino / sino estelas en la mar...* ", Antonio Machado, Proverbios y Cantares, 1909).

## **8. References**


<sup>\*</sup> Corresponding Author

areas of the occipital and temporal lobes of the macaque monkey. Journal of Comparative Neurology 1999; 412(3) 515–526, doi:10.1002/(SICI)1096- 9861(19990927)412:3<515::AID-CNE10>3.0.CO;2-1

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 23

Cajal on the Cerebral Cortex: An Annotated Translation on the Complete Writings (History of Neuroscience; Nº 1). New York: Oxford University Press; 1988, p557–621. [23] Bueno-López JL, Reblet C, López-Medina A, Gómez-Urquijo S, Grandes P, Gondra J, Hennequet L. Targets and laminar distribution of projection neurons with inverted morphology in rabbit cortex. European Journal of Neuroscience 1991; 3(5) 415–430,

[24] Larkman A, Mason A. Correlations between morphology and electrophysiology of pyramidal neurons in slices of rat visual cortex. I. Establishment of cell classes. Journal

[25] DeFelipe J, Fariñas I. The pyramidal neuron of the cerebral cortex: morphological and chemical characteristics of the synaptic inputs. Progress in Neurobiology 1992; 39(6)

[26] Innocenti GM. (1986). General Organization of Callosal Connections in the Cerebral Cortex. In: Jones EG, Peters A (eds) Cerebral cortex (vol. 5) Sensory-motor areas and

[27] Reblet C, López-Medina A, Gómez-Urquijo SM, Bueno-López JL. Widespread horizontal connections arising from layer 5/6 border inverted cells in rabbit visual cortex. European Journal of Neuroscience 1992; 4(3) 221–234, doi:10.1111/j.1460-

[28] Mendizabal-Zubiaga JL, Reblet C, Bueno-López JL. The underside of the cerebral cortex: layer V/VI spiny inverted neurons. Journal of Anatomy 2007; 211(2) 223–236,

[29] Casanovas-Aguilar C, Christensen MK, Reblet C, Martínez-García F, Pérez-Clausell J, Bueno-López JL. Callosal neurons give rise to zinc-rich boutons in the rat visual cortex.

[30] Casanovas-Aguilar C, Reblet C, Pérez-Clausell J, Bueno-López JL. Zinc-Rich afferents to the rat neocortex: projections to the visual cortex traced with intracerebral selenite injections. Journal of Chemical Neuroanatomy 1998; 72(2) 97–109, doi:10.1016/S0891-

[31] Felleman DJ, Van Essen DC. Distributed hierarchical processing in the primate cerebral

[32] Coogan TA, Burkhalter A. Hierarchical organization of areas in rat visual cortex.

[33] Gutiérrez-Ibarluzea I, Acera-Osa A, Mendizabal-Zubiaga JL, Arana-Arri A, Bueno-López JL, Reblet C. Morphology and laminar distribution of cortico-claustral neurons in different areas of the rabbit cerebral cortex. European Journal of Anatomy 1999; 3(2)

[34] Olavarría J, Van Sluyters RC. Organization and postnatal development of callosal connections in the visual cortex of the rat. Journal of Comparative Neurology 1985;

[35] Innocenti M, Manger PR, Masiello I, Colin I, Tettoni L. Architecture and callosal connections of visual areas 17, 18, 19 and 21 in the ferret (Mustela putorius). Cerebral

cortex. Cerebral Cortex 1991; 1(1) 1-47, doi:10.1093/cercor/1.1.1

Journal of Neuroscience 1993; 13(9) 3749-3772.

239(1) 1–26, doi:10.1002/cne.902390102

Cortex 2002; 12(4) 411-422, doi:10.1093/cercor/12.4.411

aspects of cortical connectivity. Plenum Press, New York; 1986, p291–353.

doi:10.1111/j.1460-9568.1991.tb00829.x

of Neuroscience 1990; 10(5) 1407–1414.

doi:10.1111/j.1469-7580.2007.00779.x

NeuroReport 1995; 6(3) 497-500.

563–607.

9568.1992.tb00870.x

0618(98)00035-0

101-109.


Cajal on the Cerebral Cortex: An Annotated Translation on the Complete Writings (History of Neuroscience; Nº 1). New York: Oxford University Press; 1988, p557–621.

[23] Bueno-López JL, Reblet C, López-Medina A, Gómez-Urquijo S, Grandes P, Gondra J, Hennequet L. Targets and laminar distribution of projection neurons with inverted morphology in rabbit cortex. European Journal of Neuroscience 1991; 3(5) 415–430, doi:10.1111/j.1460-9568.1991.tb00829.x

22 Visual Cortex – Current Status and Perspectives

doi:10.1002/cne.901600204

doi:10.1038/nrn1519

Research 1977; 136(-) 345–50.

9861(19990927)412:3<515::AID-CNE10>3.0.CO;2-1

Cortex 2009; 19(1) 41-54, doi:10.1093/cercor/bhn057

Cerebral Cortex 2007; 17(9) 2060-2071, doi:10.1093/cercor/bhl114

1986; 17(4) 991-1009, doi:10.1016/0306-4522(86)90075-8

2003; 13(1) 5–14, doi:10.1093/cercor/13.1.5

Hopkins Hospital 1965; 117(-) 228–250.

Neurology 1993; 327(3) 398–415, doi:10.1002/cne.903270307

Nº 1). New York: Oxford University Press; 1988, p23–54.

Research Program Bulletin 1974; 12(3) 306–510.

areas of the occipital and temporal lobes of the macaque monkey. Journal of Comparative Neurology 1999; 412(3) 515–526, doi:10.1002/(SICI)1096-

[8] Szentágothai J, Arbib MA. Conceptual models of neural organization. Neurosciences

[9] Jones EG. Varieties and distribution of non-pyramidal cells in the somatic sensory cortex of the squirrel monkey. Journal of Comparative Neurology 1975; 160(2) 205–267,

[10] Somogyi P. A specific 'axo-axonal' interneuron in the visual cortex of the rat. Brain

[11] Inda MC, DeFelipe J, Muñoz A. Morphology and distribution of chandelier cell axon terminals in the mouse cerebral cortex and claustroamygdaloid complex. Cerebral

[12] Markram H, Toledo-Rodríguez M, Wang Y, Gupta A, Silberberg G, Wu C. Interneurons of the neocortical inhibitory system. Nature Reviews Neuroscience 2004; 5(-) 793–807,

[13] Inda MC, DeFelipe J, Munoz A. The distribution of chandelier cell axon terminals that express the GABA plasma membrane transporter GAT-1 in the human neocortex.

[14] Howard A, Tamás G, Soltesz I. Lighting the chandelier: new vistas for axo-axonic cells.

[15] DeFelipe J, Hendry SHC, Jones EG. A correlative electron microscopic study of basket cells and large gabaergic neurons in the monkey sensory-motor cortex. Neuroscience

[16] Kisvárday ZF, Beaulieu C, Eysel UT. Network of gabaergic large basket cells in cat visual cortex (area 18): implication for lateral disinhibition. Journal of Comparative

[17] Freund T. Interneuron diversity series: rhythm and mood in perisomatic inhibition. Trends in Neurosciences 2003; 26(9) 489-495, doi:10.1016/S0166-2236(03)00227-3 [18] Thomson AM, Bannister AP. Interlaminar connections in the neocortex. Cerebral Cortex

[19] Ramón y Cajal S (1891) On the Structure of the Cerebral Cortex of Certain Mammals. (La Cellule 7: 125–176). In: DeFelipe J, Jones EG (trans. and eds.) Cajal on the Cerebral Cortex: An Annotated Translation on the Complete Writings (History of Neuroscience;

[20] Ramón y Cajal, S. (1911). Histologie du Systéme Nerveux de l'Homme et des Vertebrés.

[21] Van der Loos H. The 'improperly' oriented pyramidal cell in the cerebral cortex and its possible bearing on problems of neuronal growth and cell orientation. Bulletin Johns

[22] DeFelipe J, Jones EG. The Functional Histology of the Cerebral Cortex and the Continuing Relevance of Cajal's Observations. In: DeFelipe J, Jones EG (trans. and eds.)

Paris: Maloine (Madrid: C.S.I.C. Facsimile Edition 1972). Chapters 24–35.

Trends in Neurosciences 2005; 28(5) 310–316, doi:10.1016/j.tins.2005.04.004


[36] Molyneaux BJ, Arlotta P, Fame RM, MacDonald JL, MacQuarrie KL, Macklis JD. Novel subtype-specific genes identify distinct subpopulations of callosal projection neurons. Journal of Neuroscience 2009; 29(39) 12343–12354, doi:10.1523/ JNEUROSCI.6108- 08.2009

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 25

[50] Veinante P, Lavallé P, Deschênes M. Corticothalamic projections from layer V of the vibrissal barrel cortex in the rat. Journal of Comparative Neurology 2000; 424(2) 197-

[51] Guillery RW, Sherman SM. Thalamic relay functions and their role in corticocortical communication: generalizations from the visual system. Neuron 2002; 33(2) 163-175,

[52] Hallman LE, Schofield BR, Lin CS. Dendritic morphology and axon collaterals of corticotectal, cortico-pontine, and callosal neurons in layer V of the primary visual cortex of the hooded rat. Journal of Comparative Neurology 1990; 272(1) 149–160,

[53] Otsuka T, Kawaguchi Y. Cortical inhibitory cell types differentially form intralaminar and interlaminar subnetworks with excitatory neurons. Journal of Neuroscience 2009;

[54] Brown SP, Hestrin S. Intracortical circuits of pyramidal neurons reflect their long-range

[55] Kiritani T, Wickersham IR, Seung HS, Shepherd GMG. Hierarchical connectivity and connection-specific dynamics in the corticospinal-corticostriatal microcircuit in mouse motor cortex. Journal of Neuroscience 2012; 32(14) 4992-5001, doi:10.1523/

[56] Morishima M, Morita K, Kubota Y, Kawaguchi Y. Highly differentiated projectionspecific cortical subnetworks. Journal of Neuroscience 2011; 31(28) 10380-10391,

[57] Otsuka T, Kawaguchi Y. Cell diversity and connection specifity between callosal projection neurons in the frontal cortex. Journal of Neuroscience 2011; 31(10) 3862-3870,

[58] Binzegger T, Douglas RJ, Martin KAC. A quantitative map of the circuit of cat primary visual cortex. Journal of Neuroscience 2004; 24(39) 8441– 8453, doi:10.1523/

[59] Guillery RW. Anatomical pathways that link perception and action. In: Progress in Brain Research. Elsevier BV. 2005; 149(Chapter 17) 235-256, doi:10.1016/S0079-

[60] Thomson A.M. Neocortical layer 6, a review. Frontiers in Neuroanatomy 2010; 4(13) 4-

[61] Molnár G, Oláh S, Komlósi G, Füle M, Szabadics J, Varga C, Barzo P, Tamás, G. Complex events initiated by individual spikes in the human cerebral cortex. PLoS

[62] Woodruff A, Yuste R. State-dependent function of neocortical chandelier cells. Journal of Neuroscience 2011; 31(49), 17872-17886, doi:10.1523/jneurosci.3894-11.2011 [63] Gilbert CD, Wiesel TN. Columnar specificity of intrinsic horizontal and corticocortical connections in cat visual cortex. Journal of Neuroscience 1989; 9(7), 2432–2442. [64] Kisvárday ZF, Eysel UT. Cellular organization of reciprocal patchy networks in layer III of cat visual cortex area 17. Neuroscience 1992; 46(2) 275–286, doi: 10.1016/0306-

Biology September 2008; 6(9):e222, doi:10.1371/journal.pbio.0060222

axonal targets. Nature 2009; 457(-) 1133–1136, doi:10.1038/nature07658

204, doi:0.1002/1096-9861(20000821)424:2<197::AID-CNE1>3.0.CO;2-6

29(34) 10533-10540, doi:10.1523/jneurosci.2219-09.2009

doi:10.1016/S0896-6273(01)00582-7

doi:10.1002/cne.902720111

jneurosci.4759-11.2012

jneurosci.1400-04.2004

6123(05)49017-2

4522(92)90050-C

doi:10.1523/jneurosci.0772-11.2011

doi:10.1523/jneurosci.5795-10.2011

14, doi:10.3389/fnana.2010.00013


[50] Veinante P, Lavallé P, Deschênes M. Corticothalamic projections from layer V of the vibrissal barrel cortex in the rat. Journal of Comparative Neurology 2000; 424(2) 197- 204, doi:0.1002/1096-9861(20000821)424:2<197::AID-CNE1>3.0.CO;2-6

24 Visual Cortex – Current Status and Perspectives

doi:10.1002/cne.10769

doi:10.1002/cne.902630408

doi:10.1093/cercor/8.7.602

doi:10.3389/fnana.2010.00142

518, doi:10.1002/cne.903390404

2005; 482(1) 17-32, doi:10.1002/cne.20428

1996; 709(2) 311–315, doi:10.1016/0006-8993(95)01333-4

Cerebral Cortex 1996; 6(6) 759–770, doi:10.1093/cercor/6.6.759

2012; 32(17) 5990-5999, doi:10.1523/jneurosci.5717-11.2012

9861(19980126)390:4<515::AID-CNE5>3.0.CO;2-3

08.2009

[36] Molyneaux BJ, Arlotta P, Fame RM, MacDonald JL, MacQuarrie KL, Macklis JD. Novel subtype-specific genes identify distinct subpopulations of callosal projection neurons. Journal of Neuroscience 2009; 29(39) 12343–12354, doi:10.1523/ JNEUROSCI.6108-

[37] Arimatsu Y, Ishida M, Kaneko T, Ichinose S, Omori A. Organization and Development of Corticocortical Associative neurons expressing the orphan nuclear receptor Nurr1.

[39] Veinante P, Deschênes M. Single-cell study of motor cortex projections to the barrel field in rats. Journal of Comparative Neurology 2003; 464(1) 98 –103,

[40] Wilson CJ. Morphology and synaptic connections of crossed corticostriatal neurons in the rat. Journal of Comparative Neurology 1987; 263(4) 567-580,

[41] Levesque M, Charara A, Gagnon S, Parent A, Deschênes M. Corticostriatal projections from layer V cells in rat are collaterals of long range corticofugal axons. Brain Research

[42] Levesque M, Gagnon S, Parent A, Deschênes M. Axonal arborizations of corticostriatal and corticothalamic fibers arising from the second somatosensory area in the rat.

[43] Levesque M, Parent A. Axonal arborization of corticostriatal and corticothalamic fibers arising from prelimbic cortex in the rat. Cerebral Cortex 1998; 8(7) 602–613,

[44] Reiner A, Hart NM, Lei W, Deng Y. Corticostriatal projection neurons—dichotomous types and dichotomous functions. Frontiers in Neuroanatomy 2010. 4(Article142) 1-15,

[45] Shoemaker LD, Arlotta P. Untangling the Cortex: Advances in Understanding Specification and Differentiation of Corticospinal Motor Neurons. BioEssays 2010; 32(3) 197–206. Available online 27 January 2010; http://dx.doi.org/10.1002/bies.200900114 [46] Kasper EM, Larkman AU, Lübke J, Blakemore C. Pyramidal neurons in layer 5 of the rat visual cortex. III. Differential maturation of axon targeting, dendritic morphology and electrophysiological properties. Journal of Comparative Neurology 1994; 339(4) 495–

[47] Kita T, Kita H. The subthalamic nucleus is one of multiple innervation sites for longrange corticofugal axons: a single-axon tracing study in the rat. Journal of Neuroscience

[48] Jones E.G. The thalamic matrix and thalamocortical synchrony. Trends in

[49] Rockland KS. Convergence and branching patterns of round, type II corticopulvinar axons. Journal of Comparative Neurology 1998; 390(4) 515-536, doi:10.1002/(SICI)1096-

Neurosciences 2001; 24(10), 595-601, doi:10.1016/S0166-2236(00)01922-6

Journal of Comparative Neurology 2003; 466(2) 180–196, doi:10.1002/cne.10875 [38] Mitchel BD, Macklis JD. Large-scale maintenance of dual projections by callosal and frontal cortical projection neurons in adult mice. Journal of Comparative Neurology

	- [65] Burkhalter A. Intrinsic connections of rat primary visual cortex: laminar organization of axonal projections. Journal of Comparative Neurology 1989; 279(2) 171-186, doi: 10.1002/cne.902790202

Projections, Partaken Circuits and Axon Initial Segments of Cortical Principal Neurons 27

[80] Perez-Reyes E. Molecular physiology of low-voltage-activated t-type calcium channels.

[81] Bender KJ, Ford CP, Trussell LO. Dopaminergic modulation of axon initial segment calcium channels regulates action potential initiation. Neuron 2010; 68(3) 500 –511,

[82] DeFelipe J, Hendry SH, Hashikawa T, Jones EG. Synaptic relationships of serotoninimmunoreactive terminal baskets on GABA neurons in the cat auditory cortex. Cerebral

[83] Dianne A, Cruz MS, Stephen M, Eggan BS, Efrain C, Azmitia Ph.D, David A, Lewis MD. Serotonin1A receptors at the axon initial segment of prefrontal pyramidal neurons in schizophrenia. American Journal of Psychiatry 2004; 161(4) 739–742, doi:

[84] Peters A, Proskauer CC, Kaiserman-Abramof IR. The small pyramidal neuron of the rat cerebral cortex. The axon hillock and initial segment. Journal of Cell Biology 1968; 39(3)

[85] Somogyi P, Freund TF, Cowey A. The axo-axonic interneuron in the cerebral cortex of the rat, cat and monkey. Neuroscience 1982; 7(11) 2577–607, doi:10.1016/0306-

[86] Fariñas I, DeFelipe J. Patterns of synaptic input on corticocortical and corticothalamic cells in the cat visual cortex. II. The axon initial segment. Journal of Comparative

[87] Ogawa Y, Horresh I, Trimmer JS, Bredt DS, Peles E, Rasband MN. Postsynaptic density-93 Clusters Kv1 channels at axon initial segments independently of Caspr2. Journal of

[88] Davis JQ, Bennet V. Ankyrin binding activity shared by the neurofascin/L1/NrCAM family of nervous system cell adhesion molecules. Journal of Biological Chemistry 1994;

[89] Hedstrom KL, Ogawa Y, Rasband MN. AnkiyrinG is required for maintenance of the axon initial segment and neuronal plasticity. Journal of Cell Biology 2008; 183(4) 635-

[90] Meeks JP, Mennerick S. Action potential initiation and propagation in CA3 pyramidal axons. Journal of Neurophysiology 2007; 97(5) 3460-3472, doi:10. 1152/jn.01288. 2006 [91] Schmidt-Hieber C, Bischofberger J. Fast sodium channel gating supports localized and efficient axonal action potential initiation. Journal of Neuroscience 2010; 30(30) 10233-

[92] Krees GJ, Dowling MJ, Meeks JP, Mennerick S. High threshold, proximal initiation, and slow conduction velocity of action potentials in dentate granule neuron mossy fibers.

[93] Chara JC. Estudio paramétrico ultrastructural del segmento inicial del axon de las neuronas piramidales supra e infragranulares con proyección axonal identificada en la corteza visual de la rata. Tesis doctoral. Universidad del País Vasco (UPV/EHU); 2010.

Journal of Neurophysiology 2008; 100(1) 281–291, doi:10.1152/jn.90295.2008

Neuroscience 2008; 28(22) 5731–5739, doi: 10.1523/jneurosci.4431-07.2008

Neurology 1991; 304(1) 70-77, doi: 10.1002/cne.903040106

Physiological Reviews 2003; 83(1) 117-161, doi:10.1152/physrev.00018.2002

10.1016/j.neuron.2010.09.026

Cortex 1991; 1(2) 117-133.

10.1176/appi.ajp.161.4.739

604-619.

4522(82)90086-0

269(44) 27163-27166.

640, doi:10.1083/jcb.200806112

10242, doi:10.1523/jneurosci.6335-09.2010


[80] Perez-Reyes E. Molecular physiology of low-voltage-activated t-type calcium channels. Physiological Reviews 2003; 83(1) 117-161, doi:10.1152/physrev.00018.2002

26 Visual Cortex – Current Status and Perspectives

10.1002/cne.902790202

2007; 17(13) R496-R500.

45-56, doi:10.1038/nrn2044

2003; 421(6925) 844-848.

1002, doi:10.1038/nn.2359

08.2008

16055, doi:10.1523/jneurosci.4064-11.2011

2006; 103(8) 2920-2925, doi:10.1073/pneas.05111971.03

[65] Burkhalter A. Intrinsic connections of rat primary visual cortex: laminar organization of axonal projections. Journal of Comparative Neurology 1989; 279(2) 171-186, doi:

[66] Markov NT, Misery P, Falchier A, Lamy C, Vezoli J, Quilodran R, Gariel MA, Giroud P, Ercsey-Ravasz M, Pilaz LJ, Huissoud C, Barone P, Dehay C, Toroczkai Z, Van Essen DC, Kennedy H, Knoblauch K. Weight consistency specifies regularities of macaque cortical

[68] Galarreta M, Hestrin S. A network of fast-spiking cells in the neocortex connected by

[69] Fukuda T, Kosaka T, Singer W, Galuske RAW. Gap junctions among dendrites of cortical GABAergic neurons establish a dense and widespread intercolumnar network. Journal of Neuroscience 2006; 26(13) 3434 –3443, doi: 10.1523/jneurosci.4076-05.2006 [70] Simon A, Oláh S, Molnár G, Szabadics J, Tamás G. Gap-junctional coupling between neurogliaform cells and various interneuron types in the neocortex. Journal of

[71] Bartos M, Vida I, Jonas P. (2007). Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. Nature Reviews Neuroscience 2007; 8(-)

[72] Klausberger T, Magill PJ, Márton LF, Roberts JD, Cobden PM, Buzsáki G, Somogyi P. Brain-state- and cell-type-specific firing of hippocampal interneurons in vivo. Nature

[73] Grubb MS, Shu Y, Kuba H, Rasband MN, Wimmer VC, Bender, KJ. Short- and longterm plasticity at the axon initial segment. Journal of Neuroscience 2011; 31(45), 16049-

[74] Lorincz A, Nusser Z. Cell-type-dependent molecular composition of the axon initial segment. Journal of Neuroscience 2008; 28(53) 14329-14340, doi: 10.1523/jneurosci.4833-

[75] Grubb MS, Burrone J. Activity-dependent relocation of the axon initial segment finetunes neuronal excitability. Nature 2010; 465(-) 1070-1076, doi:10.1038/nature09160 [76] Inda MC, DeFelipe J, Muñoz A. Voltage-gated ion channels in the axon initial segment of human cortical pyramidal cells and their relationship with chandelier cells. PNAS

[77] Hu W, Tian C, Yang M, Hou H, Shu Y. Distinct contributions of Nav1.6 and Nav1.2 in action potential initiation and backpropagation. Nature Neuroscience 2009; 12(-) 996-

[78] Palmer LM, Stuart GJ. Site of action potential initiation in layer 5 pyramidal neurons. Journal of Neuroscience 2006; 26(6) 1854–1863, doi: 10.1523/jneurosci.4812-05.2006 [79] Shu Y, Duque A, Yu Y, Haider B, McCormick, D. Properties of action-potential initiation in neocortical pyramidal cells: evidence from whole cell axon recordings.

Journal of Neurophysiology 2007; 97(1) 746-760, doi:10.1152/jn.00922.2006

networks. Cerebral Cortex 2011; 21(6) 1254—1272, doi: 10.1093/cercor/bhq201 [67] Douglas RJ, Martin KAC. Recurrent neuronal circuits in the neocortex. Current Biology

electrical synapses. Nature 1999; 402(-), 72–75, doi:10.1038/47029

Neuroscience 2005; 25(27) 6278-6285, doi:10.1523/jneurosci.1431-05.2005

	- [94] Sloper JJ, Powell TPS. A study of the axon initial segment and proximal axon of neurons in the primate motor cortex y somatic sensory cortices. Philosophical Transactions of the Royal Society of London 1979. Series B, Biological Sciences 285(1006) 173-197.

**Chapter 2** 

© 2012 Brewer and Barton, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

© 2012 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution,

**Visual Field Map Organization** 

**in Human Visual Cortex** 

Alyssa A. Brewer and Brian Barton

http://dx.doi.org/10.5772/51914

**1. Introduction** 

visual cortex in human.

Additional information is available at the end of the chapter

The search for organizing principles of visual processing in cortex has proven long and fruitful, demonstrating specific types of organization arising on multiple scales (e.g., magnocellular / parvo-cellular pathways [1] and ocular dominance columns [2]). One of the more important larger scale organizing principles of visual cortical organization is the visual field map (VFM): neurons whose visual receptive fields lie next to one another in visual space are located next to one another in cortex, forming one complete representation of contralateral visual space [3]. Each VFM subserves a specific computation or set of computations; locating these VFMs allows for the systematic exploration of these computations across visual cortex [4, 5]. It has been suggested that this retinotopic organization of VFMs allows for efficient connectivity between neurons that represent nearby locations in visual space, likely necessary for such processes as lateral inhibition and gain control [6-9]. This chapter will discuss the primary neuroimaging techniques used for measuring human VFMs, our current understanding of the organization of visuospatial representations across human visual cortex, the present state of our knowledge of white matter connectivity among these representations, and how these measurements inform us about the functional divisions of

**2. Neuroimaging methods for measuring human visual field maps** 

VFMs are routinely measured in the *in vivo* human brain using functional magnetic resonance imaging (fMRI) (e.g., [10-15]). The fMRI paradigms for these measurements take advantage of knowledge gleaned from electrophysiological measurements of visual cortex in animal models about the structure and stimulus preferences of VFMs. In monkey, as in human, visual information travels from the retina through the lateral geniculate nucleus (LGN) of the thalamus to primary visual cortex (area V1) in the posterior occipital lobe [16].

and reproduction in any medium, provided the original work is properly cited.


**Chapter 2** 

## **Visual Field Map Organization in Human Visual Cortex**

Alyssa A. Brewer and Brian Barton

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/51914

## **1. Introduction**

28 Visual Cortex – Current Status and Perspectives

10.1002/cne.902310307

and Histopathology 1987; 2(-) 147-151.

doi: 10.1146/annurev.neuro.26.041002.131123

[94] Sloper JJ, Powell TPS. A study of the axon initial segment and proximal axon of neurons in the primate motor cortex y somatic sensory cortices. Philosophical Transactions of

[96] Fairén A, Valverde F. A specialized type of neuron in the visual cortex of the cat: a Golgi and electron microscope study of chandelier cells. Journal of Comparative

[97] Matute C, Contamina P, Martinez-Millán L. The axon initial segment of corticocollicular neurons in the rabbit visual cortex: An electron-microscope study with HRP. Histology

[98] Fried SI, Lasker ACW, Desail NJ, Eddington DK, Rizzo 3rd JF. Axonal sodium-channel bands shape the response to electric stimulation in retinal ganglion cells. Journal of

[99] Konishi M. Coding of auditory space. Annual Review Neuroscience 2003; 26(-) 31–55,

[100] Kuba H, Ishii TM, Ohmori H. Axonal site of spike initiation enhances auditory coincidence detection. Nature 2006; 444(-) 1069-1072, doi:10.1038/nature05347 [101] Kuba H, Oichi Y, Ohmori H. Presynaptic activity regulates Na+ channel distribution at the axon initial segment. Nature 2010; 465(-) 1075-1078, doi:10.1038/nature09087

Neurology 1980; 194(4) 761-779, doi:10.1002/cne.901940405

Neurophysiology 2009; 10(4) 1972-1987, doi:10.1152/jn.91081.2008

the Royal Society of London 1979. Series B, Biological Sciences 285(1006) 173-197. [95] DeFelipe J, Hendry SCH, Jones EG, Schmechel D. Variability in the terminations of GABAergic chandelier cell axons on initial segments of pyramidal cell axons in the monkey sensory-motor cortex. Journal of Comparative Neurology 1985; 231(3) 364-384,

> The search for organizing principles of visual processing in cortex has proven long and fruitful, demonstrating specific types of organization arising on multiple scales (e.g., magnocellular / parvo-cellular pathways [1] and ocular dominance columns [2]). One of the more important larger scale organizing principles of visual cortical organization is the visual field map (VFM): neurons whose visual receptive fields lie next to one another in visual space are located next to one another in cortex, forming one complete representation of contralateral visual space [3]. Each VFM subserves a specific computation or set of computations; locating these VFMs allows for the systematic exploration of these computations across visual cortex [4, 5]. It has been suggested that this retinotopic organization of VFMs allows for efficient connectivity between neurons that represent nearby locations in visual space, likely necessary for such processes as lateral inhibition and gain control [6-9]. This chapter will discuss the primary neuroimaging techniques used for measuring human VFMs, our current understanding of the organization of visuospatial representations across human visual cortex, the present state of our knowledge of white matter connectivity among these representations, and how these measurements inform us about the functional divisions of visual cortex in human.

## **2. Neuroimaging methods for measuring human visual field maps**

VFMs are routinely measured in the *in vivo* human brain using functional magnetic resonance imaging (fMRI) (e.g., [10-15]). The fMRI paradigms for these measurements take advantage of knowledge gleaned from electrophysiological measurements of visual cortex in animal models about the structure and stimulus preferences of VFMs. In monkey, as in human, visual information travels from the retina through the lateral geniculate nucleus (LGN) of the thalamus to primary visual cortex (area V1) in the posterior occipital lobe [16].

© 2012 Brewer and Barton, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Primate visual cortex has been parcellated into multiple visual areas defined by their unique cytoarchitectonic structures, connectivity, functional processing, and visual field topography [14, 17-20]. In the more anterior areas of the occipital lobe and in the parietal and temporal lobes, the definitions and functions of many of these areas are still being investigated. These areas were first demonstrated in monkey, and it has been possible to identify several homologous visual areas in human posterior occipital cortex (e.g., [12, 21]).

Visual Field Map Organization in Human Visual Cortex 31

typically 15–20 standard deviations above the background noise. Stimuli comprised of other shapes (e.g., faces, objects) have also been used in studies interested in measuring the retinotopic organization of higher order visual cortex, but the high contrast checkerboard

**Figure 1. Visual Field Mapping Time Series Analysis.** Each row represents the activity and analysis of a time series of a single 6-cycle scan of one type of experimental stimuli (expanding rings or rotating wedges) for a single voxel. Black dots indicate simulated raw data points of % blood oxygen leveldependent (BOLD) modulation. The red lines indicate the peak activations per cycle for an imaginary set of voxels, which are the measurements used by the traveling wave retinotopy (TWR) analysis. The blue dotted line represents a sinusoidal fit of the simulated data points, which are the measurements used by population receptive field (pRF) modeling. Rows (A) and (B) represent time series of voxels with identical %BOLD modulation, but different peak responses, which indicate different stimulus selectivity (different 'phases' of response). For example, (A) might represent a voxel with a preferred eccentricity tuning of 5° eccentric to fixation, whereas (B) might have a preferred tuning of only 2° eccentric to fixation. Rows (A) and (C) represent time series of voxels with identical peak responses, indicating identical stimulus selectivity. However, (C) has much lower %BOLD modulation than (A), which may be due to two primary factors: differences in local vasculature or broader receptive field

In each scan, only one stimulus is presented, and all of visual space is cycled through several times with each stimulus (**Figure 1**). Typically, several scans are then averaged together for each stimulus type to increase the fMRI blood oxygen level-dependent (BOLD) signal to noise ratio. These stimuli create a travelling wave of cortical activity that travels from one end of the VFM to the other along iso-angle or iso–eccentricity lines, giving TWR its name. Thus the time, or phase, of the peak modulation varies smoothly across the cortical surface. This phase defines the most effective stimulus eccentricity (ring) and polar angle (wedge) to activate that region of cortex, giving TWR its description as 'phase-encoded retinotopy.' In

tuning for (C) than (A).

stimulus has proven to drive even these regions well in many studies[26, 27].

In human measurements, the most compelling evidence for visual areas is the VFMs, also commonly called retinotopic maps. In short, a VFM is a visual area with a complete representation of visual space, where neurons that represent adjacent locations on the retina (and visual space) are also adjacent in cortex [12]. Because many computations are required to create our visual experience, our brains have many specialized VFMs which perform one or more of those computations across the entire visual scene (e.g., motion perception happens throughout our visual field, not just in the upper left quadrant). By taking advantage of the knowledge of the retinotopic organization of visual input, multiple cortical VFMs can be measured using fMRI with respect to the two orthogonal dimensions needed to identify a unique location in visual space: eccentricity and polar angle. This chapter will review two of the most powerful fMRI techniques for very detailed measurements of VFM in individual subjects: travelling wave retinotopy (TWR) [10] and population receptive field (pRF) modeling [22].

## **2.1. The standard paradigm: Travelling wave retinotopy**

TWR has been the gold standard for visual field mapping since its development in the mid 1990's (**Figure 1**) [10, 23-26]. This technique uses two types of periodic stimuli that move smoothly across a contiguous region of visual space to measure the orthogonal dimensions of polar angle and eccentricity. One stimulus is designed to elicit each voxel's preferred polar angle by presenting a high-contrast, flickering checkerboard stimulus shaped like a wedge that spans the fovea to periphery along a small range of specific polar angles (**Figure 2A**). The wedge stimulus rotates either clockwise or counterclockwise in discrete even steps around the central fixation point to sequentially activate distinct polar angle representations of visual space. The second stimulus is designed to elicit each fMRI voxel's preferred eccentricity by presenting a stimulus shaped like a ring, which expands or contracts in discrete even steps between the central fovea and the periphery (**Figure 2B**). The measurement of these *two, orthogonal* dimensions is vital for the correct definition of VFMs, as these two measurements allow for the unique mapping of the responses of the neurons within a single voxel in cortex to a unique location in visual space. If only a single dimension is measured, the cortical response can only be localized to a broad swath of visual space, which does not allow for accurate delineation of VFM boundaries, as discussed further below.

These traveling-wave stimuli are typically comprised of a set of high contrast checkerboard patterns that are designed to maximally stimulate primary visual cortex and generally elicit an fMRI signal modulation on the order of 1%–3% (**Figures 1-2**). This modulation is typically 15–20 standard deviations above the background noise. Stimuli comprised of other shapes (e.g., faces, objects) have also been used in studies interested in measuring the retinotopic organization of higher order visual cortex, but the high contrast checkerboard stimulus has proven to drive even these regions well in many studies[26, 27].

30 Visual Cortex – Current Status and Perspectives

(pRF) modeling [22].

further below.

Primate visual cortex has been parcellated into multiple visual areas defined by their unique cytoarchitectonic structures, connectivity, functional processing, and visual field topography [14, 17-20]. In the more anterior areas of the occipital lobe and in the parietal and temporal lobes, the definitions and functions of many of these areas are still being investigated. These areas were first demonstrated in monkey, and it has been possible to identify several

In human measurements, the most compelling evidence for visual areas is the VFMs, also commonly called retinotopic maps. In short, a VFM is a visual area with a complete representation of visual space, where neurons that represent adjacent locations on the retina (and visual space) are also adjacent in cortex [12]. Because many computations are required to create our visual experience, our brains have many specialized VFMs which perform one or more of those computations across the entire visual scene (e.g., motion perception happens throughout our visual field, not just in the upper left quadrant). By taking advantage of the knowledge of the retinotopic organization of visual input, multiple cortical VFMs can be measured using fMRI with respect to the two orthogonal dimensions needed to identify a unique location in visual space: eccentricity and polar angle. This chapter will review two of the most powerful fMRI techniques for very detailed measurements of VFM in individual subjects: travelling wave retinotopy (TWR) [10] and population receptive field

TWR has been the gold standard for visual field mapping since its development in the mid 1990's (**Figure 1**) [10, 23-26]. This technique uses two types of periodic stimuli that move smoothly across a contiguous region of visual space to measure the orthogonal dimensions of polar angle and eccentricity. One stimulus is designed to elicit each voxel's preferred polar angle by presenting a high-contrast, flickering checkerboard stimulus shaped like a wedge that spans the fovea to periphery along a small range of specific polar angles (**Figure 2A**). The wedge stimulus rotates either clockwise or counterclockwise in discrete even steps around the central fixation point to sequentially activate distinct polar angle representations of visual space. The second stimulus is designed to elicit each fMRI voxel's preferred eccentricity by presenting a stimulus shaped like a ring, which expands or contracts in discrete even steps between the central fovea and the periphery (**Figure 2B**). The measurement of these *two, orthogonal* dimensions is vital for the correct definition of VFMs, as these two measurements allow for the unique mapping of the responses of the neurons within a single voxel in cortex to a unique location in visual space. If only a single dimension is measured, the cortical response can only be localized to a broad swath of visual space, which does not allow for accurate delineation of VFM boundaries, as discussed

These traveling-wave stimuli are typically comprised of a set of high contrast checkerboard patterns that are designed to maximally stimulate primary visual cortex and generally elicit an fMRI signal modulation on the order of 1%–3% (**Figures 1-2**). This modulation is

homologous visual areas in human posterior occipital cortex (e.g., [12, 21]).

**2.1. The standard paradigm: Travelling wave retinotopy** 

**Figure 1. Visual Field Mapping Time Series Analysis.** Each row represents the activity and analysis of a time series of a single 6-cycle scan of one type of experimental stimuli (expanding rings or rotating wedges) for a single voxel. Black dots indicate simulated raw data points of % blood oxygen leveldependent (BOLD) modulation. The red lines indicate the peak activations per cycle for an imaginary set of voxels, which are the measurements used by the traveling wave retinotopy (TWR) analysis. The blue dotted line represents a sinusoidal fit of the simulated data points, which are the measurements used by population receptive field (pRF) modeling. Rows (A) and (B) represent time series of voxels with identical %BOLD modulation, but different peak responses, which indicate different stimulus selectivity (different 'phases' of response). For example, (A) might represent a voxel with a preferred eccentricity tuning of 5° eccentric to fixation, whereas (B) might have a preferred tuning of only 2° eccentric to fixation. Rows (A) and (C) represent time series of voxels with identical peak responses, indicating identical stimulus selectivity. However, (C) has much lower %BOLD modulation than (A), which may be due to two primary factors: differences in local vasculature or broader receptive field tuning for (C) than (A).

In each scan, only one stimulus is presented, and all of visual space is cycled through several times with each stimulus (**Figure 1**). Typically, several scans are then averaged together for each stimulus type to increase the fMRI blood oxygen level-dependent (BOLD) signal to noise ratio. These stimuli create a travelling wave of cortical activity that travels from one end of the VFM to the other along iso-angle or iso–eccentricity lines, giving TWR its name. Thus the time, or phase, of the peak modulation varies smoothly across the cortical surface. This phase defines the most effective stimulus eccentricity (ring) and polar angle (wedge) to activate that region of cortex, giving TWR its description as 'phase-encoded retinotopy.' In

TWR data, the phase of the response is represented as a color-coded overlay on anatomical data (**Figure 2**). It is important to note that these types of TWR stimuli are not only excellent for measuring cortical VFMs, but that they only produce activity in regions that are retinotopically-organized.

Visual Field Map Organization in Human Visual Cortex 33

the Fourier analysis, incorrectly reducing overall coherence. For each stimulus condition (e.g., wedge or ring), each voxel is independently assigned a coherence value, thus measuring the strength of the response of that voxel to the stimulus. Only voxels with a coherence above a chosen threshold (typically 0.15 to 0.30 coherence) are further evaluated to determine the organization of cortical visuospatial representations into specific VFMs.

**2.2. An innovative approach to measure the organization of human visual cortex:** 

Once it became clear that there were limits to the ability of TWR to deal with VFMs with large RFs, researchers at Stanford University decided to improve VFM measurements by developing a new method that models the pRFs of each voxel within VFMs [22]. This model relies on the logic that, because VFMs are retinotopically organized, the population of RFs in each voxel of a VFM is expected to have similar preferred centers and sizes, allowing their combined pRF to be estimated as a single, two-dimensional Gaussian RF. Despite the fact that there is some variability in the neural RFs of each voxel in terms of their preferred centers and sizes, termed RF scatter, the pRF provides a good, if somewhat slightly larger, estimate of the individual neural RFs in the voxel. The advantages of the method are generally stated in comparison to TWR, as the field standard for measuring VFMs. The pRF method provides an accurate estimate of not only the preferred center for each voxel's pRF (as in TWR), but also its size (**Figures 3, 4**). In addition, the method does not require two distinct stimuli to measure orthogonal dimensions of visual space as in TWR, cutting down

**Figure 3. Measurements of an Individual Voxel.** (A) A typical voxel recorded from a popular 3 Tesla MRI scanner is on the order of 1 mm3, though often slightly larger (2-3 mm3). (B) Within each typical voxel, there are on the order of ~1 million neurons, depending on the size of the voxel. For voxels in retinotopic visual cortex, the neurons each have similarly located spatial receptive fields (black outlines) with preferred centers (black dots). (C) Traveling wave retinotopy (TWR) takes advantage of the fact that nearby neurons in retinotopic cortex have similar preferred centers in order to estimate a population preferred center for the population of neurons in a given voxel. (D) Population receptive field (pRF) modeling takes advantage of the fact that nearby neurons in retinotopic cortex have similar receptive fields in order to estimate not only a preferred center, but also a pRF for the population of

**Population receptive field modeling** 

on the total number of scans necessary per subject.

neurons in a given voxel.

**Figure 2. TWR Measurements.** Traveling wave stimuli typically consist of a set of high contrast checkerboard patterns that move smoothly and periodically through a range of eccentricities (ring) or polar angles (wedge). The inflated cortical surface (inset) is labeled as follows: CC, corpus callosum; POS, parietal-occipital sulcus; CaS, calcarine sulcus. An expanded view of this surface near calcarine sulcus is overlaid with a color map showing the response phase at each location for polar angle (A) and eccentricity experiments (B) (see the colored legend insets). The stimuli covered the central 16° radius of visual space. The solid white lines indicate the boundaries of visual area V1 in the calcarine sulcus. For clarity, the colored visual responses are only overlaid on locations near the calcarine sulcus, and only voxels with a powerful response at a coherence ≥ 0.25 are colored.

The design of TWR presents all eccentricities or polar angles at a given frequency per scan (typically 6-8 cycles per scan), which allows the use of a Fourier analysis. TWR only considers activity that is at this signal frequency, excluding low-frequency physiological noise, among other things. The statistical threshold for cortical activity arising from the TWR stimulus is commonly determined by *coherence*, which is equal to the amplitude of the BOLD signal modulation at the frequency of stimulus presentation (e.g., 6 stimulus cycles per scan), divided by the square root of the power over all other frequencies except the first and second harmonic (e.g., 12 and 18 cycles per scan). These harmonic frequencies commonly can be considered signal in such analyses, but we often take a conservative approach and simply exclude their values from the calculation of coherence. Including these frequencies as noise would lead to an artificially high average for the noise frequencies in the Fourier analysis, incorrectly reducing overall coherence. For each stimulus condition (e.g., wedge or ring), each voxel is independently assigned a coherence value, thus measuring the strength of the response of that voxel to the stimulus. Only voxels with a coherence above a chosen threshold (typically 0.15 to 0.30 coherence) are further evaluated to determine the organization of cortical visuospatial representations into specific VFMs.

32 Visual Cortex – Current Status and Perspectives

retinotopically-organized.

TWR data, the phase of the response is represented as a color-coded overlay on anatomical data (**Figure 2**). It is important to note that these types of TWR stimuli are not only excellent for measuring cortical VFMs, but that they only produce activity in regions that are

**Figure 2. TWR Measurements.** Traveling wave stimuli typically consist of a set of high contrast checkerboard patterns that move smoothly and periodically through a range of eccentricities (ring) or polar angles (wedge). The inflated cortical surface (inset) is labeled as follows: CC, corpus callosum; POS, parietal-occipital sulcus; CaS, calcarine sulcus. An expanded view of this surface near calcarine sulcus is overlaid with a color map showing the response phase at each location for polar angle (A) and eccentricity experiments (B) (see the colored legend insets). The stimuli covered the central 16° radius of visual space. The solid white lines indicate the boundaries of visual area V1 in the calcarine sulcus. For clarity, the colored visual responses are only overlaid on locations near the calcarine sulcus, and only

The design of TWR presents all eccentricities or polar angles at a given frequency per scan (typically 6-8 cycles per scan), which allows the use of a Fourier analysis. TWR only considers activity that is at this signal frequency, excluding low-frequency physiological noise, among other things. The statistical threshold for cortical activity arising from the TWR stimulus is commonly determined by *coherence*, which is equal to the amplitude of the BOLD signal modulation at the frequency of stimulus presentation (e.g., 6 stimulus cycles per scan), divided by the square root of the power over all other frequencies except the first and second harmonic (e.g., 12 and 18 cycles per scan). These harmonic frequencies commonly can be considered signal in such analyses, but we often take a conservative approach and simply exclude their values from the calculation of coherence. Including these frequencies as noise would lead to an artificially high average for the noise frequencies in

voxels with a powerful response at a coherence ≥ 0.25 are colored.

## **2.2. An innovative approach to measure the organization of human visual cortex: Population receptive field modeling**

Once it became clear that there were limits to the ability of TWR to deal with VFMs with large RFs, researchers at Stanford University decided to improve VFM measurements by developing a new method that models the pRFs of each voxel within VFMs [22]. This model relies on the logic that, because VFMs are retinotopically organized, the population of RFs in each voxel of a VFM is expected to have similar preferred centers and sizes, allowing their combined pRF to be estimated as a single, two-dimensional Gaussian RF. Despite the fact that there is some variability in the neural RFs of each voxel in terms of their preferred centers and sizes, termed RF scatter, the pRF provides a good, if somewhat slightly larger, estimate of the individual neural RFs in the voxel. The advantages of the method are generally stated in comparison to TWR, as the field standard for measuring VFMs. The pRF method provides an accurate estimate of not only the preferred center for each voxel's pRF (as in TWR), but also its size (**Figures 3, 4**). In addition, the method does not require two distinct stimuli to measure orthogonal dimensions of visual space as in TWR, cutting down on the total number of scans necessary per subject.

**Figure 3. Measurements of an Individual Voxel.** (A) A typical voxel recorded from a popular 3 Tesla MRI scanner is on the order of 1 mm3, though often slightly larger (2-3 mm3). (B) Within each typical voxel, there are on the order of ~1 million neurons, depending on the size of the voxel. For voxels in retinotopic visual cortex, the neurons each have similarly located spatial receptive fields (black outlines) with preferred centers (black dots). (C) Traveling wave retinotopy (TWR) takes advantage of the fact that nearby neurons in retinotopic cortex have similar preferred centers in order to estimate a population preferred center for the population of neurons in a given voxel. (D) Population receptive field (pRF) modeling takes advantage of the fact that nearby neurons in retinotopic cortex have similar receptive fields in order to estimate not only a preferred center, but also a pRF for the population of neurons in a given voxel.

To accomplish this, the pRF model first creates a very large database of possible pRF sizes and centers that cover the field of view of the stimulus (**Figure 4**). Then, the model convolves each of the pRF possibilities with a standard hemodynamic response function (HRF). Finally, the model uses a least-squares fitting method to iteratively test each of the pRF possibilities for each voxel independently against the actual data collected. Whichever pRF best fits the data is then assigned as the pRF for that voxel. Only voxels that contain activity above a chosen threshold of variance explained as determined by the model are included for further analysis.

Visual Field Map Organization in Human Visual Cortex 35

demonstrate differences in the internal receptive field structures between, for example, a hemifield map in primary visual cortex (V1) and a hemifield map in lateral cortex (e.g., LO-1; [28]). With the traveling wave method, both maps look very similar, as only the peak time series responses are measured. However, the underlying properties of the neuronal populations within these two maps are actually quite different, with finely tuned neurons in V1 and more broadly tuned neurons in lateral cortex. The models of the underlying neuronal properties from the pRF method can measure these receptive field differences, as well as the amount of input from ipsi- and contralateral visual fields (e.g., [29]). The human pRF size estimates for V1/V2/V3 reported by Dumoulin and Wandell [22] agree well with electrophysiological receptive field measurements at a range of eccentricities in

These pRF methods have successfully been used by a small group of labs to (1) investigate the normal organization of human visual cortex (e.g., [12, 29, 30]), (2) measure developmental plasticity in achiasmatic and sight recovery patients [31, 32], and (3) examine cortical reorganization in aged-related macular degeneration [33]. Because pRF Modeling has proven so successful, it is likely that it will eventually replace TWR as the standard method for measuring VFMs. Moreover, pRF modeling has an excellent future in the measurement of the details of pRFs, which is particularly important for the measurement of visual plasticity in humans. So far, the technique has primarily used a two-dimensional Gaussian profile for the pRF estimates, but researchers are working on the use of centersurround Gaussian pRFs, multiple location pRFs, and non-classical pRF shapes, which may allow for better pRF estimation as time continues. In the future, it is likely that pRF Modeling will be very successful when used in isolation, but also excellent to use in

**2.3. Functional MRI data acquisition and analysis for individual subjects** 

The size of each VFM across the cortical surface varies significantly across individuals [22]. In fact, the size of primary visual cortex, V1, can vary by at least a factor of 3 in size, independent of overall brain size. This means that the locations of each specific VFM are necessarily shifted across individuals with respect to the underlying structural anatomy. This shift appears to be increasingly variable as measurements move anterior from primary visual cortex into regions of visual cortex that subserve higher-order computations (e.g., object recognition), the very regions that are also the most difficult to measure with TWR due to the larger RFs of the neurons here. Thus, averaging fMRI VFM data across subjects problematically blurs VFM data to a degree that should be unusable and may even obliterate VFM organization all together. Similarly, simply using coordinates from a standardized template (e.g., Talairach or MNI coordinates) to accurately estimate the location of any VFMs beyond area V1 in individual or group averaged data is not possible. The only accurate approach is to measure VFM in individual subjects. We will review here an example of one of several straightforward approaches for individual subject VFM data

corresponding locations within primate VFMs.

conjunction with other techniques.

collection and analysis.

**Figure 4. Population Receptive Field Modeling.** The parameter estimation procedure for the population receptive field (pRF) model is shown as a flow chart. The example stimulus aperture is a moving bar stimulus. Adapted from Figure 2 in [22].

Although it is technically possible to use any stimulus that systematically traverses the entire field of view, typically the stimulus takes one of two forms. First is a slightly modified version of the TWR stimuli, in which neutral gray blank periods are inserted at an offfrequency from the stimulus frequency (i.e., 4 instead of 6-8 cycles/scan, so they are separable in the Fourier analysis). The second and increasingly common stimulus is a highcontrast flickering checkerboard bar stimulus that steps across the field of view in the 8 cardinal directions, again with several interspersed neutral gray blank periods. The neutral gray blank periods allow for an estimation of a voxel's response to any visual stimulus versus just the preferred visual stimulus, which is crucial for the accurate measurement of pRF sizes. In theory, one could also tile visual space using any stimulus of interest, if the aforementioned stimuli do not drive the area well. Since the checkerboard stimuli were designed to drive activity in early visual cortex, it is possible other stimuli containing more complex may perform better in higher-order VFMs.

The pRF method has the additional benefit of measuring other neuronal population properties, such as receptive field size and laterality. These pRF measurements can demonstrate differences in the internal receptive field structures between, for example, a hemifield map in primary visual cortex (V1) and a hemifield map in lateral cortex (e.g., LO-1; [28]). With the traveling wave method, both maps look very similar, as only the peak time series responses are measured. However, the underlying properties of the neuronal populations within these two maps are actually quite different, with finely tuned neurons in V1 and more broadly tuned neurons in lateral cortex. The models of the underlying neuronal properties from the pRF method can measure these receptive field differences, as well as the amount of input from ipsi- and contralateral visual fields (e.g., [29]). The human pRF size estimates for V1/V2/V3 reported by Dumoulin and Wandell [22] agree well with electrophysiological receptive field measurements at a range of eccentricities in corresponding locations within primate VFMs.

34 Visual Cortex – Current Status and Perspectives

To accomplish this, the pRF model first creates a very large database of possible pRF sizes and centers that cover the field of view of the stimulus (**Figure 4**). Then, the model convolves each of the pRF possibilities with a standard hemodynamic response function (HRF). Finally, the model uses a least-squares fitting method to iteratively test each of the pRF possibilities for each voxel independently against the actual data collected. Whichever pRF best fits the data is then assigned as the pRF for that voxel. Only voxels that contain activity above a chosen threshold of variance explained as determined by the model are included for further analysis.

**Figure 4. Population Receptive Field Modeling.** The parameter estimation procedure for the population receptive field (pRF) model is shown as a flow chart. The example stimulus aperture is a

Although it is technically possible to use any stimulus that systematically traverses the entire field of view, typically the stimulus takes one of two forms. First is a slightly modified version of the TWR stimuli, in which neutral gray blank periods are inserted at an offfrequency from the stimulus frequency (i.e., 4 instead of 6-8 cycles/scan, so they are separable in the Fourier analysis). The second and increasingly common stimulus is a highcontrast flickering checkerboard bar stimulus that steps across the field of view in the 8 cardinal directions, again with several interspersed neutral gray blank periods. The neutral gray blank periods allow for an estimation of a voxel's response to any visual stimulus versus just the preferred visual stimulus, which is crucial for the accurate measurement of pRF sizes. In theory, one could also tile visual space using any stimulus of interest, if the aforementioned stimuli do not drive the area well. Since the checkerboard stimuli were designed to drive activity in early visual cortex, it is possible other stimuli containing more

The pRF method has the additional benefit of measuring other neuronal population properties, such as receptive field size and laterality. These pRF measurements can

moving bar stimulus. Adapted from Figure 2 in [22].

complex may perform better in higher-order VFMs.

These pRF methods have successfully been used by a small group of labs to (1) investigate the normal organization of human visual cortex (e.g., [12, 29, 30]), (2) measure developmental plasticity in achiasmatic and sight recovery patients [31, 32], and (3) examine cortical reorganization in aged-related macular degeneration [33]. Because pRF Modeling has proven so successful, it is likely that it will eventually replace TWR as the standard method for measuring VFMs. Moreover, pRF modeling has an excellent future in the measurement of the details of pRFs, which is particularly important for the measurement of visual plasticity in humans. So far, the technique has primarily used a two-dimensional Gaussian profile for the pRF estimates, but researchers are working on the use of centersurround Gaussian pRFs, multiple location pRFs, and non-classical pRF shapes, which may allow for better pRF estimation as time continues. In the future, it is likely that pRF Modeling will be very successful when used in isolation, but also excellent to use in conjunction with other techniques.

#### **2.3. Functional MRI data acquisition and analysis for individual subjects**

The size of each VFM across the cortical surface varies significantly across individuals [22]. In fact, the size of primary visual cortex, V1, can vary by at least a factor of 3 in size, independent of overall brain size. This means that the locations of each specific VFM are necessarily shifted across individuals with respect to the underlying structural anatomy. This shift appears to be increasingly variable as measurements move anterior from primary visual cortex into regions of visual cortex that subserve higher-order computations (e.g., object recognition), the very regions that are also the most difficult to measure with TWR due to the larger RFs of the neurons here. Thus, averaging fMRI VFM data across subjects problematically blurs VFM data to a degree that should be unusable and may even obliterate VFM organization all together. Similarly, simply using coordinates from a standardized template (e.g., Talairach or MNI coordinates) to accurately estimate the location of any VFMs beyond area V1 in individual or group averaged data is not possible. The only accurate approach is to measure VFM in individual subjects. We will review here an example of one of several straightforward approaches for individual subject VFM data collection and analysis.

To optimize these VFM experiments, several types of fMRI scans are obtained for each subject. First, one acquires a high-resolution structural anatomy of the whole brain (e.g., 1 mm3 resolution). Several types of pulse sequences are available, such as MPRAGE, a fast gradient echo T1- weighted inversion pulse sequence. The goal in this scan is to maximize the image contrast between white and gray cortical matter, important for the subsequent analysis. These anatomical data provide a basic coordinate frame for representing the fMRI data for each subject. Second, functional T2\*-weighted BOLD contrast images are acquired for the VFM measurements. We commonly use a gradient echo pulse sequence with a SENSE factor of 1.5 that provides whole brain coverage with slices approximately parallel to the calcarine sulcus (home of V1) and a 1.8 x 1.8 x 3 mm slice resolution (no gap). Each functional scan typically lasts will approximately 3-4 minutes, and we acquire 4-8 scans per stimulus type (e.g., wedge, ring, bar) to average together. In addition, one lower resolution anatomical inplane image is acquired before each set of functional scans, with the same slice prescription as the functional scans but with a higher spatial resolution (e.g., 1 mm x 1 mm x 3 mm voxels). These T1-weighted slices are physically in register with the functional slices and can then be used to align the functional data with the high-resolution anatomy data [34].

Visual Field Map Organization in Human Visual Cortex 37

11]). The organization of retinotopic VFMs is typically determined by manually tracing the boundaries of quarter-field or hemifield representations (**Figure 2**). These boundaries are located at the position where the measurements of visual field angle reverse direction or, for regions on the end of visually responsive cortex, end at an angular meridian or at the periphery of a VFM [5, 11]. For boundaries in a reversal, the boundary is drawn to split the reversal evenly between the two maps, unless additional functional data (e.g., motion

**Figure 5. Orthogonal Dimensions of Visual Field Maps.** Top Left: Eccentricity visual space legend. Each color represents an iso-eccentricity line in the left visual hemifield. Top Right: Polar angle visual space legend. Each color represents an iso-polar angle line in the left visual hemifield. (A) Eccentricity gradient for a visual field map (VFM). Note the gradient running from the center to more peripheral eccentricities runs from right to left. This gradient would be orthogonal to the polar angle gradient in (C), such that each isoeccentricity line has a representation of the full range of polar angles. (B) A VFM. The combination of the orthogonal gradients in (A) and (C) form one complete representation of a hemifield of visual space. This forms one half of a complete VFM, the corresponding half being located in the opposite hemisphere of the brain. Because the hemifield represented is the left, this map would be located in the right hemisphere. The black outer border indicates that each of the two gradients is located in the same portion of cortex. (C) Polar angle gradient for a VFM. Note that the colors of the cartoon in (C) are inverted with respect to the polar angle visual space legend at top. The inverted cartoon is meant to more accurately represent the inverted representation of visual space in early visual cortex. For example, in primary visual cortex (V1), the lower quarterfield of visual space is represented on the dorsal (upper) surface of the occipital lobe, and vice versa. (D) Two adjacent eccentricity gradients running in opposite directions, with adjacent representations of the central visual hemifield. (E) When the gradients in (D) are combined with adjacent polar angle gradients such as that in (F), two complete representations of the hemifield of visual space are formed. (F) Two adjacent polar angle gradients running in the same direction, with each iso-polar angle line for each gradient lying adjacent to one another. Note that if one only measured polar angle information, one would

not have the corresponding eccentricity information to know whether that portion of cortex truly

contained one (as in (B)), two (as in (E)), or more complete representations of that hemifield of visual space.

localizer) is present to suggest otherwise.

For analysis of such functional imaging data for individual subjects, several neuroimaging software packages are available that can be used. We use a Matlab-based signal processing software package called *mrVista*, which was developed by the Wandell lab at Stanford University and is now widely used for such neuroimaging analysis [35, 36]; *mrVista* is opensource software and is publicly available online at http://white.stanford.edu/software/. With this software, the location of the cortical gray matter for each subject is identified ('segmented') in the high-resolution anatomical scan using the *mrVista* automated algorithm followed by handediting to minimize errors for individual subject analyses [36]. Gray matter is then grown from the segmented white matter to form a 3-4 mm layer covering the white matter surface. To improve sensitivity, only data from this identified gray matter are analyzed. The gray matter is then rendered in 3D close to the white matter boundary or unfolded into a continuous, flat sheet to allow visualization of functional activity within the sulci. During preprocessing of the functional data, linear trends are removed from the fMRI time series, but no spatial smoothing is applied to the data to better preserve the details of the VFM organization. Motion correction algorithms can then be applied between scans in each session as well as within individual scans (*mrVista* uses a mutual information motion correction algorithm [37]); however, motion correction algorithms may themselves create artifacts, so should not be routinely applied if not needed. If motion correction fails, then scans with motion artifact greater than one voxel can be discarded. After registration to the high-resolution anatomy, the functional activity can be visualized either in its original coordinate frame (inplanes), on the segmented gray matter in anatomical volume slices, or on inflated or flattened representations of the cortical surface to allow for optimal definition of VFM boundaries.

#### **2.4. Defining visual field map boundaries**

VFMs are defined by the following criteria: 1) both a polar angle and an eccentricity gradient must be present, 2) the polar angle and eccentricity gradients are orthogonal to one another, and 3) a VFM represents a complete contralateral hemifield of visual space (**Figure 5**; e.g., [5, 11]). The organization of retinotopic VFMs is typically determined by manually tracing the boundaries of quarter-field or hemifield representations (**Figure 2**). These boundaries are located at the position where the measurements of visual field angle reverse direction or, for regions on the end of visually responsive cortex, end at an angular meridian or at the periphery of a VFM [5, 11]. For boundaries in a reversal, the boundary is drawn to split the reversal evenly between the two maps, unless additional functional data (e.g., motion localizer) is present to suggest otherwise.

36 Visual Cortex – Current Status and Perspectives

allow for optimal definition of VFM boundaries.

**2.4. Defining visual field map boundaries** 

To optimize these VFM experiments, several types of fMRI scans are obtained for each subject. First, one acquires a high-resolution structural anatomy of the whole brain (e.g., 1 mm3 resolution). Several types of pulse sequences are available, such as MPRAGE, a fast gradient echo T1- weighted inversion pulse sequence. The goal in this scan is to maximize the image contrast between white and gray cortical matter, important for the subsequent analysis. These anatomical data provide a basic coordinate frame for representing the fMRI data for each subject. Second, functional T2\*-weighted BOLD contrast images are acquired for the VFM measurements. We commonly use a gradient echo pulse sequence with a SENSE factor of 1.5 that provides whole brain coverage with slices approximately parallel to the calcarine sulcus (home of V1) and a 1.8 x 1.8 x 3 mm slice resolution (no gap). Each functional scan typically lasts will approximately 3-4 minutes, and we acquire 4-8 scans per stimulus type (e.g., wedge, ring, bar) to average together. In addition, one lower resolution anatomical inplane image is acquired before each set of functional scans, with the same slice prescription as the functional scans but with a higher spatial resolution (e.g., 1 mm x 1 mm x 3 mm voxels). These T1-weighted slices are physically in register with the functional slices and can

then be used to align the functional data with the high-resolution anatomy data [34].

For analysis of such functional imaging data for individual subjects, several neuroimaging software packages are available that can be used. We use a Matlab-based signal processing software package called *mrVista*, which was developed by the Wandell lab at Stanford University and is now widely used for such neuroimaging analysis [35, 36]; *mrVista* is opensource software and is publicly available online at http://white.stanford.edu/software/. With this software, the location of the cortical gray matter for each subject is identified ('segmented') in the high-resolution anatomical scan using the *mrVista* automated algorithm followed by handediting to minimize errors for individual subject analyses [36]. Gray matter is then grown from the segmented white matter to form a 3-4 mm layer covering the white matter surface. To improve sensitivity, only data from this identified gray matter are analyzed. The gray matter is then rendered in 3D close to the white matter boundary or unfolded into a continuous, flat sheet to allow visualization of functional activity within the sulci. During preprocessing of the functional data, linear trends are removed from the fMRI time series, but no spatial smoothing is applied to the data to better preserve the details of the VFM organization. Motion correction algorithms can then be applied between scans in each session as well as within individual scans (*mrVista* uses a mutual information motion correction algorithm [37]); however, motion correction algorithms may themselves create artifacts, so should not be routinely applied if not needed. If motion correction fails, then scans with motion artifact greater than one voxel can be discarded. After registration to the high-resolution anatomy, the functional activity can be visualized either in its original coordinate frame (inplanes), on the segmented gray matter in anatomical volume slices, or on inflated or flattened representations of the cortical surface to

VFMs are defined by the following criteria: 1) both a polar angle and an eccentricity gradient must be present, 2) the polar angle and eccentricity gradients are orthogonal to one another, and 3) a VFM represents a complete contralateral hemifield of visual space (**Figure 5**; e.g., [5,

**Figure 5. Orthogonal Dimensions of Visual Field Maps.** Top Left: Eccentricity visual space legend. Each color represents an iso-eccentricity line in the left visual hemifield. Top Right: Polar angle visual space legend. Each color represents an iso-polar angle line in the left visual hemifield. (A) Eccentricity gradient for a visual field map (VFM). Note the gradient running from the center to more peripheral eccentricities runs from right to left. This gradient would be orthogonal to the polar angle gradient in (C), such that each isoeccentricity line has a representation of the full range of polar angles. (B) A VFM. The combination of the orthogonal gradients in (A) and (C) form one complete representation of a hemifield of visual space. This forms one half of a complete VFM, the corresponding half being located in the opposite hemisphere of the brain. Because the hemifield represented is the left, this map would be located in the right hemisphere. The black outer border indicates that each of the two gradients is located in the same portion of cortex. (C) Polar angle gradient for a VFM. Note that the colors of the cartoon in (C) are inverted with respect to the polar angle visual space legend at top. The inverted cartoon is meant to more accurately represent the inverted representation of visual space in early visual cortex. For example, in primary visual cortex (V1), the lower quarterfield of visual space is represented on the dorsal (upper) surface of the occipital lobe, and vice versa. (D) Two adjacent eccentricity gradients running in opposite directions, with adjacent representations of the central visual hemifield. (E) When the gradients in (D) are combined with adjacent polar angle gradients such as that in (F), two complete representations of the hemifield of visual space are formed. (F) Two adjacent polar angle gradients running in the same direction, with each iso-polar angle line for each gradient lying adjacent to one another. Note that if one only measured polar angle information, one would not have the corresponding eccentricity information to know whether that portion of cortex truly contained one (as in (B)), two (as in (E)), or more complete representations of that hemifield of visual space.

In addition to expert manual definition of VFMs, one can also use an automated tool to help give an objective definition of the boundary reversals between VFMs. Only a few of these tools are currently available, however. One such algorithm identifies VFMs by minimizing the error between an expected visual map (atlas) and the observed data [38]. In this tool, the atlas is coarsely aligned with the data and then elastically deformed. The search algorithm minimizes the weighted sum of deviations between the predicted and measured maps and the force of the elastic deformation. This algorithm is applied to both angle and eccentricity maps simultaneously to obtain a fit between these retinotopic measurements and templates of the two expected VFMs. This automated approach thus give more objective determinations of the boundaries of hemifield and quarter field visual angle representations or of the periphery edge of eccentricity representations, which can then be used to define specific VMFs.

Visual Field Map Organization in Human Visual Cortex 39

including ocular dominance columns, pinwheel orientation columns, and blobs/interblobs have been the subject of much study and argument (e.g., [2, 40, 41]). It remains to be seen

V1, V2, and V3 each contain a foveal representation positioned at the occipital pole, with progressively more peripheral representations extending into more anteromedial cortex, forming complete eccentricity gradients (**Figure 2**; e.g., [12, 13, 15, 23]). The region where the individual foveal representations meet at the occipital pole is commonly referred to the as the *foveal confluence* [42]. Despite the apparent merging of these foveal representations into one confluent fovea in fMRI measurements of eccentricity gradients, distinct boundaries between V1, V2, and V3 have been shown to be present even within this most central foveal

**Figure 6. Medial Occipital Cortex.** The anatomical region containing early visual areas V1, V2, and V3 is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. Gray represents sulci, and white represents gyri. Cu, cuneus; CaS, calcarine sulcus; LiG, lingual gyrus; POS, parieto-occipital sulcus; ColS, collateral sulcus.

The boundaries between each map are delineated by reversals in polar angle gradients (**Figure 2, 5**; e.g., [12, 13, 15, 23]). V1 has a contiguous polar angle gradient. In contrast, V2 and V3 have split-hemifield representations (quarterfields), which are denoted by their locations dorsal or ventral to V1 (V2d, V2v, V3d, V3v). For each map, the lower visual quarterfield is represented on the dorsal surface, and the upper visual quarterfield is represented on the ventral surface. The quarterfields of V2 and V3 are connected at the fovea for each map, but are otherwise distinct. Although some details differ between the macaque and human V1, V2, and V3 maps (for example, the surface area of macaque V1 is roughly half that of human V1), they are arguably the most similar between the species in terms of structure and function [14, 17, 19, 20, 44, 45]. Beyond these three maps, even as early as hV4, the anatomical and topographical details of the maps diverge [11, 46]. As a practical matter, due to their relatively consistent anatomical locations and unique concentric organization, these three maps form the first landmarks identified in visual field mapping analyses [10, 13]. However, as noted above, these three maps can differ

how many maps throughout the visual hierarchy have similarly complex mosaics.

representation [42, 43].

In cortical regions that have undetermined or ambiguous maps, this algorithm can be used to try a variety of possible templates of map organization (i.e., quarterfield map vs. hemifield map). Further, larger scale patterns of the organization of VFM across regions of cortex can be tested. By determining the error between the atlas template prediction and the actual angle and eccentricity measurements, the best fit template of VFM organization for a particular region can be estimated [11]. These atlas estimates can also be used to average map data across our subjects within a particular region of visual cortex [28]. The fitted atlas template additionally provides definitions of iso-angle and iso-eccentricity lines within each map, which can further be examined to compare patterns of VFM organization across the subject population [5, 11, 39].

## **3. Multiple visual field maps span human visual cortex**

This section will review current human VFM organization and some of the controversies surrounding these measurements.

## **3.1. Visual field maps in medial occipital cortex**

Three hemifield representations of visual space known as V1, V2, and V3 occupy the medial wall of occipital cortex in humans (**Figures 2, 5, 6**; for a review, see [12]). V1 is very reliably located in the calcarine sulcus, bounded on either side by the unique split-hemifield representations of V2 and V3 on the cuneus and lingual gyrus. V1 is known as "primary visual cortex," because it receives direct input from the retino-geniculate pathway and is the first place in the retino-geniculo-cortical pathway where information from the two eyes is combined. Not only that, but V1 is an important site of basic calculations of orientation, color, and motion. Each computation is performed across the entire visual field, yet V1 appears at the level of fMRI measurements to be a single, smooth representation of visual space. One can think of V1 as several VFMs laid on top of one another, each of which performs a single computation (one overlapping map each for color, orientation, and motion). To accomplish this organization, a very intricate mosaic of neurons subserving these computations allows for each computation to be performed over each portion of visual space. These mosaics, including ocular dominance columns, pinwheel orientation columns, and blobs/interblobs have been the subject of much study and argument (e.g., [2, 40, 41]). It remains to be seen how many maps throughout the visual hierarchy have similarly complex mosaics.

38 Visual Cortex – Current Status and Perspectives

specific VMFs.

subject population [5, 11, 39].

surrounding these measurements.

In addition to expert manual definition of VFMs, one can also use an automated tool to help give an objective definition of the boundary reversals between VFMs. Only a few of these tools are currently available, however. One such algorithm identifies VFMs by minimizing the error between an expected visual map (atlas) and the observed data [38]. In this tool, the atlas is coarsely aligned with the data and then elastically deformed. The search algorithm minimizes the weighted sum of deviations between the predicted and measured maps and the force of the elastic deformation. This algorithm is applied to both angle and eccentricity maps simultaneously to obtain a fit between these retinotopic measurements and templates of the two expected VFMs. This automated approach thus give more objective determinations of the boundaries of hemifield and quarter field visual angle representations or of the periphery edge of eccentricity representations, which can then be used to define

In cortical regions that have undetermined or ambiguous maps, this algorithm can be used to try a variety of possible templates of map organization (i.e., quarterfield map vs. hemifield map). Further, larger scale patterns of the organization of VFM across regions of cortex can be tested. By determining the error between the atlas template prediction and the actual angle and eccentricity measurements, the best fit template of VFM organization for a particular region can be estimated [11]. These atlas estimates can also be used to average map data across our subjects within a particular region of visual cortex [28]. The fitted atlas template additionally provides definitions of iso-angle and iso-eccentricity lines within each map, which can further be examined to compare patterns of VFM organization across the

This section will review current human VFM organization and some of the controversies

Three hemifield representations of visual space known as V1, V2, and V3 occupy the medial wall of occipital cortex in humans (**Figures 2, 5, 6**; for a review, see [12]). V1 is very reliably located in the calcarine sulcus, bounded on either side by the unique split-hemifield representations of V2 and V3 on the cuneus and lingual gyrus. V1 is known as "primary visual cortex," because it receives direct input from the retino-geniculate pathway and is the first place in the retino-geniculo-cortical pathway where information from the two eyes is combined. Not only that, but V1 is an important site of basic calculations of orientation, color, and motion. Each computation is performed across the entire visual field, yet V1 appears at the level of fMRI measurements to be a single, smooth representation of visual space. One can think of V1 as several VFMs laid on top of one another, each of which performs a single computation (one overlapping map each for color, orientation, and motion). To accomplish this organization, a very intricate mosaic of neurons subserving these computations allows for each computation to be performed over each portion of visual space. These mosaics,

**3. Multiple visual field maps span human visual cortex** 

**3.1. Visual field maps in medial occipital cortex** 

V1, V2, and V3 each contain a foveal representation positioned at the occipital pole, with progressively more peripheral representations extending into more anteromedial cortex, forming complete eccentricity gradients (**Figure 2**; e.g., [12, 13, 15, 23]). The region where the individual foveal representations meet at the occipital pole is commonly referred to the as the *foveal confluence* [42]. Despite the apparent merging of these foveal representations into one confluent fovea in fMRI measurements of eccentricity gradients, distinct boundaries between V1, V2, and V3 have been shown to be present even within this most central foveal representation [42, 43].

**Figure 6. Medial Occipital Cortex.** The anatomical region containing early visual areas V1, V2, and V3 is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. Gray represents sulci, and white represents gyri. Cu, cuneus; CaS, calcarine sulcus; LiG, lingual gyrus; POS, parieto-occipital sulcus; ColS, collateral sulcus.

The boundaries between each map are delineated by reversals in polar angle gradients (**Figure 2, 5**; e.g., [12, 13, 15, 23]). V1 has a contiguous polar angle gradient. In contrast, V2 and V3 have split-hemifield representations (quarterfields), which are denoted by their locations dorsal or ventral to V1 (V2d, V2v, V3d, V3v). For each map, the lower visual quarterfield is represented on the dorsal surface, and the upper visual quarterfield is represented on the ventral surface. The quarterfields of V2 and V3 are connected at the fovea for each map, but are otherwise distinct. Although some details differ between the macaque and human V1, V2, and V3 maps (for example, the surface area of macaque V1 is roughly half that of human V1), they are arguably the most similar between the species in terms of structure and function [14, 17, 19, 20, 44, 45]. Beyond these three maps, even as early as hV4, the anatomical and topographical details of the maps diverge [11, 46]. As a practical matter, due to their relatively consistent anatomical locations and unique concentric organization, these three maps form the first landmarks identified in visual field mapping analyses [10, 13]. However, as noted above, these three maps can differ significantly in size across individuals. While V1 is always positioned along the calcarine sulcus in normal individuals, an increase in V1 size will necessarily shift the locations of V2 and V3 with respect to the specific underlying anatomy.

Visual Field Map Organization in Human Visual Cortex 41

macaque remains a split-hemifield adjacent to area V3, human V4 (designated hV4 because of the unclear homology to macaque V4) is positioned as a complete hemifield on the ventral occipital surface adjacent to V3v (**Figure 7**). Several additional VFMs containing representations of complete, contiguous hemifields lie anterior to hV4 roughly along the fusiform gyrus (**Figure 11**). These maps are named and numbered for their anatomical locations: VO-1 and VO-2, for ventral-occipital, and PHC-1 and PHC-2, for

The differences between human and macaque organization at the fourth visual area initially led to much controversy in the field regarding the organization of V4 in human, as some researchers sought a similar pattern of organization for the fourth visual area between human and macaque. To understand this controversy, it is important to review some of the

One of the early lines of investigation into the ventral surface focused on measurements of both color and retinotopic organization. Zeki and colleagues measured responses to an isoluminant pattern modulated in chromatic contrast in two regions of ventral occipitotemporal cortex: V4 and V4 alpha [48, 49]. McKeefry and Zeki [50] then demonstrated that the posterior color-responsive region of V4 was at least coarsely retinotopically organized and represented the entire contralateral hemifield. However, they

**Figure 8. Ventral Occipitotemporal Cortex***.* The anatomical region containing ventral visual areas V2v, V3v, hV4, VO-1, VO-2, PHC-1, and PHC-2 is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. CaS, calcarine sulcus; LiG, lingual gyrus; ColS, collateral sulcus; FuG, fusiform gyrus; PHG, parahippocampal gyrus. Other details

Hadjikhani et al. [51] also measured retinotopic and color organization along this region, describing two ventral retinotopic regions. The first was an upper quarterfield map, which they referred to as V4v. This putative V4v abutted the central visual field representation of V3v with an eccentricity map parallelingV1/V2/V3. Unlike the measurements of McKeefry and Zeki [50], they saw no adjacent lower quarterfield map that would form a complete contralateral hemifield. Instead, they described a hemifield map with an eccentricity representation that ran perpendicular to the putative V4v quarter field and called this VFM

parahippocampal cortex (**Figure 8**).

history of measurements in this region.

as in **Figure 6**.

did not locate this map with respect to other neighboring VFMs.

#### **3.2. Visual field maps in ventral occipitotemporal cortex**

Beyond V3v, the organization of VFMs in human cortex no longer follows that of macaque. This divergence should not be surprising given that the two species diverged from a common ancestor approximately 25 million years ago [47]. While the fourth visual area of

**Figure 7. Comparison of Human and Macaque Monkey Occipital Cortex.** (A) 3D renderings of human (top) and macaque monkey (bottom) cortex are shown for a single right hemisphere. Cortical sheet is rendered at the white-gray boundary to allow visualization into the sulci. Hemispheres are scaled to relatively match in size. Scale bar is 1 cm. (B) Cartoon representations of flattened sections of cortex are centered on the occipital pole and show eccentricity gradients of human (top) and macaque (bottom) for visual field maps (VFMs) in posterior occipital cortex. Black lines denote boundaries between VFMs. Each color represents the location in visual space that best drives this region of cortex (see color legend inset for left visual field eccentricity). (C) Cartoon representations now show polar angle gradients of human (top) and macaque (bottom) for VFMs in posterior occipital cortex. Each color represents the location in visual space that best drives this region of cortex (see color legend inset for left visual field polar angle). Arrows (center) depict the approximate anatomical orientation for the cartoon representations in (B) and (C).

macaque remains a split-hemifield adjacent to area V3, human V4 (designated hV4 because of the unclear homology to macaque V4) is positioned as a complete hemifield on the ventral occipital surface adjacent to V3v (**Figure 7**). Several additional VFMs containing representations of complete, contiguous hemifields lie anterior to hV4 roughly along the fusiform gyrus (**Figure 11**). These maps are named and numbered for their anatomical locations: VO-1 and VO-2, for ventral-occipital, and PHC-1 and PHC-2, for parahippocampal cortex (**Figure 8**).

40 Visual Cortex – Current Status and Perspectives

and V3 with respect to the specific underlying anatomy.

**3.2. Visual field maps in ventral occipitotemporal cortex** 

significantly in size across individuals. While V1 is always positioned along the calcarine sulcus in normal individuals, an increase in V1 size will necessarily shift the locations of V2

Beyond V3v, the organization of VFMs in human cortex no longer follows that of macaque. This divergence should not be surprising given that the two species diverged from a common ancestor approximately 25 million years ago [47]. While the fourth visual area of

**Figure 7. Comparison of Human and Macaque Monkey Occipital Cortex.** (A) 3D renderings of human (top) and macaque monkey (bottom) cortex are shown for a single right hemisphere. Cortical sheet is rendered at the white-gray boundary to allow visualization into the sulci. Hemispheres are scaled to relatively match in size. Scale bar is 1 cm. (B) Cartoon representations of flattened sections of cortex are centered on the occipital pole and show eccentricity gradients of human (top) and macaque (bottom) for visual field maps (VFMs) in posterior occipital cortex. Black lines denote boundaries between VFMs. Each color represents the location in visual space that best drives this region of cortex (see color legend inset for left visual field eccentricity). (C) Cartoon representations now show polar angle gradients of human (top) and macaque (bottom) for VFMs in posterior occipital cortex. Each color represents the location in visual space that best drives this region of cortex (see color legend inset for left visual field polar angle). Arrows (center) depict the approximate anatomical orientation for the cartoon representations in (B) and (C).

The differences between human and macaque organization at the fourth visual area initially led to much controversy in the field regarding the organization of V4 in human, as some researchers sought a similar pattern of organization for the fourth visual area between human and macaque. To understand this controversy, it is important to review some of the history of measurements in this region.

One of the early lines of investigation into the ventral surface focused on measurements of both color and retinotopic organization. Zeki and colleagues measured responses to an isoluminant pattern modulated in chromatic contrast in two regions of ventral occipitotemporal cortex: V4 and V4 alpha [48, 49]. McKeefry and Zeki [50] then demonstrated that the posterior color-responsive region of V4 was at least coarsely retinotopically organized and represented the entire contralateral hemifield. However, they did not locate this map with respect to other neighboring VFMs.

**Figure 8. Ventral Occipitotemporal Cortex***.* The anatomical region containing ventral visual areas V2v, V3v, hV4, VO-1, VO-2, PHC-1, and PHC-2 is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. CaS, calcarine sulcus; LiG, lingual gyrus; ColS, collateral sulcus; FuG, fusiform gyrus; PHG, parahippocampal gyrus. Other details as in **Figure 6**.

Hadjikhani et al. [51] also measured retinotopic and color organization along this region, describing two ventral retinotopic regions. The first was an upper quarterfield map, which they referred to as V4v. This putative V4v abutted the central visual field representation of V3v with an eccentricity map parallelingV1/V2/V3. Unlike the measurements of McKeefry and Zeki [50], they saw no adjacent lower quarterfield map that would form a complete contralateral hemifield. Instead, they described a hemifield map with an eccentricity representation that ran perpendicular to the putative V4v quarter field and called this VFM V8. Using harmonic stimuli that alternated in luminance and chrominance, they also showed color responsivity within V8, although this stimulus type also would stimulate regions responsive to variations in luminance. Following the model of macaque cortex, Tootell and Hadjikhani [21] searched for a quarterfield map in dorsal occipital cortex to pair with their putative V4v. They failed to find the map and concluded that it did not exist. They did not resolve why an isolated quarterfield map would exist, a strange organization which would necessitate that whatever computation was subserved by putative V4v was only performed on one quarterfield of visual space.

Visual Field Map Organization in Human Visual Cortex 43

data to be two VFMs within another distinct ventral 'clover leaf' cluster [52]. PHC-1 is located just anterior to VO-2, running from the fusiform gyrus into the parahippocampal gyrus (**Figure 8**). Both PHC-1 and PHC-2 represent a full contralateral hemifield of visual space, with the representation of the upper vertical meridian denoting the boundary

In contrast to the posterior medial occipital VFMs, the lateral occipital cortex, with the object-responsive lateral occipital complex (LOC), has been much more difficult to measure in terms of retinotopic organization (**Figure 9**). This region was initially thought to be nonretinotopic or only to contain an 'eccentricity bias' (e.g., [21, 59, 60]). Recently, two VFMs, called LO-1 and LO-2 for 'lateral occipital', were described along the dorsal aspect of the LOC (**Figure 11**). LO-1 lies just anterior to V3d, reversing from the upper vertical meridian representation at the boundary into its representation of a full hemifield of visual space [28, 61]. LO-2 is located just inferior to LO-1, with the lower vertical meridian represented at the shared boundary of these two maps. The foveal representations of these two VFMs are located with the confluent foveal representations of V1, V2, V3, and hV4 on the occipital

**Figure 9. Lateral Occipitotemporal Cortex.** The anatomical region containing lateral visual areas LO-1, LO-2, and the hMT+ cluster is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. STS, superior temporal sulcus; ITS, inferior temporal sulcus; LOG, lateral occipital gyri; AnG, angular gyrus; IPS, intraparietal sulcus; TOS,

In addition, regions just inferior to these maps have been shown to be responsive to lateralized visual stimuli, but have not yet been divided into specific VFMs [27, 62-64]. In

transverse occipital sulcus; OP, occipital pole. Other details as in **Figure 6**.

**3.3. Visual field maps in lateral occipitotemporal cortex** 

between PHC-1 and PHC-2.

pole.

With improvements in measurement techniques, we clarified the retinotopic organization of this ventral region [11, 46]. Our experiments defined three VFMs in ventral occipital cortex: hV4, VO-1, and VO-2, which we showed to be involved in the color and object processing pathways. HV4 is a hemifield map on the posterior fusiform gyrus that directly abuts the upper hemifield representation of V3v and shares a common eccentricity orientation with the confluent foveal representations of V1, V2, and V3 (**Figures 7, 8**). Anterior to the peripheral representation of hV4 is a distinct group of VFMs with a shared foveal representation separate from that of V1, V2, V3, and hV4. We have termed this organization of a discrete group of VFMs a *'clover leaf' cluster*, as described in more detail in **Section 4** below. This 'clover leaf' cluster contains at least two full hemifield representations of visual space: VO-1 and VO-2. The posterior portion of VO-1 is adjacent to the relatively peripheral visual field representation of hV4 and also abuts the peripheral V3v representation on the lingual gyrus. The posterior border of VO-1 represents the lower vertical meridian, and the anterior region represents the upper vertical meridian. This anterior upper vertical meridian reverses into the VO-2 hemifield map. The eccentricity gradient of VO-1 and VO-2 runs from the shared foveal representation on the fusiform gyrus anteromedially towards the more peripheral representation along the collateral sulcus and more anterior fusiform gyrus. Our recent measurements have suggested that additional maps (e.g., VO-3, VO-4) may be identified within this 'clover leaf' cluster in the future [52-54].

Our findings regarding hV4 and its neighbors have since been supported by measurements from several independent studies [28, 29, 39, 55-58]. Hansen et al. [57] initially continued the search for a hV4 organization more homologous to the split-hemifield of macaque V4 by proposing that a small section of dorsal lateral occipitotemporal cortex represented the inferior vertical meridian of hV4. This organization then left an hV4 division on the ventral surface that represented the full upper visual quarterfield adjacent to V3v plus some additional part of the visual field into the lower visual quarterfield. However, this organization conflicts with the now widely accepted dorsal occipitotemporal organization of LO-1 and LO-2 described below [28, 29, 39]. Further, additional measurements have now repeatedly confirmed 1) the full hemifield span of the ventral hV4 hemifield and 2) demonstrated that artifacts from a regional draining vein may in some subjects interfere with the accurate measurement of this section of hV4 [30].

Beyond hV4, VO-1, and VO-2, Arcaro et al. [55] defined two additional VFMs in this region that overlap with the parahippocampal place area, PHC-1 and PHC-2. Like the VFMs of the VO cluster, PHC-1 and PHC-2 also share a distinct foveal representation and appear in our data to be two VFMs within another distinct ventral 'clover leaf' cluster [52]. PHC-1 is located just anterior to VO-2, running from the fusiform gyrus into the parahippocampal gyrus (**Figure 8**). Both PHC-1 and PHC-2 represent a full contralateral hemifield of visual space, with the representation of the upper vertical meridian denoting the boundary between PHC-1 and PHC-2.

## **3.3. Visual field maps in lateral occipitotemporal cortex**

42 Visual Cortex – Current Status and Perspectives

only performed on one quarterfield of visual space.

identified within this 'clover leaf' cluster in the future [52-54].

with the accurate measurement of this section of hV4 [30].

V8. Using harmonic stimuli that alternated in luminance and chrominance, they also showed color responsivity within V8, although this stimulus type also would stimulate regions responsive to variations in luminance. Following the model of macaque cortex, Tootell and Hadjikhani [21] searched for a quarterfield map in dorsal occipital cortex to pair with their putative V4v. They failed to find the map and concluded that it did not exist. They did not resolve why an isolated quarterfield map would exist, a strange organization which would necessitate that whatever computation was subserved by putative V4v was

With improvements in measurement techniques, we clarified the retinotopic organization of this ventral region [11, 46]. Our experiments defined three VFMs in ventral occipital cortex: hV4, VO-1, and VO-2, which we showed to be involved in the color and object processing pathways. HV4 is a hemifield map on the posterior fusiform gyrus that directly abuts the upper hemifield representation of V3v and shares a common eccentricity orientation with the confluent foveal representations of V1, V2, and V3 (**Figures 7, 8**). Anterior to the peripheral representation of hV4 is a distinct group of VFMs with a shared foveal representation separate from that of V1, V2, V3, and hV4. We have termed this organization of a discrete group of VFMs a *'clover leaf' cluster*, as described in more detail in **Section 4** below. This 'clover leaf' cluster contains at least two full hemifield representations of visual space: VO-1 and VO-2. The posterior portion of VO-1 is adjacent to the relatively peripheral visual field representation of hV4 and also abuts the peripheral V3v representation on the lingual gyrus. The posterior border of VO-1 represents the lower vertical meridian, and the anterior region represents the upper vertical meridian. This anterior upper vertical meridian reverses into the VO-2 hemifield map. The eccentricity gradient of VO-1 and VO-2 runs from the shared foveal representation on the fusiform gyrus anteromedially towards the more peripheral representation along the collateral sulcus and more anterior fusiform gyrus. Our recent measurements have suggested that additional maps (e.g., VO-3, VO-4) may be

Our findings regarding hV4 and its neighbors have since been supported by measurements from several independent studies [28, 29, 39, 55-58]. Hansen et al. [57] initially continued the search for a hV4 organization more homologous to the split-hemifield of macaque V4 by proposing that a small section of dorsal lateral occipitotemporal cortex represented the inferior vertical meridian of hV4. This organization then left an hV4 division on the ventral surface that represented the full upper visual quarterfield adjacent to V3v plus some additional part of the visual field into the lower visual quarterfield. However, this organization conflicts with the now widely accepted dorsal occipitotemporal organization of LO-1 and LO-2 described below [28, 29, 39]. Further, additional measurements have now repeatedly confirmed 1) the full hemifield span of the ventral hV4 hemifield and 2) demonstrated that artifacts from a regional draining vein may in some subjects interfere

Beyond hV4, VO-1, and VO-2, Arcaro et al. [55] defined two additional VFMs in this region that overlap with the parahippocampal place area, PHC-1 and PHC-2. Like the VFMs of the VO cluster, PHC-1 and PHC-2 also share a distinct foveal representation and appear in our In contrast to the posterior medial occipital VFMs, the lateral occipital cortex, with the object-responsive lateral occipital complex (LOC), has been much more difficult to measure in terms of retinotopic organization (**Figure 9**). This region was initially thought to be nonretinotopic or only to contain an 'eccentricity bias' (e.g., [21, 59, 60]). Recently, two VFMs, called LO-1 and LO-2 for 'lateral occipital', were described along the dorsal aspect of the LOC (**Figure 11**). LO-1 lies just anterior to V3d, reversing from the upper vertical meridian representation at the boundary into its representation of a full hemifield of visual space [28, 61]. LO-2 is located just inferior to LO-1, with the lower vertical meridian represented at the shared boundary of these two maps. The foveal representations of these two VFMs are located with the confluent foveal representations of V1, V2, V3, and hV4 on the occipital pole.

**Figure 9. Lateral Occipitotemporal Cortex.** The anatomical region containing lateral visual areas LO-1, LO-2, and the hMT+ cluster is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. STS, superior temporal sulcus; ITS, inferior temporal sulcus; LOG, lateral occipital gyri; AnG, angular gyrus; IPS, intraparietal sulcus; TOS, transverse occipital sulcus; OP, occipital pole. Other details as in **Figure 6**.

In addition, regions just inferior to these maps have been shown to be responsive to lateralized visual stimuli, but have not yet been divided into specific VFMs [27, 62-64]. In

macaque, there has been increasing electrophysiological evidence that neurons in homologous regions may maintain a high degree of retinotopic sensitivity [65]. It is important to note again that the presence of organized representations of visual space in this region can still allow for the stimulus size and position invariance frequently described across the objectresponsive LOC. PRF sizes are expected to be large across this region, but may maintain just enough dispersion of pRF centers to allow for slightly different preferred tuning of responses to visual space. Thus, the responses to object stimuli from this cluster could remain both invariant to stimulus size and position over a wide field of view while retaining visuospatial information. With LO-1 and LO-2, the definition of additional VFMs spanning LOC would form the lateral half of an occipital pole 'clover leaf' cluster and would provide a concrete, visuospatial framework for reliably localizing specific computational regions within lateral object-responsive cortex [5, 12, 66, 67]. This inferior region of the LOC merges with the most lateral lower vertical meridian representation of hV4 [11] and is also subject to issues with the 'venous eclipse' fMRI artifact due to the presence of a draining vein here that has a somewhat variable position across subjects within this region [30]. This artifact has likely significantly contributed to the difficulties in measuring organized VFMs within this region.

Visual Field Map Organization in Human Visual Cortex 45

terminology TO-1, TO-2, TO-3, and TO-4 for these VFMs; these would correspond to Kolster

Amano et al. [29] and Kolster et al. [39] disagreed, however, on the specific organization of the eccentricity gradients between the LO VFMs and the TO/hMT+ VFMs [29, 39]. Amano et al. [29] described LO-2 and TO-1 as positioned as a strip of VFMs on the lateral occipital surface, with hemifield representations sharing meridian boundaries from V3d, to LO-1, to LO-2, to TO-1, to TO-2. In this organization, LO-2 and TO-1 would share a boundary that represents the lower vertical meridian. TO-1 and TO-2 would still meet at a discrete foveal representation that runs only from the foveal representation to a more superior peripheral representation that merges only superiorly with that of LO-2. In contrast, Kolster et al. [39] describe the TO maps as a distinct cluster anterior to LOC and the LO maps, which do not share any meridian boundaries. With this cluster organization, it is the peripheral representation that is shared between the LO and TO maps. Our measurements confirm those of Kolster et al. [5, 11, 39, 66, 67]. The four TO maps form a complete 'clover leaf' cluster that merges with the eccentricity gradients of the LO VFMs at their peripheral eccentricity representations. This VFM definition can be seen in the data in Figure 1 of Amano et al. [29], but was differently interpreted prior to the measurements of the full TO 'clover leaf' cluster. Finally, Kolster et al. [39] describe an additional putative new cluster of two VFMs inferior to the TO/hMT cluster they term putative human PIT (phPIT) based on possible homology to macaque PIT [39]. These are the first measurements of these VFMs in human and have not yet been verified by other labs. As techniques improve and we learn more about the effects of the venous eclipse artifact in this region [30], we expect that multiple VFMs will be measured

by multiple labs in this and surrounding regions in future measurements [53, 66, 67].

Beyond the medial part of the dorsal lower vertical meridian representation of human V3d, lies a series of hemifield VFMs running from the transverse occipital sulcus (TOS) up along the medial wall of the intraparietal sulcus (IPS) (**Figures 10, 11**; e.g., [12, 70]). The first maps bordering V3d are V3A and V3B [71-73]. These two maps share another discrete foveal representation within the TOS, forming a complete 'clover leaf' cluster [5]. V3A has some similarities to macaque V3A and is thought to play a role in motion processing; the computations subserved by V3B are not yet known [71, 72, 74, 75]. V3B was originally described Smith et al. [72] as at least a quarterfield of visual space next to V3A. This definition was expanded to cover a while hemifield of visual space by Press et al. [71], who noted that this region of V3A had a full hemifield polar angle gradient with an eccentricity gradient expanding concentrically from a central position within this hemifield. Such an organization necessitates that two representations of visual space are represented; thus, V3A and V3B were described in Press et al. [71]as a cluster of two VFMs sharing a distinct fovea. In comparing the initial V3B definitions from Smith et al. [72] and later definitions of LO-1 [27- 29], which is just lateral and inferior to V3B, it is possible that the original V3B definitions were measuring a part of what was later called LO-1 and not the same hemifield that was described in Press et al. [71] as V3B. The field now generally considers V3A and V3B as the cluster of two VFMs in the TOS, and LO-1 and LO-2 as the dorsal aspect of the LO [12, 27-29].

**3.4. Visual field maps in posterior parietal cortex** 

et al.'s [39] MT, pMST, pFST, and pV4t, respectively.

Anterior to LOC, along the banks of the inferior temporal sulcus, lies the TO cluster, also known as hMT+ (**Figure 9**). Amano et al. [29] were able to provide the first clear measurements of VFMs in the human motion-selective MT complex (hMT+) using the pRF methods described above [68]. They named these two new hemifield VFMs TO-1 and TO-2, for temporal-occipital areas 1 and 2. These maps are positioned just anterior to the LO maps and share another distinct foveal representation. The representations of visual of space across these maps run anteriorly from the lower vertical meridian at the posterior border of TO-1, to the shared upper vertical meridian at the border of TO-1 and TO-2, to the lower vertical meridian at the anterior border. Following measurements by Huk et al. [68] that demonstrated retinotopic organization within the MT subdivision and differentiated human MT and human MST using dissociable motion stimuli, it is likely that TO-1 is the same as human MT, and TO-2 is the same as human MST.

Recently, Kolster et al. [39] verified that both TO-1 and TO-2 represent full, organized hemifields of visual space, but used a naming scheme homologous to macaque with MT and 'putative' MST (pMST). They also expanded the VFM measurements of this region, showing that the TO/hMT+ cluster contains four full hemifields of visual space. The two additional hemifields of visual space merge at the confluent fovea of TO-1 and TO-2 and are located just inferior to TO-1 and TO-2. They suggest that these VFMs correspond to macaque FST and V4t. Human 'putative' FST (pFST) is positioned anterior to human 'putative' V4t (pV4t), and the two VFMs meet an upper vertical meridian representation. Around this cluster, MT/V5 (TO-1) then shares a lower vertical meridian border with pV4t, and pMST (TO-2) shares a lower vertical meridian border with pFST. A similar clustered organization of these motion responsive VFMs in macaque has also recently been demonstrated by Kolster et al. [69]. We confirm these measurements with data showing the 'clover leaf' cluster organization for this group of VFMs in the human MT+ complex [5, 11, 12, 66, 67]. Because the homology to monkey has not yet been verified, we prefer the anatomy-based terminology TO-1, TO-2, TO-3, and TO-4 for these VFMs; these would correspond to Kolster et al.'s [39] MT, pMST, pFST, and pV4t, respectively.

Amano et al. [29] and Kolster et al. [39] disagreed, however, on the specific organization of the eccentricity gradients between the LO VFMs and the TO/hMT+ VFMs [29, 39]. Amano et al. [29] described LO-2 and TO-1 as positioned as a strip of VFMs on the lateral occipital surface, with hemifield representations sharing meridian boundaries from V3d, to LO-1, to LO-2, to TO-1, to TO-2. In this organization, LO-2 and TO-1 would share a boundary that represents the lower vertical meridian. TO-1 and TO-2 would still meet at a discrete foveal representation that runs only from the foveal representation to a more superior peripheral representation that merges only superiorly with that of LO-2. In contrast, Kolster et al. [39] describe the TO maps as a distinct cluster anterior to LOC and the LO maps, which do not share any meridian boundaries. With this cluster organization, it is the peripheral representation that is shared between the LO and TO maps. Our measurements confirm those of Kolster et al. [5, 11, 39, 66, 67]. The four TO maps form a complete 'clover leaf' cluster that merges with the eccentricity gradients of the LO VFMs at their peripheral eccentricity representations. This VFM definition can be seen in the data in Figure 1 of Amano et al. [29], but was differently interpreted prior to the measurements of the full TO 'clover leaf' cluster.

Finally, Kolster et al. [39] describe an additional putative new cluster of two VFMs inferior to the TO/hMT cluster they term putative human PIT (phPIT) based on possible homology to macaque PIT [39]. These are the first measurements of these VFMs in human and have not yet been verified by other labs. As techniques improve and we learn more about the effects of the venous eclipse artifact in this region [30], we expect that multiple VFMs will be measured by multiple labs in this and surrounding regions in future measurements [53, 66, 67].

### **3.4. Visual field maps in posterior parietal cortex**

44 Visual Cortex – Current Status and Perspectives

macaque, there has been increasing electrophysiological evidence that neurons in homologous regions may maintain a high degree of retinotopic sensitivity [65]. It is important to note again that the presence of organized representations of visual space in this region can still allow for the stimulus size and position invariance frequently described across the objectresponsive LOC. PRF sizes are expected to be large across this region, but may maintain just enough dispersion of pRF centers to allow for slightly different preferred tuning of responses to visual space. Thus, the responses to object stimuli from this cluster could remain both invariant to stimulus size and position over a wide field of view while retaining visuospatial information. With LO-1 and LO-2, the definition of additional VFMs spanning LOC would form the lateral half of an occipital pole 'clover leaf' cluster and would provide a concrete, visuospatial framework for reliably localizing specific computational regions within lateral object-responsive cortex [5, 12, 66, 67]. This inferior region of the LOC merges with the most lateral lower vertical meridian representation of hV4 [11] and is also subject to issues with the 'venous eclipse' fMRI artifact due to the presence of a draining vein here that has a somewhat variable position across subjects within this region [30]. This artifact has likely significantly

contributed to the difficulties in measuring organized VFMs within this region.

human MT, and TO-2 is the same as human MST.

Anterior to LOC, along the banks of the inferior temporal sulcus, lies the TO cluster, also known as hMT+ (**Figure 9**). Amano et al. [29] were able to provide the first clear measurements of VFMs in the human motion-selective MT complex (hMT+) using the pRF methods described above [68]. They named these two new hemifield VFMs TO-1 and TO-2, for temporal-occipital areas 1 and 2. These maps are positioned just anterior to the LO maps and share another distinct foveal representation. The representations of visual of space across these maps run anteriorly from the lower vertical meridian at the posterior border of TO-1, to the shared upper vertical meridian at the border of TO-1 and TO-2, to the lower vertical meridian at the anterior border. Following measurements by Huk et al. [68] that demonstrated retinotopic organization within the MT subdivision and differentiated human MT and human MST using dissociable motion stimuli, it is likely that TO-1 is the same as

Recently, Kolster et al. [39] verified that both TO-1 and TO-2 represent full, organized hemifields of visual space, but used a naming scheme homologous to macaque with MT and 'putative' MST (pMST). They also expanded the VFM measurements of this region, showing that the TO/hMT+ cluster contains four full hemifields of visual space. The two additional hemifields of visual space merge at the confluent fovea of TO-1 and TO-2 and are located just inferior to TO-1 and TO-2. They suggest that these VFMs correspond to macaque FST and V4t. Human 'putative' FST (pFST) is positioned anterior to human 'putative' V4t (pV4t), and the two VFMs meet an upper vertical meridian representation. Around this cluster, MT/V5 (TO-1) then shares a lower vertical meridian border with pV4t, and pMST (TO-2) shares a lower vertical meridian border with pFST. A similar clustered organization of these motion responsive VFMs in macaque has also recently been demonstrated by Kolster et al. [69]. We confirm these measurements with data showing the 'clover leaf' cluster organization for this group of VFMs in the human MT+ complex [5, 11, 12, 66, 67]. Because the homology to monkey has not yet been verified, we prefer the anatomy-based Beyond the medial part of the dorsal lower vertical meridian representation of human V3d, lies a series of hemifield VFMs running from the transverse occipital sulcus (TOS) up along the medial wall of the intraparietal sulcus (IPS) (**Figures 10, 11**; e.g., [12, 70]). The first maps bordering V3d are V3A and V3B [71-73]. These two maps share another discrete foveal representation within the TOS, forming a complete 'clover leaf' cluster [5]. V3A has some similarities to macaque V3A and is thought to play a role in motion processing; the computations subserved by V3B are not yet known [71, 72, 74, 75]. V3B was originally described Smith et al. [72] as at least a quarterfield of visual space next to V3A. This definition was expanded to cover a while hemifield of visual space by Press et al. [71], who noted that this region of V3A had a full hemifield polar angle gradient with an eccentricity gradient expanding concentrically from a central position within this hemifield. Such an organization necessitates that two representations of visual space are represented; thus, V3A and V3B were described in Press et al. [71]as a cluster of two VFMs sharing a distinct fovea. In comparing the initial V3B definitions from Smith et al. [72] and later definitions of LO-1 [27- 29], which is just lateral and inferior to V3B, it is possible that the original V3B definitions were measuring a part of what was later called LO-1 and not the same hemifield that was described in Press et al. [71] as V3B. The field now generally considers V3A and V3B as the cluster of two VFMs in the TOS, and LO-1 and LO-2 as the dorsal aspect of the LO [12, 27-29].

Visual Field Map Organization in Human Visual Cortex 47

knowledge, with the one exception, to show anything but the polar angle responses in the

**Figure 11. The Current State of Contiguous Visual Field Maps in Occipital, Parietal, and Temporal Lobes of Human Cortex.** The figure depicts a cartoon overview of hemifield representations of visual field maps (VFMs) in a single right hemisphere if the cortical sheet of that hemisphere were removed from its usual position and laid flat. The center of the blue region is centered on the occipital pole, while the purple region is in parietal cortex, the red and yellow regions are on the ventral surface of the occipital and temporal lobes, and the green region is on the lateral surface of the temporal lobe. Each color except purple represents a likely or confirmed 'clover leaf' cluster of VFMs. Solid black lines indicate the borders between 'clover leaf' clusters and dotted lines indicate the borders between hemifield representations within clusters. In the purple region, the organization is much less clear, particularly for maps outlined in dotted rather than solid black lines (IPS-1, IPS-2, IPS-3, IPS-4, IPS-5, SPL1, and V6A). See main text for more details.

published data (main or supplemental).

**Figure 10. Posterior Parietal Cortex.** The anatomical region containing dorsal posterior parietal cortex visual areas V3A, V3B, and the IPS maps is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. STS, superior temporal sulcus; IPS, intraparietal sulcus; POS, parieto-occipital sulcus; TOS, transverse occipital sulcus; OP, occipital pole. Other details as in **Figure 6**.

Entering into the IPS, the next VFM anterior to the V3A/V3B cluster is IPS-0 (formerly called V7) [75, 76]. IPS-0 has a foveal representation distinct from the V3A/V3B cluster and represents a full hemifield of contralateral visual space, running from the lower vertical meridian representation at the posterior border to the upper vertical meridian representation anteriorly. Beyond IPS-0, a series of polar angle representations extends from IPS-0 along the medial wall of the IPS. All these polar angler representations have been primarily measured using attentionally demanding traveling wave stimuli, consistent with the description of this parietal region as having a role in spatial attention [74-79]. These representations have primarily been described as reversing smoothly through a strip of several hemifield representations from IPS-0 to IPS-5 [70, 76-78, 80-82].

It is important here to differentiate "polar angle representations" from complete "VFMs." For the initial part of the IPS, VFM IPS-0 (V7) has been relatively well characterized [76, 83]. However, VFMs IPS-1 through IPS-5 as well as human LIP (a possible homologous region to macaque LIP which has been suggested to align with IPS-1 or IPS-2) have been exclusively presented in the literature to date as only polar angle maps [70, 76-82, 84, 85], except for the one example we have found of an eccentricity representation in two hemispheres of one subject (see Figure 4 in [81]). Swisher et al. [81] note, for example, that "we reliably find a continuous gradient of eccentricity response phase along the V7(IPS-0)/ IPS-1 border," but that "the eccentricity representation in IPS-3/4 is less clear." Some of these studies have reported that eccentricity mapping was used in the VFM definitions, but failed to our knowledge, with the one exception, to show anything but the polar angle responses in the published data (main or supplemental).

46 Visual Cortex – Current Status and Perspectives

occipital pole. Other details as in **Figure 6**.

**Figure 10. Posterior Parietal Cortex.** The anatomical region containing dorsal posterior parietal cortex visual areas V3A, V3B, and the IPS maps is shown within the black dotted circle on an inflated rendering of the cortical surface of a single left hemisphere from one subject. STS, superior temporal sulcus; IPS, intraparietal sulcus; POS, parieto-occipital sulcus; TOS, transverse occipital sulcus; OP,

Entering into the IPS, the next VFM anterior to the V3A/V3B cluster is IPS-0 (formerly called V7) [75, 76]. IPS-0 has a foveal representation distinct from the V3A/V3B cluster and represents a full hemifield of contralateral visual space, running from the lower vertical meridian representation at the posterior border to the upper vertical meridian representation anteriorly. Beyond IPS-0, a series of polar angle representations extends from IPS-0 along the medial wall of the IPS. All these polar angler representations have been primarily measured using attentionally demanding traveling wave stimuli, consistent with the description of this parietal region as having a role in spatial attention [74-79]. These representations have primarily been described as reversing smoothly through a strip of

It is important here to differentiate "polar angle representations" from complete "VFMs." For the initial part of the IPS, VFM IPS-0 (V7) has been relatively well characterized [76, 83]. However, VFMs IPS-1 through IPS-5 as well as human LIP (a possible homologous region to macaque LIP which has been suggested to align with IPS-1 or IPS-2) have been exclusively presented in the literature to date as only polar angle maps [70, 76-82, 84, 85], except for the one example we have found of an eccentricity representation in two hemispheres of one subject (see Figure 4 in [81]). Swisher et al. [81] note, for example, that "we reliably find a continuous gradient of eccentricity response phase along the V7(IPS-0)/ IPS-1 border," but that "the eccentricity representation in IPS-3/4 is less clear." Some of these studies have reported that eccentricity mapping was used in the VFM definitions, but failed to our

several hemifield representations from IPS-0 to IPS-5 [70, 76-78, 80-82].

**Figure 11. The Current State of Contiguous Visual Field Maps in Occipital, Parietal, and Temporal Lobes of Human Cortex.** The figure depicts a cartoon overview of hemifield representations of visual field maps (VFMs) in a single right hemisphere if the cortical sheet of that hemisphere were removed from its usual position and laid flat. The center of the blue region is centered on the occipital pole, while the purple region is in parietal cortex, the red and yellow regions are on the ventral surface of the occipital and temporal lobes, and the green region is on the lateral surface of the temporal lobe. Each color except purple represents a likely or confirmed 'clover leaf' cluster of VFMs. Solid black lines indicate the borders between 'clover leaf' clusters and dotted lines indicate the borders between hemifield representations within clusters. In the purple region, the organization is much less clear, particularly for maps outlined in dotted rather than solid black lines (IPS-1, IPS-2, IPS-3, IPS-4, IPS-5, SPL1, and V6A). See main text for more details.

While polar angle representations (i.e., IPS-1/2/3/4/5) are excellent indicators that one or a number of VFMs may exist in a given location, they do not provide evidence of a unique, complete VFM. In the case of the IPS polar angle gradients, there could be any number of reversals of eccentricity representations along the polar iso-angle lines. Without obtaining clear, reliable orthogonal maps of both dimensions, it is impossible to know exactly how to divide up a region into multiple representations of visual space. We encountered such a division of what had appeared to be a single hemifield polar angle represenation into multiple VFMs given a central foveal representation previously in the segregation of human V3A into the 2-map cluster with V3B [71]. In those measurements, what had appeared to simply be a large single hemifield representation of V3A, we showed to actually be divided into two full hemifield maps of visual space along the centrally positioned foveal representation [5]. Current work is suggesting that the parietal cortex anterior to the V3A/V3B cluster may also in fact contain several additional 'clover leaf' clusters with VFMs that share several confluent, discrete foveae along the IPS [86, 87].

Visual Field Map Organization in Human Visual Cortex 49

attention and eye movement control (**Figure 12**). Like many of the measurements shown for the IPS region, these frontal regions also lack measurements of orthogonal eccentricity representations. It remains to be seen whether these attentional regions contain only very coarse eccentricity gradients that is difficult to measure or whether future measurements will unveil more detailed eccentricity gradients and possibly multiple VFMs within each

**Figure 12. Frontal Cortex.**SFS, superior frontal sulcus; IFS, inferior frontal sulcus; PreCS, precentral

**4. Organizational patterns of visual field maps across human cortex**

As increasing numbers of VFMs have been defined in human visual cortex, one question that has arisen is whether there is an organizing principle for the distribution of these maps across visual cortex [5, 11, 58, 60, 69, 97, 98]. A basic approach that has worked for early visual cortex has been to define strings of VFMs along contiguous strips of occipital cortex, with adjacent portions of the maps representing similar portions of visual space, but performing different computations [3]. This configuration of maps again allows for more efficient connections between neurons in different maps, such that for a given portion of

It is important to note that the stimulus paradigms used to define these regions do not differentiate between retina-centered (retinotopic) and head-, gaze-, body-, or worldcentered (spatiotopic) topographic organization. While the VFMs in the occipital lobe have been shown to be retinotopically organized [92], there is more controversy regarding if and where the visual system transforms from retinotopic to spatiotopic organization. Spatiotopic representations have been suggested to be present in some visuo-motor pathways (for review, see [93]), but have also been shown to not be necessary for such transformations [94]. Similarly, the native coordinate system of spatial attention has been shown to be retinotopic [95, 96], suggesting that retina-centered visuospatial information is propagated

sulcus; CS, central sulcus; LF, lateral fissure. Other details as in **Figure 6**.

topographic region.

throughout the visual system.

The medial wall of the parietal lobe has also been shown to contain at least three additional retinotopically organized regions. One VFM called V6 lies along the parieto-occipital sulcus just anterior to the peripheral representation of V3d [88, 89]. V6 represents the contralateral hemifield, with an inferior upper vertical meridian running superiorly to the lower vertical meridian representation and a distinct eccentricity gradient that extends far into the periphery. V6 has been shown to be involved in particular types of motion processing, such as pattern and self motion [89]. Adjacent to V6 is another VFM termed V6A [88-90]. Both areas are named to represent their expected homology to macaque. Human V6A has primarily been measured using visual stimuli that activate the far periphery and is likely involved in visuomotor integration [88, 89]. V6A contains at least a coarse representation of the contralateral hemifield and a very peripheral eccentricity gradient with expansion the representation of > 35° visual angle. The last retinotopic region so far measured here is named SPL1 (for superior parietal lobule 1) and is located just medial to the series of maps running along the IPS [70, 82, 85]. Although its position appears somewhat variable across subjects and publications with respect to the specific IPS maps it borders, it appears to be primarily located adjacent to IPS-2, IPS-3 and/or IPS-4. The polar angle gradient of SPL1 runs from its lower vertical meridian border along these IPS maps inferiorly to its upper vertical meridian border. To our knowledge, no eccentricity gradients for this region have yet been demonstrated in the literature. Like the other maps along the IPS, SPL1 may also play a role in spatial attention.

### **3.5. Visual field maps in frontal cortex**

Several topographically organized representations of the contralateral hemifield have also been demonstrated in frontal cortex by a few studies using a variety of stimuli including TWR, visual spatial attention tasks, and memory-guided saccade tasks [70, 78, 79, 84, 85, 91]. These topographic representations of the polar angle of visual space arise in the frontal eye fields (FEF), the supplementary eye fields (SEF), dorsolateral prefrontal cortex (DLPFC), and precentral cortex (pre-CC), regions involved in complex visual processing of spatial attention and eye movement control (**Figure 12**). Like many of the measurements shown for the IPS region, these frontal regions also lack measurements of orthogonal eccentricity representations. It remains to be seen whether these attentional regions contain only very coarse eccentricity gradients that is difficult to measure or whether future measurements will unveil more detailed eccentricity gradients and possibly multiple VFMs within each topographic region.

48 Visual Cortex – Current Status and Perspectives

play a role in spatial attention.

**3.5. Visual field maps in frontal cortex** 

While polar angle representations (i.e., IPS-1/2/3/4/5) are excellent indicators that one or a number of VFMs may exist in a given location, they do not provide evidence of a unique, complete VFM. In the case of the IPS polar angle gradients, there could be any number of reversals of eccentricity representations along the polar iso-angle lines. Without obtaining clear, reliable orthogonal maps of both dimensions, it is impossible to know exactly how to divide up a region into multiple representations of visual space. We encountered such a division of what had appeared to be a single hemifield polar angle represenation into multiple VFMs given a central foveal representation previously in the segregation of human V3A into the 2-map cluster with V3B [71]. In those measurements, what had appeared to simply be a large single hemifield representation of V3A, we showed to actually be divided into two full hemifield maps of visual space along the centrally positioned foveal representation [5]. Current work is suggesting that the parietal cortex anterior to the V3A/V3B cluster may also in fact contain several additional 'clover leaf' clusters with VFMs

The medial wall of the parietal lobe has also been shown to contain at least three additional retinotopically organized regions. One VFM called V6 lies along the parieto-occipital sulcus just anterior to the peripheral representation of V3d [88, 89]. V6 represents the contralateral hemifield, with an inferior upper vertical meridian running superiorly to the lower vertical meridian representation and a distinct eccentricity gradient that extends far into the periphery. V6 has been shown to be involved in particular types of motion processing, such as pattern and self motion [89]. Adjacent to V6 is another VFM termed V6A [88-90]. Both areas are named to represent their expected homology to macaque. Human V6A has primarily been measured using visual stimuli that activate the far periphery and is likely involved in visuomotor integration [88, 89]. V6A contains at least a coarse representation of the contralateral hemifield and a very peripheral eccentricity gradient with expansion the representation of > 35° visual angle. The last retinotopic region so far measured here is named SPL1 (for superior parietal lobule 1) and is located just medial to the series of maps running along the IPS [70, 82, 85]. Although its position appears somewhat variable across subjects and publications with respect to the specific IPS maps it borders, it appears to be primarily located adjacent to IPS-2, IPS-3 and/or IPS-4. The polar angle gradient of SPL1 runs from its lower vertical meridian border along these IPS maps inferiorly to its upper vertical meridian border. To our knowledge, no eccentricity gradients for this region have yet been demonstrated in the literature. Like the other maps along the IPS, SPL1 may also

Several topographically organized representations of the contralateral hemifield have also been demonstrated in frontal cortex by a few studies using a variety of stimuli including TWR, visual spatial attention tasks, and memory-guided saccade tasks [70, 78, 79, 84, 85, 91]. These topographic representations of the polar angle of visual space arise in the frontal eye fields (FEF), the supplementary eye fields (SEF), dorsolateral prefrontal cortex (DLPFC), and precentral cortex (pre-CC), regions involved in complex visual processing of spatial

that share several confluent, discrete foveae along the IPS [86, 87].

**Figure 12. Frontal Cortex.**SFS, superior frontal sulcus; IFS, inferior frontal sulcus; PreCS, precentral sulcus; CS, central sulcus; LF, lateral fissure. Other details as in **Figure 6**.

It is important to note that the stimulus paradigms used to define these regions do not differentiate between retina-centered (retinotopic) and head-, gaze-, body-, or worldcentered (spatiotopic) topographic organization. While the VFMs in the occipital lobe have been shown to be retinotopically organized [92], there is more controversy regarding if and where the visual system transforms from retinotopic to spatiotopic organization. Spatiotopic representations have been suggested to be present in some visuo-motor pathways (for review, see [93]), but have also been shown to not be necessary for such transformations [94]. Similarly, the native coordinate system of spatial attention has been shown to be retinotopic [95, 96], suggesting that retina-centered visuospatial information is propagated throughout the visual system.

## **4. Organizational patterns of visual field maps across human cortex**

As increasing numbers of VFMs have been defined in human visual cortex, one question that has arisen is whether there is an organizing principle for the distribution of these maps across visual cortex [5, 11, 58, 60, 69, 97, 98]. A basic approach that has worked for early visual cortex has been to define strings of VFMs along contiguous strips of occipital cortex, with adjacent portions of the maps representing similar portions of visual space, but performing different computations [3]. This configuration of maps again allows for more efficient connections between neurons in different maps, such that for a given portion of

space, the length of the axons between neurons performing different sets of computations in the process of building the visual percept is minimized [6, 7, 99]. As additional VFMs have been defined in higher order visual cortex, more complex organizing principles for visual cortex have been proposed [5, 11, 58, 60, 69, 97, 98, 100, 101]. One such principle describes VFMs as organized into roughly circular clusters that we have termed 'clover leaf clusters' (**Figure 13**) [5, 11, 66, 67]. These clusters are hypothesized to each contain multiple VFMs which share similar processing properties across that cluster [5, 11, 69]. Several clusters have been partially described, and two have been repeatedly mapped in full: V3A/V3B and TO/hMT+ clusters [5, 11, 12, 29, 39].

Visual Field Map Organization in Human Visual Cortex 51

center to the periphery of the cluster, effectively spanning the radius of the cluster like a spoke on a wheel. We refer to this type of organization – one dimension of space (polar angle) represented radially from center to periphery of a cluster and the other dimension (eccentricity) represented in concentric, circular bands from center to periphery – as being radially orthogonal. Furthermore, we expect that these clusters have consistent locations relative to one another, but that the maps within each cluster may be oriented somewhat differently. Finally, VFMs within each 'clover leaf' clusters are proposed to perform similar types of computations. This broad scale organizational pattern of VFM is currently under investigation by several groups [5, 7, 12, 39, 66, 67, 69], and we expect it to extend across

**5. Additional measurements to refine human visual field map definitions** 

In order to fully investigate visuospatial processing in the human brain, it is important to measure not only patterns of functional VFM organization across cerebral cortex, but also the organization of the white matter tracts connecting these functional regions. Feedforward and feedback information from each VFM must be passed on to other maps up and down the hierarchy of visual processing [18]. The retinotopic human visual system processes portions of visual space in parallel, requiring connections for one type of processing in any particular location of the visual field to be passed to the same location of visual space in the next VFM for the next step in visual processing [102]. An elegant solution to this connectivity problem would be to maintain visuospatial organization in the white matter tracts between clusters. Such a solution allows for molecular cues to guide and maintain the relative structure of VFMs. Any other solution would necessarily require a breakdown of

**Figure 14. White matter connections in human visual cortex.** (a) An example of separate sets of diffusion tensor imaging fibers estimated among maps in parietal cortex. Fibers originated from a 5 mm

hemispheric U-fibers connecting neighboring locations in left intraparietal sulcus (IPS). Blue and green fibers depict 2 subsets of a group of inter-hemispheric fiber tracks that connect to the contralateral IPS.

seed point located in the dorsal polar angle gradients IPS-1. Fibers colored red illustrate intra-

**5.1. White matter connectivity among human visual field maps** 

human visual cortex.

**Figure 13. Orthogonal Dimensions of 'Clover Leaf' Clusters.** Top Left: Eccentricity visual space legend. Each color represents an iso-eccentricity line in the left visual hemifield. Top Right: Polar angle visual space legend. Each color represents an iso-polar angle line in the left visual hemifield. (A) Eccentricity gradients of a typical 'clover leaf' cluster, containing 4 complete representations of the left visual hemifield. Each eccentricity gradient from the center to periphery of visual space runs physically from the center to the periphery of the cluster, such that iso-eccentricity lines form concentric circles about the center of the cluster. (B) A typical 'clover leaf' cluster with 4 complete representations of a hemifield of visual space. Note that, due to the radially orthogonal organization of 'clover leaf' clusters, each individual map must be shaped like a piece of pie. (C) Polar angle gradients of a typical 'clover leaf' cluster. Each iso-polar angle line runs from the center to periphery of the cluster like a 'spoke of the wheel' of the cluster, such that each polar angle gradient runs from one 'spoke of the wheel' of the cluster around to another 'spoke'. Note that, due to the fact that polar angle reversals (adjacent spokes with the same polar angle preference) denote the boundaries between hemifield representations within a 'clover leaf' cluster, there must always be an even number of hemifield representations in each 'clover leaf' cluster. To have an odd number would necessarily require a discrete jump in polar angle representation between hemifield representations, which has so far not been observed in human visual field mapping.

The VFMs within a 'clover leaf' clusters are organized such that the central fovea is represented in the center of the cluster, with more peripheral representations of space represented in more peripheral positions in the cluster in a smooth, orderly fashion. The representation of any given polar angle of space for any given VFM extends out from the center to the periphery of the cluster, effectively spanning the radius of the cluster like a spoke on a wheel. We refer to this type of organization – one dimension of space (polar angle) represented radially from center to periphery of a cluster and the other dimension (eccentricity) represented in concentric, circular bands from center to periphery – as being radially orthogonal. Furthermore, we expect that these clusters have consistent locations relative to one another, but that the maps within each cluster may be oriented somewhat differently. Finally, VFMs within each 'clover leaf' clusters are proposed to perform similar types of computations. This broad scale organizational pattern of VFM is currently under investigation by several groups [5, 7, 12, 39, 66, 67, 69], and we expect it to extend across human visual cortex.

## **5. Additional measurements to refine human visual field map definitions**

## **5.1. White matter connectivity among human visual field maps**

50 Visual Cortex – Current Status and Perspectives

TO/hMT+ clusters [5, 11, 12, 29, 39].

space, the length of the axons between neurons performing different sets of computations in the process of building the visual percept is minimized [6, 7, 99]. As additional VFMs have been defined in higher order visual cortex, more complex organizing principles for visual cortex have been proposed [5, 11, 58, 60, 69, 97, 98, 100, 101]. One such principle describes VFMs as organized into roughly circular clusters that we have termed 'clover leaf clusters' (**Figure 13**) [5, 11, 66, 67]. These clusters are hypothesized to each contain multiple VFMs which share similar processing properties across that cluster [5, 11, 69]. Several clusters have been partially described, and two have been repeatedly mapped in full: V3A/V3B and

**Figure 13. Orthogonal Dimensions of 'Clover Leaf' Clusters.** Top Left: Eccentricity visual space legend. Each color represents an iso-eccentricity line in the left visual hemifield. Top Right: Polar angle visual space legend. Each color represents an iso-polar angle line in the left visual hemifield. (A) Eccentricity gradients of a typical 'clover leaf' cluster, containing 4 complete representations of the left visual hemifield. Each eccentricity gradient from the center to periphery of visual space runs physically from the center to the periphery of the cluster, such that iso-eccentricity lines form concentric circles about the center of the cluster. (B) A typical 'clover leaf' cluster with 4 complete representations of a hemifield of visual space. Note that, due to the radially orthogonal organization of 'clover leaf' clusters, each individual map must be shaped like a piece of pie. (C) Polar angle gradients of a typical 'clover leaf' cluster. Each iso-polar angle line runs from the center to periphery of the cluster like a 'spoke of the wheel' of the cluster, such that each polar angle gradient runs from one 'spoke of the wheel' of the cluster around to another 'spoke'. Note that, due to the fact that polar angle reversals (adjacent spokes with the same polar angle preference) denote the boundaries between hemifield representations within a 'clover leaf' cluster, there must always be an even number of hemifield representations in each 'clover leaf' cluster. To have an odd number would necessarily require a discrete jump in polar angle representation between hemifield representations, which has so far not been observed in human visual field mapping.

The VFMs within a 'clover leaf' clusters are organized such that the central fovea is represented in the center of the cluster, with more peripheral representations of space represented in more peripheral positions in the cluster in a smooth, orderly fashion. The representation of any given polar angle of space for any given VFM extends out from the In order to fully investigate visuospatial processing in the human brain, it is important to measure not only patterns of functional VFM organization across cerebral cortex, but also the organization of the white matter tracts connecting these functional regions. Feedforward and feedback information from each VFM must be passed on to other maps up and down the hierarchy of visual processing [18]. The retinotopic human visual system processes portions of visual space in parallel, requiring connections for one type of processing in any particular location of the visual field to be passed to the same location of visual space in the next VFM for the next step in visual processing [102]. An elegant solution to this connectivity problem would be to maintain visuospatial organization in the white matter tracts between clusters. Such a solution allows for molecular cues to guide and maintain the relative structure of VFMs. Any other solution would necessarily require a breakdown of

**Figure 14. White matter connections in human visual cortex.** (a) An example of separate sets of diffusion tensor imaging fibers estimated among maps in parietal cortex. Fibers originated from a 5 mm seed point located in the dorsal polar angle gradients IPS-1. Fibers colored red illustrate intrahemispheric U-fibers connecting neighboring locations in left intraparietal sulcus (IPS). Blue and green fibers depict 2 subsets of a group of inter-hemispheric fiber tracks that connect to the contralateral IPS.

the organization inherent in connections coming out of a cluster into some other organization, and a rebuilding of the organization inherent in connections going into the next cluster, which is more costly, difficult, and complicated. The MR protocol of diffusion tensor imaging (DTI) is beginning to provide such measurements of connectivity among VFMs early in the visual processing hierarchy (**Figure 14**).

Visual Field Map Organization in Human Visual Cortex 53

invariant object recognition can occur in VFMs without any cost. It is possible that the majority of higher order visual areas are retinotopic, maintaining retinotopically organized, dispersed receptive field centers despite increasingly large receptive field sizes. Why would we evolve an entirely different way to deal with one type of visual information when the simple solution of using large receptive fields and population codes within VFMs does not require a change in organization or connectivity? We contend that any failure to demonstrate retinotopic responses in a visually responsive area must be first carefully evaluated as a failure of the measurement methodology. This is a question whose answer is vital to understanding higher-order cognition, which need not abandon the visuospatial

Our understanding the visuospatial organization of human visual cortex is crucial for our further exploration of the computations subserved by these visual pathways. Because every human brain shares common functional topography that is somewhat variably located anatomically, it is vital to correctly localize common functional areas in individual subjects in order to then study which specific computations are carried out by each area. The widespread technique of mapping anatomy to a common atlas not only destroys information about individual subjects, but also blurs data from adjacent areas within each

This chapter has reviewed the primary measurement techniques for investigating the organization of human visual cortex as well as the present state of knowledge of the visuospatial representations across cortex. Until recently, the best technique for localizing visual field maps was TWR, which is excellent at identifying the centers of pRFs in fMRI voxels. Now, that technique has been surpassed by pRF modeling, which measures not only pRF centers, but the spread of each pRF. Both techniques have been used to successfully localize numerous visual field maps, most of which have been confirmed by numerous laboratories. However, it would be a mistake to assume that the current organizational model is entirely correct or complete. Some polar angle gradients still need to have eccentricity gradients measured in order to correctly determine the number of VFMs in parts of cortex such as posterior parietal cortex. Other maps simply need confirmation in the form of replication by independent laboratories. Perhaps most importantly, the current organizational model will need to be reconsidered in the face of evidence that VFMs are organized into 'clover leaf' clusters. These data will surely be updated as additional measurements both expand the known retinotopic representations in human visual cortex

subject, making it impossible to differentiate computations in adjacent areas.

knowledge gained from low-level visual processing.

and clarify the current VFM definitions.

Alyssa A. Brewer and Brian Barton

*Department of Cognitive Sciences, University of California, Irvine, USA* 

**Author details** 

**6. Conclusion** 

DTI is based upon a computer-assisted analysis of multiple diffusion-weighted images and uses a Gaussian model of diffusion, called an ellipsoid or 'tensor', to measure the mobility of water molecules within human tissue [103]. Traditional DTI measurements include fractional anisotropy (FA) and mean diffusivity (MD); custom software packages allow one to now also estimate the shapes, properties and destinations of white matter fiber tracts from DTI data (e.g., [104, 105]; http://white.stanford.edu/software/). The combination of DTI and fiber-tracking algorithms that combine tensors with similar principal diffusion directions can produce estimates of the major fiber bundles in the brain of individual subjects [106, 107]. FA, MD, and estimated fiber track locations and densities can be compared across individuals and between groups of subjects [108].

DTI has now successfully been used to measure the organization of the human optic tracts and optic radiations from the retina to the LGN to V1 [50, 109]. These measurements of the initial visual tracts match those measured by other methods [12, 110] and are now being used in studies examining how the optic radiations differ in specific patient populations (e.g., [32]). In cortex, DTI measurements have further demonstrated that the retinotopic organization of V1 is maintained in connections between homologous VFMs across hemispheres, with upper and lower visual field sections running together through the splenium of the corpus callosum [108, 111, 112]. These emerging measurements of white matter connectivity among VFMs will aid us both in refining our definitions of specific VFM boundaries throughout the visual processing hierarchy, especially in regions with more ambiguous VFM measurements, and in expanding our understanding of specific computational pathways.

## **5.2. Integration of visual field maps with other functional measurements in human visual cortex**

The findings from combined structural and functional measurements have a profound impact on the study of vision in the brain. Any visual area that is organized retinotopically is subject to the constraints common to all VFMs in the human brain. Many areas of the brain have been considered "non-retinotopic" or containing only an "eccentricity bias," because laboratories using the field standard TWR were not able to demonstrate complete, precise retinotopic organization (e.g., [21, 97, 113]). Typically, this has been interpreted to mean that the early portion of the visual system is the only part that is retinotopic, and, at some point midway in the hierarchy, there must be a fundamental change in the way that the visual system is constructed. Not only is this a major theoretical claim, which would require a potentially complex transformation from a retinotopic framework to some other non-spatial organization, but it side-steps the basic fact that properties such as retinal-sizeinvariant object recognition can occur in VFMs without any cost. It is possible that the majority of higher order visual areas are retinotopic, maintaining retinotopically organized, dispersed receptive field centers despite increasingly large receptive field sizes. Why would we evolve an entirely different way to deal with one type of visual information when the simple solution of using large receptive fields and population codes within VFMs does not require a change in organization or connectivity? We contend that any failure to demonstrate retinotopic responses in a visually responsive area must be first carefully evaluated as a failure of the measurement methodology. This is a question whose answer is vital to understanding higher-order cognition, which need not abandon the visuospatial knowledge gained from low-level visual processing.

## **6. Conclusion**

52 Visual Cortex – Current Status and Perspectives

computational pathways.

**human visual cortex** 

VFMs early in the visual processing hierarchy (**Figure 14**).

compared across individuals and between groups of subjects [108].

the organization inherent in connections coming out of a cluster into some other organization, and a rebuilding of the organization inherent in connections going into the next cluster, which is more costly, difficult, and complicated. The MR protocol of diffusion tensor imaging (DTI) is beginning to provide such measurements of connectivity among

DTI is based upon a computer-assisted analysis of multiple diffusion-weighted images and uses a Gaussian model of diffusion, called an ellipsoid or 'tensor', to measure the mobility of water molecules within human tissue [103]. Traditional DTI measurements include fractional anisotropy (FA) and mean diffusivity (MD); custom software packages allow one to now also estimate the shapes, properties and destinations of white matter fiber tracts from DTI data (e.g., [104, 105]; http://white.stanford.edu/software/). The combination of DTI and fiber-tracking algorithms that combine tensors with similar principal diffusion directions can produce estimates of the major fiber bundles in the brain of individual subjects [106, 107]. FA, MD, and estimated fiber track locations and densities can be

DTI has now successfully been used to measure the organization of the human optic tracts and optic radiations from the retina to the LGN to V1 [50, 109]. These measurements of the initial visual tracts match those measured by other methods [12, 110] and are now being used in studies examining how the optic radiations differ in specific patient populations (e.g., [32]). In cortex, DTI measurements have further demonstrated that the retinotopic organization of V1 is maintained in connections between homologous VFMs across hemispheres, with upper and lower visual field sections running together through the splenium of the corpus callosum [108, 111, 112]. These emerging measurements of white matter connectivity among VFMs will aid us both in refining our definitions of specific VFM boundaries throughout the visual processing hierarchy, especially in regions with more ambiguous VFM measurements, and in expanding our understanding of specific

**5.2. Integration of visual field maps with other functional measurements in** 

The findings from combined structural and functional measurements have a profound impact on the study of vision in the brain. Any visual area that is organized retinotopically is subject to the constraints common to all VFMs in the human brain. Many areas of the brain have been considered "non-retinotopic" or containing only an "eccentricity bias," because laboratories using the field standard TWR were not able to demonstrate complete, precise retinotopic organization (e.g., [21, 97, 113]). Typically, this has been interpreted to mean that the early portion of the visual system is the only part that is retinotopic, and, at some point midway in the hierarchy, there must be a fundamental change in the way that the visual system is constructed. Not only is this a major theoretical claim, which would require a potentially complex transformation from a retinotopic framework to some other non-spatial organization, but it side-steps the basic fact that properties such as retinal-sizeOur understanding the visuospatial organization of human visual cortex is crucial for our further exploration of the computations subserved by these visual pathways. Because every human brain shares common functional topography that is somewhat variably located anatomically, it is vital to correctly localize common functional areas in individual subjects in order to then study which specific computations are carried out by each area. The widespread technique of mapping anatomy to a common atlas not only destroys information about individual subjects, but also blurs data from adjacent areas within each subject, making it impossible to differentiate computations in adjacent areas.

This chapter has reviewed the primary measurement techniques for investigating the organization of human visual cortex as well as the present state of knowledge of the visuospatial representations across cortex. Until recently, the best technique for localizing visual field maps was TWR, which is excellent at identifying the centers of pRFs in fMRI voxels. Now, that technique has been surpassed by pRF modeling, which measures not only pRF centers, but the spread of each pRF. Both techniques have been used to successfully localize numerous visual field maps, most of which have been confirmed by numerous laboratories. However, it would be a mistake to assume that the current organizational model is entirely correct or complete. Some polar angle gradients still need to have eccentricity gradients measured in order to correctly determine the number of VFMs in parts of cortex such as posterior parietal cortex. Other maps simply need confirmation in the form of replication by independent laboratories. Perhaps most importantly, the current organizational model will need to be reconsidered in the face of evidence that VFMs are organized into 'clover leaf' clusters. These data will surely be updated as additional measurements both expand the known retinotopic representations in human visual cortex and clarify the current VFM definitions.

## **Author details**

Alyssa A. Brewer and Brian Barton *Department of Cognitive Sciences, University of California, Irvine, USA* 

## **Acknowledgement**

This work was supported by University of California, Irvine, startup funds to A.A.B.

#### **7. References**

[1] W. H. Merigan and J. H. Maunsell, "How parallel are the primate visual pathways?," *Annu Rev Neurosci,* vol. 16, pp. 369-402, 1993.

Visual Field Map Organization in Human Visual Cortex 55

[17] D. C. Van Essen, W. T. Newsome, and J. H. Maunsell, "The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual

[18] D. J. Felleman and D. C. V. Essen, "Distributed hierarchical processing in the primate

[19] D. C. Van Essen, *Organization of visual areas in macaque and human cerebral cortex*. Boston:

[20] D. C. Van Essen, D. J. Felleman, E. A. DeYoe, J. Olavarria, and J. Knierim, "Modular and hierarchical organization of extrastriate visual cortex in the macaque monkey," *Cold* 

[21] R. B. Tootell and N. Hadjikhani, "Where is 'dorsal V4' in human visual cortex? Retinotopic, topographic and functional evidence," *Cereb Cortex,* vol. 11, pp. 298-311,

[22] S. O. Dumoulin and B. A. Wandell, "Population receptive field estimates in human

[23] S. A. Engel, G. H. Glover, and B. A. Wandell, "Retinotopic organization in human visual cortex and the spatial precision of functional MRI," *Cereb Cortex,* vol. 7, pp. 181-92, Mar

[24] E. A. DeYoe, G. J. Carman, P. Bandettini, S. Glickman, J. Wieser, R. Cox, D. Miller, and J. Neitz, "Mapping striate and extrastriate visual areas in human cerebral cortex," *Proc.* 

[25] M. I. Sereno, C. T. McDonald, and J. M. Allman, "Analysis of retinotopic maps in

[26] B. A. Wandell and J. Winawer, "Imaging retinotopic maps in the human brain," *Vision* 

[27] R. Sayres and K. Grill-Spector, "Relating retinotopic and object-selective responses in

[28] J. Larsson and D. J. Heeger, "Two retinotopic visual areas in human lateral occipital

[29] K. Amano, B. A. Wandell, and S. O. Dumoulin, "Visual field maps, population receptive field sizes, and visual field coverage in the human MT+ complex," *J Neurophysiol,* vol.

[30] J. Winawer, H. Horiguchi, R. A. Sayres, K. Amano, and B. A. Wandell, "Mapping hV4

[31] S. Prakash, S. O. Dumoulin, N. Fischbein, B. A. Wandell, and Y. J. Liao, "Congenital achiasma and see-saw nystagmus in VACTERL syndrome," *J Neuroophthalmol,* vol. 30,

[32] N. Levin, S. O. Dumoulin, J. Winawer, R. F. Dougherty, and B. A. Wandell, "Cortical maps and white matter tracts following long period of visual deprivation and retinal

[33] H. A. Baseler, A. Gouws, K. V. Haak, C. Racey, M. D. Crossland, A. Tufail, G. S. Rubin, F. W. Cornelissen, and A. B. Morland, "Large-scale remapping of visual cortex is absent

human lateral occipital cortex," *J Neurophysiol,* vol. 100, pp. 249-67, Jul 2008.

and ventral occipital cortex: the venous eclipse," *J Vis,* vol. 10, p. 1, 2010.

variability," *Vision Res,* vol. 24, pp. 429-48, 1984.

Bradford, 2003.

Apr 2001.

1997.

cerebral cortex," *Cerebral Cortex,* vol. 1, pp. 1-47, 1991.

*Spring Harb Symp Quant Biol,* vol. 55, pp. 679-96, 1990.

visual cortex," *Neuroimage,* vol. 39, pp. 647-60, Jan 15 2008.

extrastriate cortex," *Cerebral Cortex,* vol. 6, pp. 601-620, 1994.

*Natl. Acad. Sci. (USA),* vol. 93, pp. 2382-2386, 1996.

cortex," *J Neurosci,* vol. 26, pp. 13128-42, Dec 20 2006.

image restoration," *Neuron,* vol. 65, pp. 21-31, Jan 14 2010.

*Res,* vol. 51, pp. 718-37, Apr 13 2011.

102, pp. 2704-18, Nov 2009.

pp. 45-8, Mar 2010.


[17] D. C. Van Essen, W. T. Newsome, and J. H. Maunsell, "The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual variability," *Vision Res,* vol. 24, pp. 429-48, 1984.

54 Visual Cortex – Current Status and Perspectives

*Annu Rev Neurosci,* vol. 16, pp. 369-402, 1993.

*Bulletin,* vol. 44, pp. 107–112, 1997.

*Biol Sci,* vol. 245, pp. 151-8, Aug 22 1991.

cortex," *J Vis,* vol. 9, pp. 13 1-22, 2009.

*Neuron,* vol. 56, pp. 366-83, Oct 25 2007.

atlases," *Vision Res,* vol. 41, pp. 1359-78, 2001.

*Natl Acad Sci U S A,* vol. 93, pp. 2382-6, Mar 19 1996.

*Trans Am Ophthalmol Soc,* vol. 93, pp. 123-33, 1995.

them?," *Annu Rev Neurosci,* vol. 27, pp. 369-92, 2004.

visual cortex," *Prog Brain Res,* vol. 165, pp. 21-32, 2007.

This work was supported by University of California, Irvine, startup funds to A.A.B.

[1] W. H. Merigan and J. H. Maunsell, "How parallel are the primate visual pathways?,"

[2] D. L. Adams, L. C. Sincich, and J. C. Horton, "Complete pattern of ocular dominance columns in human primary visual cortex," *J Neurosci,* vol. 27, pp. 10391-403, Sep 26 2007. [3] B. A. Wandell, S. O. Dumoulin, and A. A. Brewer, "Visual cortex in humans," in *Encyclopedia of Neuroscience*. vol. 10, L. R. Squire, Ed., ed Oxford: Academic Press, 2009,

[4] J. H. Kaas, "Topographic Maps are Fundamental to Sensory Processing," *Brain Research* 

[5] B. A. Wandell, A. A. Brewer, and R. F. Dougherty, "Visual field map clusters in human

[6] G. Mitchison, "Neuronal branching patterns and the economy of cortical wiring," *Proc* 

[7] D. B. Chklovskii and A. A. Koulakov, "Maps in the brain: what can we learn from

[8] R. Shapley, M. Hawken, and D. Xing, "The dynamics of visual responses in the primary

[9] F. Moradi and D. J. Heeger, "Inter-ocular contrast normalization in human visual

[10] S. A. Engel, D. E. Rumelhart, B. A. Wandell, A. T. Lee, G. H. Glover, E. J. Chichilnisky, and M. N. Shadlen, "fMRI of human visual cortex," *Nature,* vol. 369, p. 525, Jun 16 1994. [11] A. A. Brewer, J. Liu, A. R. Wade, and B. A. Wandell, "Visual field maps and stimulus selectivity in human ventral occipital cortex," *Nat Neurosci,* vol. 8, pp. 1102-9, Aug 2005. [12] B. A. Wandell, S. O. Dumoulin, and A. A. Brewer, "Visual field maps in human cortex,"

[13] M. I. Sereno, A. M. Dale, J. B. Reppas, K. K. Kwong, J. W. Belliveau, T. J. Brady, B. R. Rosen, and R. B. Tootell, "Borders of multiple visual areas in humans revealed by

[14] D. C. Van Essen, J. W. Lewis, H. A. Drury, N. Hadjikhani, R. B. Tootell, M. Bakircioglu, and M. I. Miller, "Mapping visual cortex in monkeys and humans using surface-based

[15] E. A. DeYoe, G. J. Carman, P. Bandettini, S. Glickman, J. Wieser, R. Cox, D. Miller, and J. Neitz, "Mapping striate and extrastriate visual areas in human cerebral cortex," *Proc* 

[16] D. C. Van Essen, "Behind the optic nerve: an inside view of the primate visual system,"

functional magnetic resonance imaging," *Science,* vol. 268, pp. 889-93., 1995.

cortex," *Philos Trans R Soc Lond B Biol Sci,* vol. 360, pp. 693-707, Apr 29 2005.

**Acknowledgement** 

**7. References** 

pp. 251-257.


in adult humans with macular degeneration," *Nat Neurosci,* vol. 14, pp. 649-55, May 2011.

Visual Field Map Organization in Human Visual Cortex 57

[49] S. Zeki, J. D. G. Watson, C. J. Lueck, K. J. Friston, C. Kennard, and R. S. J. Frackowiak, "A direct demonstration of functional specialization in human visual cortex.," *J.* 

[50] D. J. McKeefry and S. Zeki, "The position and topography of the human colour centre as revealed by functional magnetic resonance imaging," *Brain,* vol. 120, pp. 2229-42, 1997. [51] N. Hadjikhani, A. K. Liu, A. M. Dale, P. Cavanagh, and R. B. Tootell, "Retinotopy and color sensitivity in human visual cortical area V8," *Nat Neurosci,* vol. 1, pp. 235-41, Jul

[52] S. A. Drew, D. E. Asher, B. Barton, and A. A. Brewer, "Pinwheel cartography: New visual field map cluster in the human posterior parahippocampal complex," presented

[53] D. E. Asher and A. A. Brewer, "Hemispheric differences of color responses in human ventral visual cortex," *Vision Sciences Society. Journal of Vision,* vol. 9, pp. 776-776, 2009. [54] D. E. Asher, S. A. Drew, B. Barton, and A. A. Brewer, "Pinwheel cartography: Novel visual field map cluster within human ventro-lateral occipital cortex. ," presented at the

[55] M. J. Arcaro, S. A. McMains, B. D. Singer, and S. Kastner, "Retinotopic organization of

[56] S. Kastner, P. De Weerd, M. A. Pinsk, M. I. Elizondo, R. Desimone, and L. G. Ungerleider, "Modulation of sensory suppression: implications for receptive field sizes

[57] K. A. Hansen, K. N. Kay, and J. L. Gallant, "Topographic organization in and near

[58] C. W. Tyler and A. R. Wade, "Extended concepts of occipital retinotopy," *Current* 

[59] U. Hasson, T. Hendler, D. Ben Bashat, and R. Malach, "Vase or face? A neural correlate of shape-selective grouping processes in the human brain," *J Cogn Neurosci,* vol. 13, pp.

[60] R. Malach, I. Levy, and U. Hasson, "The topography of high-order human object areas,"

[61] J. Larsson, M. S. Landy, and D. J. Heeger, "Orientation-selective adaptation to first- and second-order patterns in human visual cortex," *J Neurophysiol,* vol. 95, pp. 862-81, Feb

[62] A. McKyton and E. Zohary, "Beyond retinotopic mapping: the spatial representation of objects in the human lateral occipital complex," *Cereb Cortex,* vol. 17, pp. 1164-72, May

[63] C. C. Hemond, N. G. Kanwisher, and H. P. Op de Beeck, "A preference for contralateral stimuli in human object- and face-selective cortex," *PLoS One,* vol. 2, p. e574, 2007. [64] M. Niemeier, H. C. Goltz, A. Kuchinad, D. B. Tweed, and T. Vilis, "A contralateral preference in the lateral occipital area: sensory and attentional mechanisms," *Cereb* 

human ventral visual cortex," *J Neurosci,* vol. 29, pp. 10638-52, Aug 26 2009.

in the human visual cortex," *J Neurophysiol,* vol. 86, pp. 1398-411, Sep 2001.

human visual area V4," *J Neurosci,* vol. 27, pp. 11896-911, Oct 31 2007.

*Neuroscience,* vol. 11, pp. 641-649, 1991.

at the Society for Neuroscience, San Diego, CA, 2010.

Society for Neuroscience, San Diego, CA, 2010.

*Medical Imaging Review,* vol. 1, p. 3190329, 2005.

*Trends Cogn Sci,* vol. 6, pp. 176-184, Apr 1 2002.

*Cortex,* vol. 15, pp. 325-31, Mar 2005.

744-53, Aug 15 2001.

2006.

2007.

1998.


[49] S. Zeki, J. D. G. Watson, C. J. Lueck, K. J. Friston, C. Kennard, and R. S. J. Frackowiak, "A direct demonstration of functional specialization in human visual cortex.," *J. Neuroscience,* vol. 11, pp. 641-649, 1991.

56 Visual Cortex – Current Status and Perspectives

vol. 16, pp. 187-98, Apr 1997.

2011.

63, Dec 1997.

Jul 21 2010.

2010.

360, pp. 665-91, Apr 29 2005.

*Sci,* vol. 357, pp. 963-73, Aug 29 2002.

cortex of man," *Nature,* vol. 340, pp. 386-389, 1989.

*Genetics,* vol. 19, pp. 200-206, 2003.

in adult humans with macular degeneration," *Nat Neurosci,* vol. 14, pp. 649-55, May

[34] F. Maes, A. Collignon, D. Vandermeulen, G. Marchal, and P. Suetens, "Multimodality image registration by maximization of mutual information," *IEEE Trans Med Imaging,* 

[35] B. A. Wandell, S. Chial, and B. T. Backus, "Visualization and measurement of the

[36] P. C. Teo, G. Sapiro, and B. A. Wandell, "Creating connected representations of cortical gray matter for functional MRI visualization," *IEEE Trans Med Imaging,* vol. 16, pp. 852-

[37] O. Nestares and D. J. Heeger, "Robust multiresolution alignment of MRI brain

[38] R. F. Dougherty, V. M. Koch, A. A. Brewer, B. Fischer, J. Modersitzki, and B. A. Wandell, "Visual field representations and locations of visual areas V1/2/3 in human

[39] H. Kolster, R. Peeters, and G. A. Orban, "The retinotopic organization of the human middle temporal area MT/V5 and its cortical neighbors," *J Neurosci,* vol. 30, pp. 9801-20,

[40] E. Bartfeld and A. Grinvald, "Relationships between orientation-preference pinwheels, cytochrome oxidase blobs, and ocular-dominance columns in primate striate cortex,"

[41] M. S. Livingstone and D. H. Hubel, "Anatomy and physiology of a color system in the

[42] M. M. Schira, C. W. Tyler, M. Breakspear, and B. Spehar, "The foveal confluence in

[43] M. M. Schira, C. W. Tyler, B. Spehar, and M. Breakspear, "Modeling magnification and anisotropy in the primate foveal confluence," *PLoS Comput Biol,* vol. 6, p. e1000651, Jan

[44] M. G. Rosa, "Visuotopic organisation of primate extrastriate cortex.," in *Cerebral Cortex*. vol. 12, K. J. Rockland KS, Peters A, Ed., ed New York: Plenum Press, 1997, pp. 127–203. [45] M. G. Rosa and R. Tweedale, "Brain maps, great and small: lessons from comparative studies of primate visual cortical organization," *Philos Trans R Soc Lond B Biol Sci,* vol.

[46] A. R. Wade, A. A. Brewer, J. W. Rieger, and B. A. Wandell, "Functional measurements of human ventral occipital cortex: retinotopy and colour," *Philos Trans R Soc Lond B Biol* 

[47] S. B. Hedges and S. Kumar, "Genomic clocks and evolutionary timescales," *Trends in* 

[48] C. J. Lueck, S. Zeki, K. J. Friston, M.-P. Deiber, P. Cope, V. J. Cunningham, A. A. Lammertsma, C. Kennard, and R. S. J. Frackowiak, "The color center in the cerebral

cortical surface," *J Cogn Neurosci,* vol. 12, pp. 739-52, Sep 2000.

volumes," *Magn Reson Med,* vol. 43, pp. 705-15, May 2000.

*Proc Natl Acad Sci U S A,* vol. 89, pp. 11905-9, Dec 15 1992.

primate visual cortex," *J Neurosci,* vol. 4, pp. 309-56, Jan 1984.

human visual cortex," *J Neurosci,* vol. 29, pp. 9050-8, Jul 15 2009.

visual cortex," *J Vis,* vol. 3, pp. 586-98, 2003.


[65] J. J. DiCarlo and J. H. Maunsell, "Anterior inferotemporal neurons of monkeys engaged in object recognition can be highly sensitive to object retinal position," *J Neurophysiol,*  vol. 89, pp. 3264-78, Jun 2003.

Visual Field Map Organization in Human Visual Cortex 59

[80] D. Schluppeck, P. Glimcher, and D. J. Heeger, "Topographic organization for delayed saccades in human posterior parietal cortex," *J Neurophysiol,* vol. 94, pp. 1372-84, Aug

[81] J. D. Swisher, M. A. Halko, L. B. Merabet, S. A. McMains, and D. C. Somers, "Visual topography of human intraparietal sulcus," *J Neurosci,* vol. 27, pp. 5326-37, May 16 2007. [82] C. S. Konen and S. Kastner, "Representation of eye movements and stimulus motion in topographically organized areas of human posterior parietal cortex," *J Neurosci,* vol. 28,

[83] T. Z. Lauritzen, M. D'Esposito, D. J. Heeger, and M. A. Silver, "Top-down flow of visual spatial attention signals from parietal to occipital cortex," *J Vis,* vol. 9, pp. 18 1-14, 2009. [84] D. J. Hagler, Jr. and M. I. Sereno, "Spatial maps in frontal and prefrontal cortex,"

[85] S. M. Szczepanski, C. S. Konen, and S. Kastner, "Mechanisms of spatial attention control

[86] A. A. Brewer and B. Barton, "Pinwheel cartography: Visual field map clusters in posterior parietal cortex that subserve visual attention and working memory,"

[87] H. Kolster, R. Peeters, and G. A. Orban, "Ten retinotopically organized areas in human parietal cortex," presented at the Society for Neuroscience, Washington, D.C., 2011. [88] S. Pitzalis, C. Galletti, R. S. Huang, F. Patria, G. Committeri, G. Galati, P. Fattori, and M. I. Sereno, "Wide-field retinotopy defines human cortical visual area v6," *J Neurosci,* vol.

[89] S. Pitzalis, M. I. Sereno, G. Committeri, P. Fattori, G. Galati, F. Patria, and C. Galletti, "Human v6: the medial motion area," *Cereb Cortex,* vol. 20, pp. 411-24, Feb 2010. [90] P. Dechent and J. Frahm, "Characterization of the human visual V6 complex by functional magnetic resonance imaging," *Eur J Neurosci,* vol. 17, pp. 2201-11, May 2003. [91] D. J. Hagler, Jr., L. Riecke, and M. I. Sereno, "Parietal and superior frontal visuospatial maps activated by pointing and saccades," *Neuroimage,* vol. 35, pp. 1562-77, May 1 2007. [92] J. L. Gardner, E. P. Merriam, J. A. Movshon, and D. J. Heeger, "Maps of visual space in human occipital cortex are retinotopic, not spatiotopic," *J Neurosci,* vol. 28, pp. 3988-99,

[93] D. C. Burr and M. C. Morrone, "Spatiotopic coding and remapping in humans," *Philos* 

[94] A. P. Batista, C. A. Buneo, L. H. Snyder, and R. A. Andersen, "Reach plans in eye-

[95] J. D. Golomb, M. M. Chun, and J. A. Mazer, "The native coordinate system of spatial

[96] J. D. Golomb, A. C. Marino, M. M. Chun, and J. A. Mazer, "Attention doesn't slide: spatiotopic updating after eye movements instantiates a new, discrete attentional

[97] U. Hasson, I. Levy, M. Behrmann, T. Hendler, and R. Malach, "Eccentricity bias as an organizing principle for human high-order object areas," *Neuron,* vol. 34, pp. 479-90,

in frontal and parietal cortex," *J Neurosci,* vol. 30, pp. 148-60, Jan 6 2010.

presented at the Society for Neuroscience, San Diego, CA, 2010.

*Trans R Soc Lond B Biol Sci,* vol. 366, pp. 504-15, Feb 27 2011.

centered coordinates," *Science,* vol. 285, pp. 257-60, Jul 9 1999.

locus," *Atten Percept Psychophys,* vol. 73, pp. 7-14, Jan 2011.

attention is retinotopic," *J Neurosci,* vol. 28, pp. 10654-62, Oct 15 2008.

2005.

pp. 8361-75, Aug 13 2008.

26, pp. 7962-73, Jul 26 2006.

Apr 9 2008.

Apr 25 2002.

*Neuroimage,* vol. 29, pp. 567-77, Jan 15 2006.


[80] D. Schluppeck, P. Glimcher, and D. J. Heeger, "Topographic organization for delayed saccades in human posterior parietal cortex," *J Neurophysiol,* vol. 94, pp. 1372-84, Aug 2005.

58 Visual Cortex – Current Status and Perspectives

vol. 89, pp. 3264-78, Jun 2003.

vol. 29, pp. 7031-9, May 27 2009.

*Cortex,* vol. 11, pp. 1182-1190., 2001.

cortex," *J Neurosci,* vol. 17, pp. 7060-78, Sep 15 1997.

*Trends Cogn Sci,* vol. 13, pp. 488-95, Nov 2009.

Deigo, CA, 2010.

2001.

1998.

2001.

2007.

[65] J. J. DiCarlo and J. H. Maunsell, "Anterior inferotemporal neurons of monkeys engaged in object recognition can be highly sensitive to object retinal position," *J Neurophysiol,* 

[66] A. A. Brewer and B. Barton, "'Clover Leaf' Cartography: Connectivity Among Visual Field Map Clusters," presented at the Society for Neuroscience, Washington, D.C., 2011. [67] B. Barton and A. A. Brewer, "Pinwheel cartography: A fundamental organizing principle of the human visual system," presented at the Society for Neuroscience, San

[68] A. C. Huk, R. F. Dougherty, and D. J. Heeger, "Retinotopy and functional subdivision of

[69] H. Kolster, J. B. Mandeville, J. T. Arsenault, L. B. Ekstrom, L. L. Wald, and W. Vanduffel, "Visual field map clusters in macaque extrastriate visual cortex," *J Neurosci,* 

[70] M. A. Silver and S. Kastner, "Topographic maps in human frontal and parietal cortex,"

[71] W. A. Press, A. A. Brewer, R. F. Dougherty, A. R. Wade, and B. A. Wandell, "Visual areas and spatial summation in human visual cortex," *Vision Res,* vol. 41, pp. 1321-32,

[72] A. T. Smith, M. W. Greenlee, K. D. Singh, F. M. Kraemer, and J. Hennig, "The processing of first- and second-order motion in human visual cortex assessed by functional

[73] A. T. Smith, K. D. Singh, A. L. Williams, and M. W. Greenlee, "Estimating Receptive Field Size from fMRI Data in Human Striate and Extrastriate Visual Cortex," *Cereb* 

[74] R. B. Tootell, N. Hadjikhani, E. K. Hall, S. Marrett, W. Vanduffel, J. T. Vaughan, and A. M. Dale, "The retinotopy of visual spatial attention," *Neuron,* vol. 21, pp. 1409-22, Dec

[75] R. B. Tootell, J. D. Mendola, N. K. Hadjikhani, P. J. Ledden, A. K. Liu, J. B. Reppas, M. I. Sereno, and A. M. Dale, "Functional analysis of V3A and related areas in human visual

[76] M. A. Silver, D. Ress, and D. J. Heeger, "Topographic maps of visual spatial attention in

[77] M. I. Sereno, S. Pitzalis, and A. Martinez, "Mapping of contralateral space in retinotopic coordinates by a parietal cortical area in humans," *Science,* vol. 294, pp. 1350-4, Nov 9

[78] S. Kastner, K. DeSimone, C. S. Konen, S. M. Szczepanski, K. S. Weiner, and K. A. Schneider, "Topographic maps in human frontal cortex revealed in memory-guided saccade and spatial working-memory tasks," *J Neurophysiol,* vol. 97, pp. 3494-507, May

[79] A. P. Saygin and M. I. Sereno, "Retinotopy and attention in human occipital, temporal,

human parietal cortex," *J Neurophysiol,* vol. 94, pp. 1358-71, Aug 2005.

parietal, and frontal cortex," *Cereb Cortex,* vol. 18, pp. 2158-68, Sep 2008.

magnetic resonance imaging (fMRI)," *J Neurosci,* vol. 18, pp. 3816-30, 1998.

human areas MT and MST," *J Neurosci,* vol. 22, pp. 7195-205, Aug 15 2002.


[98] H. P. Op de Beeck, J. Haushofer, and N. G. Kanwisher, "Interpreting fMRI data: maps, modules and dimensions," *Nat Rev Neurosci,* vol. 9, pp. 123-35, Feb 2008.

**Chapter 3** 

© 2012 Peyrin and Musel, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

© 2012 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution,

**Figure 1.** Coarse-to-fine sequence of spatial frequency processing (from low-to-high spatial frequencies)

and reproduction in any medium, provided the original work is properly cited.

**On the Specific Role of the Occipital** 

In recent years there has been mounting scientific excitement about the perception of scenes containing more realistic and complex stimuli than simple objects or drawings. Visual recognition of scenes is a fast, automatic and reliable process. Experimental studies have shown that complex natural scenes can be categorized in a very short time (under 150 ms [1]), suggesting a simple and efficient coding process. Many studies have attested to the importance of the Fourier components of images in scene categorization. In terms of signal representation, an image can be expressed in the Fourier domain as amplitude and phase spectra [2-5]. The amplitude spectrum highlights the dominant spatial scales (spatial frequencies) and the dominant orientations of the image, while the phase spectrum describes the relationship between spatial frequencies. It is now well established that the primary visual cortex is mainly dominated by complex cells which respond preferentially to orientations and spatial frequencies [6-8]. Simulation and psychophysical experiments have shown that information from low/medium frequencies of the amplitude spectrum is sufficient to enable scene categorization [9, 10]. These data support current influential models of scene perception [11-14].

**Cortex in Scene Perception** 

Additional information is available at the end of the chapter

Carole Peyrin and Benoit Musel

http://dx.doi.org/10.5772/48438

**1. Introduction** 

during scene perception


**Chapter 3** 

## **On the Specific Role of the Occipital Cortex in Scene Perception**

Carole Peyrin and Benoit Musel

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/48438

## **1. Introduction**

60 Visual Cortex – Current Status and Perspectives

Plenum Press, 1997, pp. 91–125.

*Neurosci,* vol. 2, pp. 265-83, 1982.

*Graph,* vol. 11, pp. 419-30, Jul-Aug 2005.

*A,* vol. 102, pp. 7350-5, May 17 2005.

*Mapp,* vol. 7, pp. 29-37, 1999.

cortex," *Trends Neurosci,* vol. 11, pp. 219-26, May 1988.

callosum in humans," *Neuroimage,* vol. 52, pp. 1224-9, Oct 1 2010.

tractography," *J Vis,* vol. 8, pp. 15 1-16, 2008.

pp. 649-77, 2004.

Jun 1996.

2008.

2001.

[98] H. P. Op de Beeck, J. Haushofer, and N. G. Kanwisher, "Interpreting fMRI data: maps,

[99] A. A. Koulakov and D. B. Chklovskii, "Orientation preference patterns in mammalian visual cortex: a wire length minimization approach," *Neuron,* vol. 29, pp. 519-27, Feb

[100] J. Kaas, "Theories of visual cortex organisation in primates," in *Cerebral Cortex, vol 12. Extrastriate cortex in primates*. vol. 12, K. J. Rockland KS, Peters A, Ed., ed New York:

[101] K. Grill-Spector and R. Malach, "The human visual cortex," *Annu Rev Neurosci,* vol. 27,

[102] D. C. Van Essen, W. T. Newsome, and J. L. Bixby, "The pattern of interhemispheric connections and its relationship to extrastriate visual areas in the macaque monkey," *J* 

[103] P. J. Basser and C. Pierpaoli, "Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI," *J Magn Reson B,* vol. 111, pp. 209-19,

[104] A. Sherbondy, D. Akers, R. Mackenzie, R. Dougherty, and B. Wandell, "Exploring connectivity of the brain's white matter with dynamic queries," *IEEE Trans Vis Comput* 

[105] A. J. Sherbondy, R. F. Dougherty, M. Ben-Shachar, S. Napel, and B. A. Wandell, "ConTrack: finding the most likely pathways between brain regions using diffusion

[106] T. E. Conturo, N. F. Lori, T. S. Cull, E. Akbudak, A. Z. Snyder, J. S. Shimony, R. C. McKinstry, H. Burton, and M. E. Raichle, "Tracking neuronal fiber pathways in the

[109] A. J. Sherbondy, R. F. Dougherty, S. Napel, and B. A. Wandell, "Identifying the human optic radiation using diffusion imaging and fiber tractography," *J Vis,* vol. 8, pp. 12 1-11,

[110] E. A. DeYoe and D. C. Van Essen, "Concurrent processing streams in monkey visual

[111] K. S. Rockland and D. N. Pandya, "Topography of occipital lobe commissural connections in the rhesus monkey," *Brain Res,* vol. 365, pp. 174-8, Feb 12 1986. [112] M. Saenz and I. Fine, "Topographic organization of V1 projections through the corpus

[113] E. Halgren, A. M. Dale, M. I. Sereno, R. B. Tootell, K. Marinkovic, and B. R. Rosen, "Location of human face-selective cortex with respect to retinotopic areas," *Hum Brain* 

living human brain," *Proc Natl Acad Sci U S A,* vol. 96, pp. 10422-7, Aug 31 1999. [107] P. J. Basser, S. Pajevic, C. Pierpaoli, J. Duda, and A. Aldroubi, "In vivo fiber tractography using DT-MRI data," *Magn Reson Med,* vol. 44, pp. 625-32, Oct 2000. [108] R. F. Dougherty, M. Ben-Shachar, R. Bammer, A. A. Brewer, and B. A. Wandell, "Functional organization of human occipital-callosal fiber tracts," *Proc Natl Acad Sci U S* 

modules and dimensions," *Nat Rev Neurosci,* vol. 9, pp. 123-35, Feb 2008.

In recent years there has been mounting scientific excitement about the perception of scenes containing more realistic and complex stimuli than simple objects or drawings. Visual recognition of scenes is a fast, automatic and reliable process. Experimental studies have shown that complex natural scenes can be categorized in a very short time (under 150 ms [1]), suggesting a simple and efficient coding process. Many studies have attested to the importance of the Fourier components of images in scene categorization. In terms of signal representation, an image can be expressed in the Fourier domain as amplitude and phase spectra [2-5]. The amplitude spectrum highlights the dominant spatial scales (spatial frequencies) and the dominant orientations of the image, while the phase spectrum describes the relationship between spatial frequencies. It is now well established that the primary visual cortex is mainly dominated by complex cells which respond preferentially to orientations and spatial frequencies [6-8]. Simulation and psychophysical experiments have shown that information from low/medium frequencies of the amplitude spectrum is sufficient to enable scene categorization [9, 10]. These data support current influential models of scene perception [11-14].

**Figure 1.** Coarse-to-fine sequence of spatial frequency processing (from low-to-high spatial frequencies) during scene perception

© 2012 Peyrin and Musel, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2012 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

On the basis of convergent data from the functional neuroanatomy of magnocellular and parvocellular visual pathways [15], neurophysiological recordings in primates [12], and psychophysical results in humans [3, 4], these models have speculated that spatial frequency content may impose a specific temporal hierarchy on the processing of visual inputs. According to these models, visual analysis starts with the parallel extraction of different elementary attributes at different spatial frequencies, in a predominantly coarse-to-fine (low-to-high spatial frequencies) sequence which favours low spatial frequencies (LSF) in the initial stages of visual processing and high spatial frequencies (HSF) in the later stages (Figure 1). The LSF in a scene, conveyed by fast magnocellular visual channels, might therefore activate visual pathways and subsequently reach high-order areas in the dorsal stream (parietal and frontal) more rapidly than HSF, allowing an initial perceptual parsing of visual inputs prior to their complete propagation along the ventral stream (inferotemporal cortex) which ultimately mediates object recognition. This initial low-pass visual analysis might serve to refine the subsequent processing of HSF, which are conveyed more slowly by parvocellular visual channels to the ventral stream.

On the Specific Role of the Occipital Cortex in Scene Perception 63

analysis follows a predominantly coarse-to-fine processing sequence (LSF are extracted first, followed by HSF). The first experimental evidence in support of this type of coarse-to-fine processing sequence in human vision comes from psychophysical studies using hierarchical stimuli (global forms composed of several local elements [22, 23]). Usually, these forms represent a large global letter form made up of small local letters. The subject's task is to identify a target letter either at global level, at local level, or at both levels. Using this paradigm, two main findings have emerged. Firstly, global form identification is faster than local form identification. This phenomenon is known as the global precedence effect. Secondly, while inconsistent global information slows down local information identification, the identification of local information has no effect on global identification. This asymmetrical effect is known as the global interference effect. However, these effects decrease or even vanish when the hierarchical forms are high-pass filtered (i.e. LSF are cut off). They are also affected by a subject's adaptation to a given frequency band (low vs. high), suggesting that LSF carry global information, whereas HSF carry local information [24-26]. Based on the assumption that global information is preferentially conveyed by LSF, and that local information is conveyed by HSF, the global-to-local processing sequence has been interpreted as reflecting a fundamental principle of the coarse-to-fine processing

Additional evidence of a coarse-to-fine processing sequence was provided by psychophysical studies using more ecological stimuli, such as natural scenes and faces [14, 27-30]. Schyns and Oliva [14], for example, used hybrid stimuli consisting of two superimposed images of natural scenes, taken from different semantic categories and containing different spatial frequencies (e.g., a highway scene in LSF superimposed on a city scene in HSF). The perception of these hybrid scenes was dominated by LSF information when presentation time was very brief (30 ms). However when presentation time was longer (150 ms), perception was dominated by HSF information, suggesting precedence of LSF over HSF in the visual processing time-course. Furthermore, when two successive hybrids displayed a coarse-to-fine sequence for a given scene (e.g., the highway scene in LSF in the first hybrid and then in HSF in the second hybrid) and a fine-to-coarse sequence simultaneously for another scene (e.g., the city scene in HSF in the first hybrid and then in LSF in the second hybrid), scene perception was more frequently based on a coarse-to-fine

Quite exactly how and where in the brain LSF and HSF information is differentially analyzed and eventually merged during visual processing remains an unresolved question. Traditional models have generally assumed that different visual cues are combined at successive stages along the cortical hierarchy [31, 32], and suggest that LSF and HSF might converge only in higher-level visual areas in the inferior temporal cortex (such as the fusiform or parahippocampal cortex [33, 34]). However, drawing on evidence from neurophysiological recordings in nonhuman primates [35], Bullier [12] proposed that rapid LSF analysis, predominantly carried out in the dorsal visual stream, might be "retroinjected" through feedback signals into low-level areas (e.g., primary visual cortex) where it would influence subsequent HSF analysis, and guide further processing through the ventral

sequence.

rather than a fine-to-coarse sequence.

The majority of current visual models are, therefore, based on the neurophysiological properties of spatial frequency processing. However exactly how spatial frequencies are processed within the occipital cortex remains unclear. Initially, on the basis of neurophysiological recordings in nonhuman primates, Bullier [12] proposed that rapid LSF analysis, which takes place predominantly in the dorsal visual stream, might be "retroinjected" through feedback signals into low-level areas (e.g., primary visual cortex, V1), where it would influence subsequent HSF analysis and guide further processing through the ventral visual stream. The occipital cortex might therefore serve as an "active blackboard" integrating computations carried out by higher order cortical areas. Secondly, the issues of cerebral asymmetries and/or retinotopic organization of spatial frequency processing within the occipital cortex are still being debated in the literature. Many studies reveal retinotopic organization of spatial frequency processing in the occipital cortex [16-18]. HSF sinusoidal grating processing, for example, activates the foveal representation in all retinotopic areas (such as V1) of the occipital cortex, and LSF sinusoidal grating processing activates more peripheral representations in the same cortical areas. Despite studies showing retinotopic mapping of spatial frequency processing in the occipital cortex, many experimental arguments assume a certain hemispheric specialization of spatial frequency processing in the occipital cortex. The right occipital cortex appears to be preferentially specialized for LSF information processing, while the left occipital cortex seems to be preferentially specialized for HSF information processing [19-21]. It is, therefore, essential to determine whether hemispheric specialization and retinotopic processing can co-occur in the occipital cortex. While addressing this issue, the present chapter also aims to clarify the different attributes of the occipital cortex during scene recognition.

## **2. Coarse-to-fine analysis in visual perception**

Results from various neurophysiological, computational, and behavioral studies all indicate that the totality of visual information is not immediately conveyed, but that information analysis follows a predominantly coarse-to-fine processing sequence (LSF are extracted first, followed by HSF). The first experimental evidence in support of this type of coarse-to-fine processing sequence in human vision comes from psychophysical studies using hierarchical stimuli (global forms composed of several local elements [22, 23]). Usually, these forms represent a large global letter form made up of small local letters. The subject's task is to identify a target letter either at global level, at local level, or at both levels. Using this paradigm, two main findings have emerged. Firstly, global form identification is faster than local form identification. This phenomenon is known as the global precedence effect. Secondly, while inconsistent global information slows down local information identification, the identification of local information has no effect on global identification. This asymmetrical effect is known as the global interference effect. However, these effects decrease or even vanish when the hierarchical forms are high-pass filtered (i.e. LSF are cut off). They are also affected by a subject's adaptation to a given frequency band (low vs. high), suggesting that LSF carry global information, whereas HSF carry local information [24-26]. Based on the assumption that global information is preferentially conveyed by LSF, and that local information is conveyed by HSF, the global-to-local processing sequence has been interpreted as reflecting a fundamental principle of the coarse-to-fine processing sequence.

62 Visual Cortex – Current Status and Perspectives

slowly by parvocellular visual channels to the ventral stream.

of the occipital cortex during scene recognition.

**2. Coarse-to-fine analysis in visual perception** 

On the basis of convergent data from the functional neuroanatomy of magnocellular and parvocellular visual pathways [15], neurophysiological recordings in primates [12], and psychophysical results in humans [3, 4], these models have speculated that spatial frequency content may impose a specific temporal hierarchy on the processing of visual inputs. According to these models, visual analysis starts with the parallel extraction of different elementary attributes at different spatial frequencies, in a predominantly coarse-to-fine (low-to-high spatial frequencies) sequence which favours low spatial frequencies (LSF) in the initial stages of visual processing and high spatial frequencies (HSF) in the later stages (Figure 1). The LSF in a scene, conveyed by fast magnocellular visual channels, might therefore activate visual pathways and subsequently reach high-order areas in the dorsal stream (parietal and frontal) more rapidly than HSF, allowing an initial perceptual parsing of visual inputs prior to their complete propagation along the ventral stream (inferotemporal cortex) which ultimately mediates object recognition. This initial low-pass visual analysis might serve to refine the subsequent processing of HSF, which are conveyed more

The majority of current visual models are, therefore, based on the neurophysiological properties of spatial frequency processing. However exactly how spatial frequencies are processed within the occipital cortex remains unclear. Initially, on the basis of neurophysiological recordings in nonhuman primates, Bullier [12] proposed that rapid LSF analysis, which takes place predominantly in the dorsal visual stream, might be "retroinjected" through feedback signals into low-level areas (e.g., primary visual cortex, V1), where it would influence subsequent HSF analysis and guide further processing through the ventral visual stream. The occipital cortex might therefore serve as an "active blackboard" integrating computations carried out by higher order cortical areas. Secondly, the issues of cerebral asymmetries and/or retinotopic organization of spatial frequency processing within the occipital cortex are still being debated in the literature. Many studies reveal retinotopic organization of spatial frequency processing in the occipital cortex [16-18]. HSF sinusoidal grating processing, for example, activates the foveal representation in all retinotopic areas (such as V1) of the occipital cortex, and LSF sinusoidal grating processing activates more peripheral representations in the same cortical areas. Despite studies showing retinotopic mapping of spatial frequency processing in the occipital cortex, many experimental arguments assume a certain hemispheric specialization of spatial frequency processing in the occipital cortex. The right occipital cortex appears to be preferentially specialized for LSF information processing, while the left occipital cortex seems to be preferentially specialized for HSF information processing [19-21]. It is, therefore, essential to determine whether hemispheric specialization and retinotopic processing can co-occur in the occipital cortex. While addressing this issue, the present chapter also aims to clarify the different attributes

Results from various neurophysiological, computational, and behavioral studies all indicate that the totality of visual information is not immediately conveyed, but that information Additional evidence of a coarse-to-fine processing sequence was provided by psychophysical studies using more ecological stimuli, such as natural scenes and faces [14, 27-30]. Schyns and Oliva [14], for example, used hybrid stimuli consisting of two superimposed images of natural scenes, taken from different semantic categories and containing different spatial frequencies (e.g., a highway scene in LSF superimposed on a city scene in HSF). The perception of these hybrid scenes was dominated by LSF information when presentation time was very brief (30 ms). However when presentation time was longer (150 ms), perception was dominated by HSF information, suggesting precedence of LSF over HSF in the visual processing time-course. Furthermore, when two successive hybrids displayed a coarse-to-fine sequence for a given scene (e.g., the highway scene in LSF in the first hybrid and then in HSF in the second hybrid) and a fine-to-coarse sequence simultaneously for another scene (e.g., the city scene in HSF in the first hybrid and then in LSF in the second hybrid), scene perception was more frequently based on a coarse-to-fine rather than a fine-to-coarse sequence.

Quite exactly how and where in the brain LSF and HSF information is differentially analyzed and eventually merged during visual processing remains an unresolved question. Traditional models have generally assumed that different visual cues are combined at successive stages along the cortical hierarchy [31, 32], and suggest that LSF and HSF might converge only in higher-level visual areas in the inferior temporal cortex (such as the fusiform or parahippocampal cortex [33, 34]). However, drawing on evidence from neurophysiological recordings in nonhuman primates [35], Bullier [12] proposed that rapid LSF analysis, predominantly carried out in the dorsal visual stream, might be "retroinjected" through feedback signals into low-level areas (e.g., primary visual cortex) where it would influence subsequent HSF analysis, and guide further processing through the ventral visual stream. The occipital cortex might therefore serve as an "active blackboard" integrating computations made by higher-order cortical areas. However, to date, the neural architecture and temporal dynamics of such top–down mechanisms have never been systematically investigated via direct testing of the preferential coarse-to-fine processing sequence during visual scene perception in humans.

On the Specific Role of the Occipital Cortex in Scene Perception 65

visual cortex. However, many studies have highlighted the fact that in humans, the left and right hemispheres do not deal with all aspects of visual information processing with equal ability. The two hemispheres might in fact make complementary contributions to the

**4. Hemispheric specialization of spatial frequency processing during** 

hemispheric specialization in global and local processing.

Many experimental arguments assume that spatial frequency processing is shared between the two hemispheres, with right hemispheric predominance for LSF processing and left hemispheric predominance for HSF processing. This hemispheric specialization has been observed either through behavioral studies on healthy subjects [21, 37-48] and neurological patients [49-51], or through functional neuroimaging studies [19, 20, 52-65]. However, the hemispheric asymmetries in question were largely inferred from studies assessing

Using hierarchical visual stimuli consisting of a global form made up of several local elements [22, 23] displayed either in the left or right visual field, Sergent [46] demonstrated that global forms were identified more quickly when they were presented in the left visual hemifield, projecting directly to the right hemisphere, while recognition of local forms was faster when they were presented in the right visual hemifield, projecting directly to the left hemisphere [52] (see also [38]). These results suggest right hemispheric dominance for the recognition of global forms, and left hemispheric dominance for the recognition of local forms. Since global processing can be considered to be mediated by low-pass spatial analysis, and local processing by high-pass spatial analysis [24-26], the hemispheric specialization patterns observed in global and local processing have been interpreted as

Unfortunately, the relationship between local and global information and spatial frequencies in hierarchical forms is far from univocal [66]. Global information could, for example, be conveyed not only by LSF but also by HSF. The hypothesis of hemispheric asymmetry in spatial frequency processing was subsequently directly tested by making explicit changes in the spatial frequency spectrum of stimuli, using sinusoidal gratings as stimuli [40-42], or more complex visual stimuli such as images of natural scenes [21, 43, 44]. It should be noted that hierarchical forms do not allow this type of manipulation, because low-pass filtering cancels out the local form, thus rendering execution of the task impossible. In a series of behavioral studies conducted by our team [21, 43, 44], we evaluated hemispheric asymmetry in healthy subjects using natural scenes, while manipulating the spatial frequency components of the scenes, which were presented in divided visual fields. In the princeps study [43], participants were asked to recognize either an LSF or an HSF filtered target scene (a city or a highway), displayed in either the left or the right visual field. Results showed that LSF filtered scenes were recognized more quickly when they were presented in the left

reflecting the hemispheric specialization of LSF and HSF, respectively [46].

processing of LSF and HSF.

**4.1. Behavioral arguments** 

**scene perception** 

## **3. Neural basis of the retro-injection mechanism during scene perception**

In order to test the coarse-to-fine processing sequence and to identify its neural substrates in the human brain, we presented in rapid succession sequences of two spatial frequencyfiltered scenes, with either an LSF image followed by an HSF image (coarse-to-fine sequence), or an HSF image followed by an LSF image (fine-to-coarse sequence) during fMRI and ERP recordings on the same participants [36]. Each scene in a sequence belonged to one of three categories (city, beach, or indoor). Half of the sequences displayed two scenes from the same category, and the other half displayed two scenes from different categories. Participants had to judge whether the two successive scenes belonged to the same category. This study also addressed the crucial issue of delayed "retro-injection" and spatial frequency integration in the occipital cortex.

Examination by fMRI showed selective increases in coarse-to-fine sequences (relative to fineto-coarse sequences) in early-stage occipital areas, as well as in frontal and temporo-parietal areas. ERP topography and source analyses highlighted a similar network of cortical areas, but were in addition able to determine the time-course of activation in these regions, involving either LSF or HSF images in the different sequences. Higher-order areas in frontal and temporo-parietal regions responded more to LSF stimuli when these were presented first, whereas the occipital visual cortex responded more to HSF presented after LSF. More specifically, our results demonstrate that LSF in scenes (conveyed by fast magnocellular channels) can rapidly activate high-order areas, providing spatial (via the frontal eye fields) and semantic information (via the left prefrontal cortex and temporal areas), as well as attentional signals (via the temporo-parietal junction), all of which may promote the ongoing perceptual organization and categorization of visual input. This first coarse analysis may possibly be refined by further processing in the visual cortices of HSF in scenes (conveyed more slowly by the parvocellular channels). In order for this to occur, feedback from the first low-pass computations carried out in frontal and temporo-parietal areas might be "retro-injected" into lower level areas, such as the occipital cortex, at the level of the primary visual cortex, with a view to guiding the high-pass analysis and selecting the relevant finer details necessary for the recognition and categorization of scenes. These results provide critical support for recent models of vision, and for the retro-injection mechanism proposed by Bullier [12]. They also highlight the necessity for further investigation of the neural mechanisms of spatial frequency processing in the occipital cortex.

The majority of visual models assume, therefore, a predominantly coarse-to-fine sequence of spatial frequency processing in the whole brain, based on the functional properties of the visual cortex. However, many studies have highlighted the fact that in humans, the left and right hemispheres do not deal with all aspects of visual information processing with equal ability. The two hemispheres might in fact make complementary contributions to the processing of LSF and HSF.

## **4. Hemispheric specialization of spatial frequency processing during scene perception**

Many experimental arguments assume that spatial frequency processing is shared between the two hemispheres, with right hemispheric predominance for LSF processing and left hemispheric predominance for HSF processing. This hemispheric specialization has been observed either through behavioral studies on healthy subjects [21, 37-48] and neurological patients [49-51], or through functional neuroimaging studies [19, 20, 52-65]. However, the hemispheric asymmetries in question were largely inferred from studies assessing hemispheric specialization in global and local processing.

## **4.1. Behavioral arguments**

64 Visual Cortex – Current Status and Perspectives

sequence during visual scene perception in humans.

frequency integration in the occipital cortex.

cortex.

visual stream. The occipital cortex might therefore serve as an "active blackboard" integrating computations made by higher-order cortical areas. However, to date, the neural architecture and temporal dynamics of such top–down mechanisms have never been systematically investigated via direct testing of the preferential coarse-to-fine processing

**3. Neural basis of the retro-injection mechanism during scene perception** 

In order to test the coarse-to-fine processing sequence and to identify its neural substrates in the human brain, we presented in rapid succession sequences of two spatial frequencyfiltered scenes, with either an LSF image followed by an HSF image (coarse-to-fine sequence), or an HSF image followed by an LSF image (fine-to-coarse sequence) during fMRI and ERP recordings on the same participants [36]. Each scene in a sequence belonged to one of three categories (city, beach, or indoor). Half of the sequences displayed two scenes from the same category, and the other half displayed two scenes from different categories. Participants had to judge whether the two successive scenes belonged to the same category. This study also addressed the crucial issue of delayed "retro-injection" and spatial

Examination by fMRI showed selective increases in coarse-to-fine sequences (relative to fineto-coarse sequences) in early-stage occipital areas, as well as in frontal and temporo-parietal areas. ERP topography and source analyses highlighted a similar network of cortical areas, but were in addition able to determine the time-course of activation in these regions, involving either LSF or HSF images in the different sequences. Higher-order areas in frontal and temporo-parietal regions responded more to LSF stimuli when these were presented first, whereas the occipital visual cortex responded more to HSF presented after LSF. More specifically, our results demonstrate that LSF in scenes (conveyed by fast magnocellular channels) can rapidly activate high-order areas, providing spatial (via the frontal eye fields) and semantic information (via the left prefrontal cortex and temporal areas), as well as attentional signals (via the temporo-parietal junction), all of which may promote the ongoing perceptual organization and categorization of visual input. This first coarse analysis may possibly be refined by further processing in the visual cortices of HSF in scenes (conveyed more slowly by the parvocellular channels). In order for this to occur, feedback from the first low-pass computations carried out in frontal and temporo-parietal areas might be "retro-injected" into lower level areas, such as the occipital cortex, at the level of the primary visual cortex, with a view to guiding the high-pass analysis and selecting the relevant finer details necessary for the recognition and categorization of scenes. These results provide critical support for recent models of vision, and for the retro-injection mechanism proposed by Bullier [12]. They also highlight the necessity for further investigation of the neural mechanisms of spatial frequency processing in the occipital

The majority of visual models assume, therefore, a predominantly coarse-to-fine sequence of spatial frequency processing in the whole brain, based on the functional properties of the Using hierarchical visual stimuli consisting of a global form made up of several local elements [22, 23] displayed either in the left or right visual field, Sergent [46] demonstrated that global forms were identified more quickly when they were presented in the left visual hemifield, projecting directly to the right hemisphere, while recognition of local forms was faster when they were presented in the right visual hemifield, projecting directly to the left hemisphere [52] (see also [38]). These results suggest right hemispheric dominance for the recognition of global forms, and left hemispheric dominance for the recognition of local forms. Since global processing can be considered to be mediated by low-pass spatial analysis, and local processing by high-pass spatial analysis [24-26], the hemispheric specialization patterns observed in global and local processing have been interpreted as reflecting the hemispheric specialization of LSF and HSF, respectively [46].

Unfortunately, the relationship between local and global information and spatial frequencies in hierarchical forms is far from univocal [66]. Global information could, for example, be conveyed not only by LSF but also by HSF. The hypothesis of hemispheric asymmetry in spatial frequency processing was subsequently directly tested by making explicit changes in the spatial frequency spectrum of stimuli, using sinusoidal gratings as stimuli [40-42], or more complex visual stimuli such as images of natural scenes [21, 43, 44]. It should be noted that hierarchical forms do not allow this type of manipulation, because low-pass filtering cancels out the local form, thus rendering execution of the task impossible. In a series of behavioral studies conducted by our team [21, 43, 44], we evaluated hemispheric asymmetry in healthy subjects using natural scenes, while manipulating the spatial frequency components of the scenes, which were presented in divided visual fields. In the princeps study [43], participants were asked to recognize either an LSF or an HSF filtered target scene (a city or a highway), displayed in either the left or the right visual field. Results showed that LSF filtered scenes were recognized more quickly when they were presented in the left visual hemifield, projecting directly to the right hemisphere, while recognition of the HSF filtered scenes was faster when these were presented in the right visual hemifield, projecting directly to the left hemisphere. This study clearly demonstrated right hemispheric superiority for LSF and left hemispheric superiority for HSF processing, and therefore supports Sergent's assumption [46] that visual tasks which require the processing of LSF information (such as global letter identification in hierarchical forms) would result in a left visual field/right hemisphere advantage, whereas tasks requiring the processing of HSF information (such as local letter identification in hierarchical forms) would result in right visual field/left hemisphere superiority.

On the Specific Role of the Occipital Cortex in Scene Perception 67

demonstrate cerebral asymmetries in first-stage visual areas. Instead, their results, based on event-related potentials (ERPs), showed long latency asymmetries (260-360 latency range) for global versus local processing, suggesting that hemispheric specialization was present

Furthermore, some functional imaging data have revealed an attentional cortical mechanism which exerts control over the perceptual processes involved in global and local processing [50, 51, 53-55, 63-65]. This mechanism operates on the attentional selection of information presented either at global level, at local level, or at both levels depending on task constraints. This mechanism is located in the temporo-partial junction. Using Using ERPs, Yamaguchi and collaborators [65] investigated the neural substrates of attentional allocation to global and local components of a hierarchical form. For this purpose ERPs were recorded while participants shifted their attention to the global or local level of a hierarchical form. Shift direction was controlled by a preceding cue stimulus. Hemispheric asymmetries arose not only during the task in which global–local processing was actually being performed, but also in the time interval during which attention was directed towards global or local levels by the cues. Therefore, in addition to hemispheric asymmetry during "bottom-up" processing, this study demonstrated the existence of neural substrates for a top-down mechanism of hemispheric asymmetry in global and local selection. ERPs to the cue showed greater amplitude in the right hemisphere during attentional allocation at global level, and greater amplitude in the left hemisphere during attentional allocation at local level. The neural activity in question was located in the right temporo-parietal junction for the global shift, and in the left temporo-parietal junction for the local shift. These electrophysiological results provided an asymmetrical neural basis for the "top-down" allocation of attention to global and local features, and revealed the contribution of the temporal-parietal cortex to

On the whole, the imaging studies mentioned previously have provided conflicting results on hemispheric specialization using hierarchical stimuli. By directly manipulating the spatial frequency content of stimuli, subsequent studies revealed hemispheric specialization in certain occipito-temporal areas [19, 20]. In one fMRI study [20], we investigated the neural correlates and the hemispheric specialization of spatial frequency processing during the perception of scene stimuli which allowed an explicit change in the spatial frequency spectrum. For this purpose, we used a categorization task of small LSF and HSF scene

By comparing LSF to HSF scene categorization (Figure 2a), we observed significant activation in the right anterior temporal cortex and the right parahippocampal gyrus. As these regions are known to be involved in scene processing, these results suggest that scene recognition is based mainly on LSF extraction and analysis, following a coarse-to-fine processing sequence. Significant activation was also obtained in the right inferior parietal lobule, and this probably reflects attentional modulation during spatial frequency selection. Compared to HSF scene recognition, LSF scene recognition also activated the bilateral posterior part of the superior temporal cortex. This result contradicts neuropsychological studies [49-51], which have shown specialization of the right superior temporal cortex in the

only in the higher levels of visual analysis.

this attentional mechanism.

images (at a visual angle of 4°).

Following this, we examined whether the temporal characteristics of spatial frequency analysis (i.e. the temporal precedence of LSF over HSF as postulated by the coarse-to-fine processing sequence) might interfere with hemispheric specialization. We did this by manipulating the exposure duration of filtered natural scene images (30 vs. 150 ms [44]). Results showed the classical hemispheric specialization pattern for brief exposure duration (the right hemisphere was predominantly involved in LSF scene processing and the left in HSF scene processing), and a tendency towards right hemisphere advantage, irrespective of the spatial frequency content for longer exposure durations. These results suggest that the hemispheric specialization pattern for visual information processing ought to be considered as a dynamic system, within which the superiority of one hemisphere over the other could change according to the level of temporal constraints. The higher the temporal constraints of the task, the more the hemispheres become specialized in spatial frequency processing. In a subsequent study [21], we provided evidence for hemispheric specialization in spatial frequency processing in men, but not in women. These results are consistent with studies showing that the functional cerebral organization of women is less lateralized than that of men [67, 68].

### **4.2. Neural correlates of hemispheric specialization**

Neuropsychological and neuroimaging studies conducted on hierarchical visual stimuli have reported conflicting results on which cortical structures present hemispheric specialization. Robertson and collaborators [51] showed, for example, that unilateral damage of the temporo–parietal junction could impair patients' performance in the hierarchical form paradigm. Patients with a lesion in the left superior temporal gyrus thus exhibited a slowing down in local form identification, whereas the performance of patients with a lesion situated in the right temporo–parietal region was impaired during global form identification. These data suggest that two independent perceptual sub-systems may be involved; the right temporo–parietal junction which emphasizes global information, and the left temporo–parietal junction which emphasizes local information. However, using positron emission tomography, Fink and collaborators [53-55] reported cerebral asymmetries in the occipital cortex. The right lingual gyrus was more highly activated during the processing of global as opposed to local forms, while the left inferior occipital gyrus was more highly activated during the processing of local rather than global forms. Using electroencephalographic recordings, Heinze and collaborators [57, 60] failed to demonstrate cerebral asymmetries in first-stage visual areas. Instead, their results, based on event-related potentials (ERPs), showed long latency asymmetries (260-360 latency range) for global versus local processing, suggesting that hemispheric specialization was present only in the higher levels of visual analysis.

66 Visual Cortex – Current Status and Perspectives

visual field/left hemisphere superiority.

**4.2. Neural correlates of hemispheric specialization** 

men [67, 68].

visual hemifield, projecting directly to the right hemisphere, while recognition of the HSF filtered scenes was faster when these were presented in the right visual hemifield, projecting directly to the left hemisphere. This study clearly demonstrated right hemispheric superiority for LSF and left hemispheric superiority for HSF processing, and therefore supports Sergent's assumption [46] that visual tasks which require the processing of LSF information (such as global letter identification in hierarchical forms) would result in a left visual field/right hemisphere advantage, whereas tasks requiring the processing of HSF information (such as local letter identification in hierarchical forms) would result in right

Following this, we examined whether the temporal characteristics of spatial frequency analysis (i.e. the temporal precedence of LSF over HSF as postulated by the coarse-to-fine processing sequence) might interfere with hemispheric specialization. We did this by manipulating the exposure duration of filtered natural scene images (30 vs. 150 ms [44]). Results showed the classical hemispheric specialization pattern for brief exposure duration (the right hemisphere was predominantly involved in LSF scene processing and the left in HSF scene processing), and a tendency towards right hemisphere advantage, irrespective of the spatial frequency content for longer exposure durations. These results suggest that the hemispheric specialization pattern for visual information processing ought to be considered as a dynamic system, within which the superiority of one hemisphere over the other could change according to the level of temporal constraints. The higher the temporal constraints of the task, the more the hemispheres become specialized in spatial frequency processing. In a subsequent study [21], we provided evidence for hemispheric specialization in spatial frequency processing in men, but not in women. These results are consistent with studies showing that the functional cerebral organization of women is less lateralized than that of

Neuropsychological and neuroimaging studies conducted on hierarchical visual stimuli have reported conflicting results on which cortical structures present hemispheric specialization. Robertson and collaborators [51] showed, for example, that unilateral damage of the temporo–parietal junction could impair patients' performance in the hierarchical form paradigm. Patients with a lesion in the left superior temporal gyrus thus exhibited a slowing down in local form identification, whereas the performance of patients with a lesion situated in the right temporo–parietal region was impaired during global form identification. These data suggest that two independent perceptual sub-systems may be involved; the right temporo–parietal junction which emphasizes global information, and the left temporo–parietal junction which emphasizes local information. However, using positron emission tomography, Fink and collaborators [53-55] reported cerebral asymmetries in the occipital cortex. The right lingual gyrus was more highly activated during the processing of global as opposed to local forms, while the left inferior occipital gyrus was more highly activated during the processing of local rather than global forms. Using electroencephalographic recordings, Heinze and collaborators [57, 60] failed to Furthermore, some functional imaging data have revealed an attentional cortical mechanism which exerts control over the perceptual processes involved in global and local processing [50, 51, 53-55, 63-65]. This mechanism operates on the attentional selection of information presented either at global level, at local level, or at both levels depending on task constraints. This mechanism is located in the temporo-partial junction. Using Using ERPs, Yamaguchi and collaborators [65] investigated the neural substrates of attentional allocation to global and local components of a hierarchical form. For this purpose ERPs were recorded while participants shifted their attention to the global or local level of a hierarchical form. Shift direction was controlled by a preceding cue stimulus. Hemispheric asymmetries arose not only during the task in which global–local processing was actually being performed, but also in the time interval during which attention was directed towards global or local levels by the cues. Therefore, in addition to hemispheric asymmetry during "bottom-up" processing, this study demonstrated the existence of neural substrates for a top-down mechanism of hemispheric asymmetry in global and local selection. ERPs to the cue showed greater amplitude in the right hemisphere during attentional allocation at global level, and greater amplitude in the left hemisphere during attentional allocation at local level. The neural activity in question was located in the right temporo-parietal junction for the global shift, and in the left temporo-parietal junction for the local shift. These electrophysiological results provided an asymmetrical neural basis for the "top-down" allocation of attention to global and local features, and revealed the contribution of the temporal-parietal cortex to this attentional mechanism.

On the whole, the imaging studies mentioned previously have provided conflicting results on hemispheric specialization using hierarchical stimuli. By directly manipulating the spatial frequency content of stimuli, subsequent studies revealed hemispheric specialization in certain occipito-temporal areas [19, 20]. In one fMRI study [20], we investigated the neural correlates and the hemispheric specialization of spatial frequency processing during the perception of scene stimuli which allowed an explicit change in the spatial frequency spectrum. For this purpose, we used a categorization task of small LSF and HSF scene images (at a visual angle of 4°).

By comparing LSF to HSF scene categorization (Figure 2a), we observed significant activation in the right anterior temporal cortex and the right parahippocampal gyrus. As these regions are known to be involved in scene processing, these results suggest that scene recognition is based mainly on LSF extraction and analysis, following a coarse-to-fine processing sequence. Significant activation was also obtained in the right inferior parietal lobule, and this probably reflects attentional modulation during spatial frequency selection. Compared to HSF scene recognition, LSF scene recognition also activated the bilateral posterior part of the superior temporal cortex. This result contradicts neuropsychological studies [49-51], which have shown specialization of the right superior temporal cortex in the

perceptual processing of global (LSF) information, and specialization of the left superior temporal cortex in the perceptual processing of local (HSF) information. Finally, comparisons between HSF and LSF scene categorization failed to show any significant activation, suggesting a bias towards the processing of LSF information.

On the Specific Role of the Occipital Cortex in Scene Perception 69

asymmetries, since it allows the cancelling out of any main effect deriving from spatial

**Figure 3.** The method of direct inter-hemispheric comparison. In this method, two sets of functional volumes, obtained from functional scans, are compared at both individual and group level. One set is represented by functional volumes in accordance with neurological convention (the left hemisphere - LH appears on the left side of images) and the other set is represented by the same functional volumes this time in accordance with radiological convention (the right hemisphere – RH appears on the left side of images). Images from the second set were "flipped" by 180° in the midsagital plane, thus providing "mirror" images of the first set. Contrasts between "unflipped" and "left-right flipped" images were then calculated for each of the spatial frequency components of natural scenes. In order to assess hemispheric predominance during the perception of LSF scenes, for instance, the following contrast was calculated: LSF unflip > LSF flip. Regions which were statistically more highly activated in the left hemisphere than in the right hemisphere appear on the left side, and regions which were statistically more highly activated in the right hemisphere than in the left hemisphere appear on the right side.

Using a neuropsychological approach [21], we further investigated the role of the right occipital cortex in LSF processing in a female neurological patient with a focal lesion in this region following the embolization of an arterioveinous malformation. As a result, she suffered from a left homonymous hemianopia. The study was conducted 1 week before and 6 months after the surgical intervention. As expected, after the embolization, LSF scene recognition was more severely impaired than HSF scene recognition. These data support the hypothesis of a preferential specialization of the right occipital cortex for LSF information processing, and suggest more generally a hemispheric specialization in spatial frequency processing in females, although this is difficult to demonstrate behaviourally in healthy

What is important here is that although LSF information may be perceptually available before HSF, this does not necessarily imply that it is always used first to support visual

**4.3. Cerebral asymmetries during coarse-to-fine analysis of scenes** 

women.

frequency bias (i.e. stronger global cerebral activation for LSF than for HSF scenes).

**Figure 2.** Hemispheric specialization of spatial frequency processing during scene perception

Based on behavioural experiments in which direct comparisons were made between the performances of the two visual fields [21, 40-44, 46], we suggested that any assessment of visual cerebral asymmetries must make direct comparisons between activation in the two hemispheres. In order to do so, we created a new method of fMRI data analysis. The method of direct inter-hemispheric comparison examines contrasts between ''unflipped'' and ''left– right flipped'' functional images from the same experimental condition (Figure 3), in order to compare activity in one hemisphere with activity in homologous regions of the other hemisphere [19, 20, 69, 70]. Using this method, we demonstrated higher levels of activation in the right middle occipital gyrus than in the left during the recognition of LSF scenes, and higher levels of activation in the left middle occipital gyrus than in the right during the recognition of HSF scenes (Figure 2b).

This result provides supplementary evidence for hemispheric specialization in the early stages of visual analysis when spatial frequencies are being processed. Another important point was that when analysing the fMRI data using a more traditional approach which contrasts spatial frequencies to one another, we observed stronger cerebral activation for LSF than for HSF scenes, while the reverse contrast did not reveal any significant activation. Results therefore differ according to the method of data analysis applied. A direct interhemispheric method of comparison seems more appropriate for the assessment of cerebral asymmetries, since it allows the cancelling out of any main effect deriving from spatial frequency bias (i.e. stronger global cerebral activation for LSF than for HSF scenes).

68 Visual Cortex – Current Status and Perspectives

recognition of HSF scenes (Figure 2b).

perceptual processing of global (LSF) information, and specialization of the left superior temporal cortex in the perceptual processing of local (HSF) information. Finally, comparisons between HSF and LSF scene categorization failed to show any significant

activation, suggesting a bias towards the processing of LSF information.

**Figure 2.** Hemispheric specialization of spatial frequency processing during scene perception

Based on behavioural experiments in which direct comparisons were made between the performances of the two visual fields [21, 40-44, 46], we suggested that any assessment of visual cerebral asymmetries must make direct comparisons between activation in the two hemispheres. In order to do so, we created a new method of fMRI data analysis. The method of direct inter-hemispheric comparison examines contrasts between ''unflipped'' and ''left– right flipped'' functional images from the same experimental condition (Figure 3), in order to compare activity in one hemisphere with activity in homologous regions of the other hemisphere [19, 20, 69, 70]. Using this method, we demonstrated higher levels of activation in the right middle occipital gyrus than in the left during the recognition of LSF scenes, and higher levels of activation in the left middle occipital gyrus than in the right during the

This result provides supplementary evidence for hemispheric specialization in the early stages of visual analysis when spatial frequencies are being processed. Another important point was that when analysing the fMRI data using a more traditional approach which contrasts spatial frequencies to one another, we observed stronger cerebral activation for LSF than for HSF scenes, while the reverse contrast did not reveal any significant activation. Results therefore differ according to the method of data analysis applied. A direct interhemispheric method of comparison seems more appropriate for the assessment of cerebral

**Figure 3.** The method of direct inter-hemispheric comparison. In this method, two sets of functional volumes, obtained from functional scans, are compared at both individual and group level. One set is represented by functional volumes in accordance with neurological convention (the left hemisphere - LH appears on the left side of images) and the other set is represented by the same functional volumes this time in accordance with radiological convention (the right hemisphere – RH appears on the left side of images). Images from the second set were "flipped" by 180° in the midsagital plane, thus providing "mirror" images of the first set. Contrasts between "unflipped" and "left-right flipped" images were then calculated for each of the spatial frequency components of natural scenes. In order to assess hemispheric predominance during the perception of LSF scenes, for instance, the following contrast was calculated: LSF unflip > LSF flip. Regions which were statistically more highly activated in the left hemisphere than in the right hemisphere appear on the left side, and regions which were statistically more highly activated in the right hemisphere than in the left hemisphere appear on the right side.

Using a neuropsychological approach [21], we further investigated the role of the right occipital cortex in LSF processing in a female neurological patient with a focal lesion in this region following the embolization of an arterioveinous malformation. As a result, she suffered from a left homonymous hemianopia. The study was conducted 1 week before and 6 months after the surgical intervention. As expected, after the embolization, LSF scene recognition was more severely impaired than HSF scene recognition. These data support the hypothesis of a preferential specialization of the right occipital cortex for LSF information processing, and suggest more generally a hemispheric specialization in spatial frequency processing in females, although this is difficult to demonstrate behaviourally in healthy women.

## **4.3. Cerebral asymmetries during coarse-to-fine analysis of scenes**

What is important here is that although LSF information may be perceptually available before HSF, this does not necessarily imply that it is always used first to support visual recognition in all tasks. Indeed, the global precedence effect can be turned into a local precedence effect by simple experimental manipulation (e.g., by changing the visual angle [22, 71] or the number of local elements [72]). In Schyns and Oliva's experiments [14], a substantial proportion (29%) of hybrid sequences were in fact categorized in accordance with a fine-to-coarse, rather than a coarse-to-fine sequence. Although the coarse-to-fine processing sequence appears to be the predominant way of operating, the processing sequence of spatial scale information has been found to be relatively flexible, and dependent on task demands [14, 29, 30]. A subsequent study by Schyns and Oliva [29] showed that the spatial scale preferentially processed in hybrid images can be constrained by a phase of prior sensitization which implicitly "primes" visual processing in favor of a particular scale (coarse or fine). After initial exposure to LSF information, the subsequent categorization of hybrid images was preferentially performed following LSF cues, whereas it was biased towards HSF information after priming by HSF. By using hybrid faces instead of scenes, Schyns and Oliva [30] showed that HSF information was preferentially used to determine whether a face was expressive or not, whereas LSF information was preferentially used to categorize emotion (e.g., happy, angry). The demands of a categorization task may, therefore, determine which range of spatial frequencies is extracted, and subsequently processed, from hybrid stimuli (even when using brief presentation times, such as 45 ms). In all, these studies suggest that all spatial frequencies were available at the beginning of categorization, and that both types of processing sequence may coexist in the visual system. The selection of spatial frequencies during the recognition of natural scenes may depend on dynamic interactions between the information requirements of a given recognition task and the perceptual information available. While a coarse-to-fine sequence of spatial frequency processing may preferentially arise for normal visual inputs containing both LSF and HSF information, our visual system should nonetheless be able to prioritize the processing of HSF in certain situations, such as when searching for a target known to be defined by specific local features rather than global visual properties (e.g., find something with a striped texture).

On the Specific Role of the Occipital Cortex in Scene Perception 71

results therefore indicate that the hemisphere functionally specialized in the processing of the visual sequence of different spatial frequency inputs is the same hemisphere that is specialized in the processing of the first spatial frequency-band appearing in this sequence (i.e. right hemispheric dominance for coarse-to-fine, but also for LSF information analysis when presented alone; and conversely, left hemispheric dominance for fine-to-coarse, but also for HSF information analysis alone). This pattern suggests that the hemisphere preferentially engaged during the sequential processing of different spatial frequencies might be determined by the initial spatial frequency-band appearing in this sequence, and that both a coarse-to-fine and fine-to-coarse analysis might take place independently in both hemispheres. Our findings indicate that the visual system might be equipped with two types of cortical apparatus which are able to support scene perception differentially and flexibly, according to task demands or input sequence. The apparatus in the right occipital cortex would give priority to LSF analysis and the one in the left occipital cortex would give priority to HSF analysis. However, although a considerable number of studies postulate hemispheric specialization of spatial frequency processing in the occipital cortex, others

**5. Retinotopic organization of spatial frequency processing during scene** 

The distribution of retinal photoreceptors and retinal ganglion cells is nonhomogeneous throughout the visual system [73, 74]. The density of cones and midget ganglion cells, which are used to process HSF information, is greatest in the fovea, while the density of rods and parasol ganglion cells, which are used to process LSF information, increase with foveal eccentricity. Different imaging data obtained from patients with cerebral lesions [75, 76] and from healthy participants [77, 78] show that the human primary visual cortex is retinotopically organized. Representation of the visual field ranges from the posterior to the anterior visual cortex, and shifts from the centre to the periphery. Since the fovea is represented in the posterior areas of the visual cortex, it could well be that HSF information (conveyed by the parvocellular pathway to the visual cortex) is predominantly processed in these areas, which are devoted to foveal vision. Similarly, since the peripheral retina is represented in progressively more anterior areas of the visual cortex, LSF information (conveyed by the magnocellular pathway to the visual cortex) might well be predominantly

A large number of neurophysiological studies performed on cats [79, 80], primates [6, 81-84] and humans [16-18] have mapped representations of the different spatial frequencies in retinotopic areas. Using retinotopic encoding with achromatic sinusoidal gratings, Sasaki and collaborators [17] have, in particular, shown that LSF are mapped in occipital areas in accordance with the cortical representation of the peripheral visual field, whereas HSF are mapped in accordance with the central visual field. Other studies have demonstrated that

highlight retinotopic processing of spatial frequencies.

processed in these areas, which are devoted to peripheral vision.

**5.1. Spatial frequency tuning in retinotopic visual areas** 

**categorization** 

The cerebral occipital asymmetries observed in spatial frequency processing raise the fundamental question of the legitimacy of the coarse-to-fine sequence in the whole brain. It remains unclear whether coarse-to-fine analysis is used in both hemispheres and/or whether this sequence predominates in only one hemisphere. In an event-related fMRI experiment, we wondered whether hemispheric specialization of spatial frequency processing might underlie the flexibility of the temporal sequence used for spatial frequency analysis during scene perception [70]. In order to constrain spatial frequency processing according to different time-courses, we asked healthy participants to perform a matching task between two successive images of natural scenes (LSF or HSF), which were displayed either in a coarse-to-fine sequence (LSF scene presented first followed by HSF scene), or in a reverse fine-to-coarse sequence. Our direct inter-hemispheric comparison of the neural responses evoked by each spatial frequency sequence revealed greater activation in the right occipitotemporal cortex than in the left for the coarse-to-fine sequence, and greater activation in the left occipito-temporal cortex than in the right for the fine-to-coarse sequence. These fMRI results therefore indicate that the hemisphere functionally specialized in the processing of the visual sequence of different spatial frequency inputs is the same hemisphere that is specialized in the processing of the first spatial frequency-band appearing in this sequence (i.e. right hemispheric dominance for coarse-to-fine, but also for LSF information analysis when presented alone; and conversely, left hemispheric dominance for fine-to-coarse, but also for HSF information analysis alone). This pattern suggests that the hemisphere preferentially engaged during the sequential processing of different spatial frequencies might be determined by the initial spatial frequency-band appearing in this sequence, and that both a coarse-to-fine and fine-to-coarse analysis might take place independently in both hemispheres. Our findings indicate that the visual system might be equipped with two types of cortical apparatus which are able to support scene perception differentially and flexibly, according to task demands or input sequence. The apparatus in the right occipital cortex would give priority to LSF analysis and the one in the left occipital cortex would give priority to HSF analysis. However, although a considerable number of studies postulate hemispheric specialization of spatial frequency processing in the occipital cortex, others highlight retinotopic processing of spatial frequencies.

70 Visual Cortex – Current Status and Perspectives

striped texture).

recognition in all tasks. Indeed, the global precedence effect can be turned into a local precedence effect by simple experimental manipulation (e.g., by changing the visual angle [22, 71] or the number of local elements [72]). In Schyns and Oliva's experiments [14], a substantial proportion (29%) of hybrid sequences were in fact categorized in accordance with a fine-to-coarse, rather than a coarse-to-fine sequence. Although the coarse-to-fine processing sequence appears to be the predominant way of operating, the processing sequence of spatial scale information has been found to be relatively flexible, and dependent on task demands [14, 29, 30]. A subsequent study by Schyns and Oliva [29] showed that the spatial scale preferentially processed in hybrid images can be constrained by a phase of prior sensitization which implicitly "primes" visual processing in favor of a particular scale (coarse or fine). After initial exposure to LSF information, the subsequent categorization of hybrid images was preferentially performed following LSF cues, whereas it was biased towards HSF information after priming by HSF. By using hybrid faces instead of scenes, Schyns and Oliva [30] showed that HSF information was preferentially used to determine whether a face was expressive or not, whereas LSF information was preferentially used to categorize emotion (e.g., happy, angry). The demands of a categorization task may, therefore, determine which range of spatial frequencies is extracted, and subsequently processed, from hybrid stimuli (even when using brief presentation times, such as 45 ms). In all, these studies suggest that all spatial frequencies were available at the beginning of categorization, and that both types of processing sequence may coexist in the visual system. The selection of spatial frequencies during the recognition of natural scenes may depend on dynamic interactions between the information requirements of a given recognition task and the perceptual information available. While a coarse-to-fine sequence of spatial frequency processing may preferentially arise for normal visual inputs containing both LSF and HSF information, our visual system should nonetheless be able to prioritize the processing of HSF in certain situations, such as when searching for a target known to be defined by specific local features rather than global visual properties (e.g., find something with a

The cerebral occipital asymmetries observed in spatial frequency processing raise the fundamental question of the legitimacy of the coarse-to-fine sequence in the whole brain. It remains unclear whether coarse-to-fine analysis is used in both hemispheres and/or whether this sequence predominates in only one hemisphere. In an event-related fMRI experiment, we wondered whether hemispheric specialization of spatial frequency processing might underlie the flexibility of the temporal sequence used for spatial frequency analysis during scene perception [70]. In order to constrain spatial frequency processing according to different time-courses, we asked healthy participants to perform a matching task between two successive images of natural scenes (LSF or HSF), which were displayed either in a coarse-to-fine sequence (LSF scene presented first followed by HSF scene), or in a reverse fine-to-coarse sequence. Our direct inter-hemispheric comparison of the neural responses evoked by each spatial frequency sequence revealed greater activation in the right occipitotemporal cortex than in the left for the coarse-to-fine sequence, and greater activation in the left occipito-temporal cortex than in the right for the fine-to-coarse sequence. These fMRI

## **5. Retinotopic organization of spatial frequency processing during scene categorization**

The distribution of retinal photoreceptors and retinal ganglion cells is nonhomogeneous throughout the visual system [73, 74]. The density of cones and midget ganglion cells, which are used to process HSF information, is greatest in the fovea, while the density of rods and parasol ganglion cells, which are used to process LSF information, increase with foveal eccentricity. Different imaging data obtained from patients with cerebral lesions [75, 76] and from healthy participants [77, 78] show that the human primary visual cortex is retinotopically organized. Representation of the visual field ranges from the posterior to the anterior visual cortex, and shifts from the centre to the periphery. Since the fovea is represented in the posterior areas of the visual cortex, it could well be that HSF information (conveyed by the parvocellular pathway to the visual cortex) is predominantly processed in these areas, which are devoted to foveal vision. Similarly, since the peripheral retina is represented in progressively more anterior areas of the visual cortex, LSF information (conveyed by the magnocellular pathway to the visual cortex) might well be predominantly processed in these areas, which are devoted to peripheral vision.

## **5.1. Spatial frequency tuning in retinotopic visual areas**

A large number of neurophysiological studies performed on cats [79, 80], primates [6, 81-84] and humans [16-18] have mapped representations of the different spatial frequencies in retinotopic areas. Using retinotopic encoding with achromatic sinusoidal gratings, Sasaki and collaborators [17] have, in particular, shown that LSF are mapped in occipital areas in accordance with the cortical representation of the peripheral visual field, whereas HSF are mapped in accordance with the central visual field. Other studies have demonstrated that

more complex cognitive functions, such as visual spatial attention, are also mapped consistently by cortical retinotopy [17, 85-89]. Using very large hierarchical visual stimuli in a block design fMRI study, Sasaki and collaborators [17] found evidence for retinotopic mapping of global and local attention in the occipital cortex. During "attend global" blocks, participants were required to deliberately focus their attention on the global form (at a visual angle of 29.4°) involving their peripheral vision, while during "attend local" blocks, they had to focus on local elements (at a visual angle of 2.4°), involving more foveal vision. FMRI data were analyzed using a traditional approach based on comparisons between local and global levels. Results showed that when attention was directed at local level (as opposed to global level), activation was consistent with the cortical representation of the fovea, which is also sensitive to HSF gratings. When attention was directed at global level (as opposed to local level), activation was consistent with the cortical representation of the periphery, which is also sensitive to LSF gratings.

On the Specific Role of the Occipital Cortex in Scene Perception 73

Our results provided first of all evidence of retinotopic organization of spatial frequency processing in the human visual cortex. LSF (as opposed to HSF) scene categorization elicited medial occipital activation in the anterior half of the calcarine fissures in correspondence with the peripheral visual field, whereas HSF (as opposed to LSF) scene categorization elicited more lateral occipital activation in the posterior part of the occipital lobes in correspondence with the fovea, in accordance with retinotopic organization in visual areas (Figure 4a). By contrasting spatial frequency blocks to one another, we were, therefore, able to show that the processing of spatial frequencies is related to the organization of retinotopic eccentricity in the occipital cortex. In addition to the retinotopic activation obtained by contrasting spatial frequencies, cerebral asymmetries were also demonstrated. In order to identify cerebral asymmetries, we made direct comparisons between the two hemispheres by contrasting "unflipped" to "left-right flipped" functional images for each particular spatial frequency band (LSF and HSF). As expected from previous studies, and in accordance with our own previous results [20, 21], the inter-hemispheric method of comparison highlights occipital cortex predominance on the right (as opposed to on the left) for LSF scene categorization, and temporal cortex predominance on the left (as opposed to

**Figure 4.** Retinotopic organization and hemispheric specialization of spatial frequency processing

Using stimuli filtered in spatial frequencies covering a large part of the visual field, and depending on the method of data analysis used, we succeeded in showing that the processing of spatial frequencies is specifically organized in the visual areas of each hemisphere, as well as between the two hemispheres, according to functional lateralization. By using several different approaches on the same data, our results enabled us to reconcile

on the right) for HSF scene categorization (Figure 4b).

during scene perception

The studies mentioned previously either postulate retinotopic processing of spatial frequencies [17], or demonstrate hemispheric specialization of spatial frequency processing in the occipital cortex [19, 20]. We conducted an fMRI study to reconcile the fact that spatial frequency processing could not only be retinotopically mapped, it could also be lateralized between both hemispheres.

## **5.2. How can retinotopy and cerebral asymmetries for spatial frequencies be reconciled?**

The results obtained by Sasaki and collaborators [17] showed no hemispheric specialization for spatial frequency processing. However, the authors used a traditional method of data analysis, comparing global and local experimental conditions to one another, rather than the direct inter-hemispheric comparison method that we had used previously [20], and which had produced different results.

In a recent fMRI study, we used a categorization task (indoors vs. outdoors) of natural scenes filtered in LSF and HSF scenes, in order to evaluate, on the one hand the retinotopy, and on the other hand, functional lateralization in spatial frequency processing. With this aim in mind, we used larger scene images (at a visual angle of 24° x 18°) than in our previous studies (which used scenes with a visual angle of 4° x 4° [20, 21, 36, 43, 44, 70]), covering as broad a visual field as had Sasaki and collaborators [17]. We used a block-design fMRI paradigm, in which large LSF or HSF were displayed in separate experimental blocks. The retinotopy of spatial frequency processing was assessed using a traditional method of fMRI data analysis based on comparisons between LSF and HSF scene categorization. According to previous retinotopy studies, when processing spatial frequencies, the categorization of LSF scenes (compared to HSF) would recruit areas devoted to peripheral vision, whereas the categorization of HSF scenes (compared to LSF) would recruit areas devoted to foveal vision. Cerebral asymmetries were assessed using the inter-hemispheric comparison method. We expected a higher level of activation in right hemisphere than in the left during the processing of LSF, and more involvement of the left hemisphere during HSF processing.

Our results provided first of all evidence of retinotopic organization of spatial frequency processing in the human visual cortex. LSF (as opposed to HSF) scene categorization elicited medial occipital activation in the anterior half of the calcarine fissures in correspondence with the peripheral visual field, whereas HSF (as opposed to LSF) scene categorization elicited more lateral occipital activation in the posterior part of the occipital lobes in correspondence with the fovea, in accordance with retinotopic organization in visual areas (Figure 4a). By contrasting spatial frequency blocks to one another, we were, therefore, able to show that the processing of spatial frequencies is related to the organization of retinotopic eccentricity in the occipital cortex. In addition to the retinotopic activation obtained by contrasting spatial frequencies, cerebral asymmetries were also demonstrated. In order to identify cerebral asymmetries, we made direct comparisons between the two hemispheres by contrasting "unflipped" to "left-right flipped" functional images for each particular spatial frequency band (LSF and HSF). As expected from previous studies, and in accordance with our own previous results [20, 21], the inter-hemispheric method of comparison highlights occipital cortex predominance on the right (as opposed to on the left) for LSF scene categorization, and temporal cortex predominance on the left (as opposed to on the right) for HSF scene categorization (Figure 4b).

72 Visual Cortex – Current Status and Perspectives

periphery, which is also sensitive to LSF gratings.

between both hemispheres.

had produced different results.

**reconciled?** 

HSF processing.

more complex cognitive functions, such as visual spatial attention, are also mapped consistently by cortical retinotopy [17, 85-89]. Using very large hierarchical visual stimuli in a block design fMRI study, Sasaki and collaborators [17] found evidence for retinotopic mapping of global and local attention in the occipital cortex. During "attend global" blocks, participants were required to deliberately focus their attention on the global form (at a visual angle of 29.4°) involving their peripheral vision, while during "attend local" blocks, they had to focus on local elements (at a visual angle of 2.4°), involving more foveal vision. FMRI data were analyzed using a traditional approach based on comparisons between local and global levels. Results showed that when attention was directed at local level (as opposed to global level), activation was consistent with the cortical representation of the fovea, which is also sensitive to HSF gratings. When attention was directed at global level (as opposed to local level), activation was consistent with the cortical representation of the

The studies mentioned previously either postulate retinotopic processing of spatial frequencies [17], or demonstrate hemispheric specialization of spatial frequency processing in the occipital cortex [19, 20]. We conducted an fMRI study to reconcile the fact that spatial frequency processing could not only be retinotopically mapped, it could also be lateralized

**5.2. How can retinotopy and cerebral asymmetries for spatial frequencies be** 

The results obtained by Sasaki and collaborators [17] showed no hemispheric specialization for spatial frequency processing. However, the authors used a traditional method of data analysis, comparing global and local experimental conditions to one another, rather than the direct inter-hemispheric comparison method that we had used previously [20], and which

In a recent fMRI study, we used a categorization task (indoors vs. outdoors) of natural scenes filtered in LSF and HSF scenes, in order to evaluate, on the one hand the retinotopy, and on the other hand, functional lateralization in spatial frequency processing. With this aim in mind, we used larger scene images (at a visual angle of 24° x 18°) than in our previous studies (which used scenes with a visual angle of 4° x 4° [20, 21, 36, 43, 44, 70]), covering as broad a visual field as had Sasaki and collaborators [17]. We used a block-design fMRI paradigm, in which large LSF or HSF were displayed in separate experimental blocks. The retinotopy of spatial frequency processing was assessed using a traditional method of fMRI data analysis based on comparisons between LSF and HSF scene categorization. According to previous retinotopy studies, when processing spatial frequencies, the categorization of LSF scenes (compared to HSF) would recruit areas devoted to peripheral vision, whereas the categorization of HSF scenes (compared to LSF) would recruit areas devoted to foveal vision. Cerebral asymmetries were assessed using the inter-hemispheric comparison method. We expected a higher level of activation in right hemisphere than in the left during the processing of LSF, and more involvement of the left hemisphere during

**Figure 4.** Retinotopic organization and hemispheric specialization of spatial frequency processing during scene perception

Using stimuli filtered in spatial frequencies covering a large part of the visual field, and depending on the method of data analysis used, we succeeded in showing that the processing of spatial frequencies is specifically organized in the visual areas of each hemisphere, as well as between the two hemispheres, according to functional lateralization. By using several different approaches on the same data, our results enabled us to reconcile

for the first time retinotopic and lateralized processing of spatial frequencies in the human occipital-temporal cortex.

On the Specific Role of the Occipital Cortex in Scene Perception 75

patient and healthy participants, at both behavioural and neurobiological levels. The present findings point to a specific deficit in the processing of HSF information contained in photographs of natural scenes in AMD, linked with hypo-activation in the occipital cortex. LSF information processing was relatively well preserved. These results could also provide interesting perspectives for the diagnosis of AMD and monitoring of future

**Figure 5.** Hypoactivation in the occipital cortex during HSF scene perception in age-related macular

Our findings demonstrate that LSF information may reach high-order areas rapidly to enable coarse initial parsing of the visual scene, which could then be retro-injected through feedback into the occipital cortex to guide a finer analysis based on HSF. Furthermore, spatial frequency processing may be retinotopically mapped and lateralized in the occipital cortex. Using stimuli filtered in spatial frequencies covering a large part of the visual field, and depending on the method of data analysis, we succeeded in showing that the processing of spatial frequencies is specifically organized in the visual areas of each hemisphere, as well as between the two hemispheres, according to functional lateralization. Critically, we provided evidence that a method of fMRI data analysis based on a direct interhemispheric comparison was more appropriate than the classical method of inter-condition comparison in the evaluation of hemispheric dominance. Using a method of interhemispheric comparison, we demonstrated greater activation in right occipital areas than in left during LSF scene perception, but greater activation in left than in right occipital areas during HSF scene perception, while the inter-condition comparison revealed retinotopic processing. HSF (compared to LSF) scenes activate the foveal representation in retinotopic areas of the occipital cortex, and LSF (compared to HSF) scenes activate more peripheral representations in the same cortical areas. Even if the hypothesis was not directly tested here, we suggest that retinotopic processing may result from bottom-up visual processes, while hemispheric specialization may be controlled by top-down attentional processes (in the temporo-parietal region). Finally, our findings indicate that in scene perception, the

treatments.

degeneration (AMD)

**6. Conclusion** 

In addition to neuroimaging studies on healthy subjects, patients with retinal disorders constitute pathological models which enable the specific investigation of retinotopic mapping of spatial frequency processing in the occipital cortex through the relationship between the position of the lesion on the retina and the processing of spatial frequencies. We specifically explored the relationship between central retinal lesions in age-related macular degeneration (AMD) patients and the processing of spatial frequencies during scene categorization.

## **5.3. Scene perception and spatial frequency processing in age-related macular degeneration**

AMD, characterized by a central vision loss caused by the destruction of macular photoreceptors [90], is the primary cause of vision loss in the elderly population [91-93]. Owing to the central position of the retinal lesion, and the neurophysiology of the parvocellular and magnocellular pathways, AMD patients would be expected to be deficient in the categorization of HSF scenes compared to age-matched healthy participants. Many studies have demonstrated impairment of low-level visual processes in AMD patients (e.g., contrast sensitivity in gratings [94-96]). However, research on the ability of AMD patients to process and recognize complex visual stimuli filtered in LSF and HSF is scarce. Recent studies have shown impairment of scene perception in AMD patients [97-99]. In face perception tasks, AMD patients were able to identify facial emotions when the decision was thought to be based on LSF processing [100]. Perception of details in facial emotions, conveyed by HSF, was impaired. However, in this study, the HSF processing deficit was inferred rather than clearly demonstrated, because the spatial frequency content of faces was not manipulated explicitly. In order to test this assumption, we recently conducted behavioural experiments [101], in which AMD patients and healthy age-matched participants performed categorization tasks of large scene images (indoors vs. outdoors) filtered in LSF and HSF. The results showed that AMD patients made more non-responses and had longer reaction times for the categorization of HSF than for that of LSF scenes, whereas healthy participants' performance was not differentially affected by the spatial frequency content of scenes.

Furthermore, retinal lesions caused by AMD induce a lack of stimulation in the part of the visual cortex which is devoted to the processing of the central visual field, suggesting a reorganization of the human cortex. If HSF processing activates the foveal representation in the occipital cortex, the specific impairment of HSF processing in AMD may result in atypical occipital activation. Using our categorization task of LSF and HSF scenes under fMRI, we recently investigated the functional cerebral reorganization of spatial frequency processing in an AMD patient. The patient showed a deficit in the processing of HSF, linked with hypoactivation in the occipital cortex, compared to age-matched healthy participants (Figure 5). However, LSF processing was relatively similar in the AMD patient and healthy participants, at both behavioural and neurobiological levels. The present findings point to a specific deficit in the processing of HSF information contained in photographs of natural scenes in AMD, linked with hypo-activation in the occipital cortex. LSF information processing was relatively well preserved. These results could also provide interesting perspectives for the diagnosis of AMD and monitoring of future treatments.

**Figure 5.** Hypoactivation in the occipital cortex during HSF scene perception in age-related macular degeneration (AMD)

## **6. Conclusion**

74 Visual Cortex – Current Status and Perspectives

occipital-temporal cortex.

**macular degeneration** 

frequency content of scenes.

categorization.

for the first time retinotopic and lateralized processing of spatial frequencies in the human

In addition to neuroimaging studies on healthy subjects, patients with retinal disorders constitute pathological models which enable the specific investigation of retinotopic mapping of spatial frequency processing in the occipital cortex through the relationship between the position of the lesion on the retina and the processing of spatial frequencies. We specifically explored the relationship between central retinal lesions in age-related macular degeneration (AMD) patients and the processing of spatial frequencies during scene

AMD, characterized by a central vision loss caused by the destruction of macular photoreceptors [90], is the primary cause of vision loss in the elderly population [91-93]. Owing to the central position of the retinal lesion, and the neurophysiology of the parvocellular and magnocellular pathways, AMD patients would be expected to be deficient in the categorization of HSF scenes compared to age-matched healthy participants. Many studies have demonstrated impairment of low-level visual processes in AMD patients (e.g., contrast sensitivity in gratings [94-96]). However, research on the ability of AMD patients to process and recognize complex visual stimuli filtered in LSF and HSF is scarce. Recent studies have shown impairment of scene perception in AMD patients [97-99]. In face perception tasks, AMD patients were able to identify facial emotions when the decision was thought to be based on LSF processing [100]. Perception of details in facial emotions, conveyed by HSF, was impaired. However, in this study, the HSF processing deficit was inferred rather than clearly demonstrated, because the spatial frequency content of faces was not manipulated explicitly. In order to test this assumption, we recently conducted behavioural experiments [101], in which AMD patients and healthy age-matched participants performed categorization tasks of large scene images (indoors vs. outdoors) filtered in LSF and HSF. The results showed that AMD patients made more non-responses and had longer reaction times for the categorization of HSF than for that of LSF scenes, whereas healthy participants' performance was not differentially affected by the spatial

Furthermore, retinal lesions caused by AMD induce a lack of stimulation in the part of the visual cortex which is devoted to the processing of the central visual field, suggesting a reorganization of the human cortex. If HSF processing activates the foveal representation in the occipital cortex, the specific impairment of HSF processing in AMD may result in atypical occipital activation. Using our categorization task of LSF and HSF scenes under fMRI, we recently investigated the functional cerebral reorganization of spatial frequency processing in an AMD patient. The patient showed a deficit in the processing of HSF, linked with hypoactivation in the occipital cortex, compared to age-matched healthy participants (Figure 5). However, LSF processing was relatively similar in the AMD

**5.3. Scene perception and spatial frequency processing in age-related** 

Our findings demonstrate that LSF information may reach high-order areas rapidly to enable coarse initial parsing of the visual scene, which could then be retro-injected through feedback into the occipital cortex to guide a finer analysis based on HSF. Furthermore, spatial frequency processing may be retinotopically mapped and lateralized in the occipital cortex. Using stimuli filtered in spatial frequencies covering a large part of the visual field, and depending on the method of data analysis, we succeeded in showing that the processing of spatial frequencies is specifically organized in the visual areas of each hemisphere, as well as between the two hemispheres, according to functional lateralization. Critically, we provided evidence that a method of fMRI data analysis based on a direct interhemispheric comparison was more appropriate than the classical method of inter-condition comparison in the evaluation of hemispheric dominance. Using a method of interhemispheric comparison, we demonstrated greater activation in right occipital areas than in left during LSF scene perception, but greater activation in left than in right occipital areas during HSF scene perception, while the inter-condition comparison revealed retinotopic processing. HSF (compared to LSF) scenes activate the foveal representation in retinotopic areas of the occipital cortex, and LSF (compared to HSF) scenes activate more peripheral representations in the same cortical areas. Even if the hypothesis was not directly tested here, we suggest that retinotopic processing may result from bottom-up visual processes, while hemispheric specialization may be controlled by top-down attentional processes (in the temporo-parietal region). Finally, our findings indicate that in scene perception, the

predominantly coarse-to-fine analysis seems to be preferentially performed in the right hemisphere, from the occipital to the inferior temporal cortex.

On the Specific Role of the Occipital Cortex in Scene Perception 77

**Author details** 

**Acknowledgement** 

manuscript.

**7. References** 

107.

Nature 381: 520-2.

Physiol Opt 12: 229-32.

Carole Peyrin and Benoit Musel

*Laboratoire de Psychologie et NeuroCognition, France* 

This work was supported by a research grant from the Fyssen Foundation to Carole Peyrin, by a doctoral fellowship from the Région Rhone-Alpes to Benoit Musel and by the National Centre for Scientific Research in France. The authors extend warm thanks to the "Délégation à la Recherche Clinique et à l'Innovation" of the University Hospital of Grenoble for sponsoring their work. We thank Catherine Dal Molin for the English revision of the

[1] Thorpe S, Fize D, Marlot C (1996) Speed of processing in the human visual system.

[2] Field DJ (1987) Relations between the statistics of natural images and the response

[3] Ginsburg AP (1986) Spatial filtering and visual form perception. In: Boff K, Kauman L, Thomas J, editors. Hanbook of perception and human performance. NY: Wiley. pp. 1-41. [4] Hughes HC, Nozawa G, Kitterle FL (1996) Global precedence, spatial frequency channels, and the statistic of the natural image. Journal of Cognitive Neuroscience 8: 197-230. [5] Tolhurst DJ, Tadmor Y, Chao T (1992) Amplitude spectra of natural images. Ophthalmic

[6] De Valois RL, Albrecht DG, Thorell LG.(1982) Spatial frequency selectivity of cells in

[7] De Valois RL, Yund EW, Hepler N (1982) The orientation and direction selectivity of

[8] Poggio GF (1972) Spatial properties of neurons in striate cortex of unanesthetized

[9] Guyader N, Chauvin A, Peyrin C, Herault J, Marendaz C (2004) Image phase or amplitude? Rapid scene categorization is an amplitude-based process. C R Biol 327: 313-8. [10] Torralba A, Oliva A (2003) Statistics of natural image categories. Network 14: 391-412. [11] Bar M (2003) A cortical mechanism for triggering top-down facilitation in visual object

[12] Bullier J (2001) Integrated model of visual processing. Brain Research Reviews 36: 96-

[13] Hegde J (2008) Time course of visual perception: coarse-to-fine processing and beyond.

[14] Schyns PG, Oliva A (1994) From blobs to boundary edges: Evidence for time- and

spatial-scale-dependant scene recognition. Psychol Sci 5: 195-200.

properties of cortical cells. J Opt Soc Am A 4: 2379-94.

macaque visual cortex. Vision Res 22: 545-59.

cells in macaque visual cortex. Vision Res 22, 531-44.

macaque monkey. Invest Ophthalmol 11: 368-77.

recognition. J Cogn Neurosci 15: 600-9.

Prog Neurobiol 84: 405-39.

Our findings also demonstrate that pathology constitutes an interesting way of investigating cognitive models of spatial frequency processing, both at behavioural and cerebral levels. Future studies conducted in patients with visual field defects (following peripheral or cerebral damage) are needed to fully investigate spatial frequency processing in the occipital cortex. Behavioural studies in hemianopic patients following occipital lobe damage would specifically allow the investigation of hemispheric specialization in spatial frequency processing. Categorization of LSF scenes ought to be more impaired in patients with left hemianopic (right occipital lesion), while categorization of HSF scenes ought to be more impaired in patients with right hemianopic (left occipital lesion). In a complementary way, studies conducted on patients with contrasting retinal diseases (e.g., AMD patients characterized by a central vision loss, and retinitis pigmentosa characterized by a peripheral vision loss) would allow further investigation of the retinotopic processing of spatial frequencies. We would expect to observe a differential reorganization of the occipital cortex depending on the retinal lesion site which mirrors the dissociation of the visual disorder in question. According to the retinotopy of spatial frequency processing, if HSF processing activates the foveal representation in the occipital cortex, a specific impairment of HSF processing in AMD may result in atypical activation in occipital areas corresponding to the fovea (compared to age-matched healthy participants). Similarly, a specific impairment of LSF processing in patients with retinitis pigmentosa may result in atypical activation in occipital areas corresponding to the periphery.

On the whole, the results obtained suggest that the occipital cortex could be the point of convergence of both afferent connections from the thalamus, and feedback connections from high-order visual areas. Coarse visual information would be rapidly forwarded to high level visual areas, more specifically to the frontal and temporo-parietal areas, via the magnocellular visual pathways. This information would provide the spatial and semantic characteristics required for the identification of the visual scene. Feedback connections from frontal and temporo-parietal areas to occipital areas might then modulate the processing of fine information, slowly conveyed by the parvocellular pathway. In the occipital cortex, the hemisphere preferentially involved in visual processing may depend on the spatial frequency band required in the task. In coarse-to-fine sequence processing, the right occipital cortex would in that case be more involved than the left. Visual information would then be sent through the right ventral visual stream, from occipital to infero-temporal areas. In tasks requiring the use of fine details, visual information would be sent preferentially through the left ventral visual stream. A more advanced analysis of the scene would be performed at the very end of the ventral visual stream (e.g., in the parahippocampal place area [102]). In addition, spatial frequencies in scenes were also retinotopically mapped in the occipital cortex. However, additional studies on the time course of activation induced by spatial frequencies are necessary to specify whether retinotopic and lateralized representations emerge from ascendant or descendant processes.

## **Author details**

76 Visual Cortex – Current Status and Perspectives

hemisphere, from the occipital to the inferior temporal cortex.

occipital areas corresponding to the periphery.

processes.

predominantly coarse-to-fine analysis seems to be preferentially performed in the right

Our findings also demonstrate that pathology constitutes an interesting way of investigating cognitive models of spatial frequency processing, both at behavioural and cerebral levels. Future studies conducted in patients with visual field defects (following peripheral or cerebral damage) are needed to fully investigate spatial frequency processing in the occipital cortex. Behavioural studies in hemianopic patients following occipital lobe damage would specifically allow the investigation of hemispheric specialization in spatial frequency processing. Categorization of LSF scenes ought to be more impaired in patients with left hemianopic (right occipital lesion), while categorization of HSF scenes ought to be more impaired in patients with right hemianopic (left occipital lesion). In a complementary way, studies conducted on patients with contrasting retinal diseases (e.g., AMD patients characterized by a central vision loss, and retinitis pigmentosa characterized by a peripheral vision loss) would allow further investigation of the retinotopic processing of spatial frequencies. We would expect to observe a differential reorganization of the occipital cortex depending on the retinal lesion site which mirrors the dissociation of the visual disorder in question. According to the retinotopy of spatial frequency processing, if HSF processing activates the foveal representation in the occipital cortex, a specific impairment of HSF processing in AMD may result in atypical activation in occipital areas corresponding to the fovea (compared to age-matched healthy participants). Similarly, a specific impairment of LSF processing in patients with retinitis pigmentosa may result in atypical activation in

On the whole, the results obtained suggest that the occipital cortex could be the point of convergence of both afferent connections from the thalamus, and feedback connections from high-order visual areas. Coarse visual information would be rapidly forwarded to high level visual areas, more specifically to the frontal and temporo-parietal areas, via the magnocellular visual pathways. This information would provide the spatial and semantic characteristics required for the identification of the visual scene. Feedback connections from frontal and temporo-parietal areas to occipital areas might then modulate the processing of fine information, slowly conveyed by the parvocellular pathway. In the occipital cortex, the hemisphere preferentially involved in visual processing may depend on the spatial frequency band required in the task. In coarse-to-fine sequence processing, the right occipital cortex would in that case be more involved than the left. Visual information would then be sent through the right ventral visual stream, from occipital to infero-temporal areas. In tasks requiring the use of fine details, visual information would be sent preferentially through the left ventral visual stream. A more advanced analysis of the scene would be performed at the very end of the ventral visual stream (e.g., in the parahippocampal place area [102]). In addition, spatial frequencies in scenes were also retinotopically mapped in the occipital cortex. However, additional studies on the time course of activation induced by spatial frequencies are necessary to specify whether retinotopic and lateralized representations emerge from ascendant or descendant Carole Peyrin and Benoit Musel *Laboratoire de Psychologie et NeuroCognition, France* 

## **Acknowledgement**

This work was supported by a research grant from the Fyssen Foundation to Carole Peyrin, by a doctoral fellowship from the Région Rhone-Alpes to Benoit Musel and by the National Centre for Scientific Research in France. The authors extend warm thanks to the "Délégation à la Recherche Clinique et à l'Innovation" of the University Hospital of Grenoble for sponsoring their work. We thank Catherine Dal Molin for the English revision of the manuscript.

## **7. References**


[15] Van Essen DC, DeYoe EA (1995) Concurrent processing in the primate visual cortex. In: Gazzaniga M, editor. The cognitive Neurosciences. Cambridge: Bradford Book. pp. 383- 400

On the Specific Role of the Occipital Cortex in Scene Perception 79

[32] Riesenhuber M, Poggio T (1999) Hierarchical models of object recognition in cortex. Nat

[33] Bar M, Tootell RB, Schacter DL, Greve DN, Fischl B, Mendola JD, Rosen BR, Dale AM (2001) Cortical mechanisms specific to explicit visual object recognition. Neuron 29: 529-

[34] Rotshtein P, Vuilleumier P, Winston J, Driver J, Dolan R (2007) Distinct and convergent visual processing of high and low spatial frequency information in faces. Cereb Cortex

[35] Hupe JM, James AC, Girard P, Lomber SG, Payne BR, Bullier J (2001) Feedback connections act on the early part of the responses in monkey visual cortex. J

[36] Peyrin C, Michel CM, Schwartz S, Thut G, Seghier M, Landis T, Marendaz C, Vuilleumier P (2010) The neural substrates and timing of top-down processes during coarse-to-fine categorization of visual scenes: a combined fMRI and ERP study. J Cogn

[37] Chokron S, Bartolomeo P, Colliot P, Brickman AM, Tabert M, Wei T, Buchsbaum MS (2003) Selective attention, inhibition for repeated events and hemispheric specialization.

[38] Chokron S, Brickman AM, Wei T, Buchsbaum MS (2000) Hemispheric asymmetry for

[39] Hellige JB, Sergent J (1986) Role of task factors in visual field asymmetries. Brain Cogn

[40] Kitterle FL, Christman S, Hellige JB (1990) Hemispheric differences are found in the identification, but not the detection, of low versus high spatial frequencies. Percept

[41] Kitterle FL, Hellige JB, Christman S (1992) Visual hemispheric asymmetries depend on

[42] Kitterle FL, Selig LM (1991) Visual field effects in the discrimination of sine-wave

[43] Peyrin C, Chauvin A, Chokron S, Marendaz C (2003) Hemispheric specialization for spatial frequency processing in the analysis of natural scenes. Brain Cogn 53: 278-82. [44] Peyrin C, Mermillod M, Chokron S, Marendaz C (2006) Effect of temporal constraints on hemispheric asymmetries during spatial frequency processing. Brain Cogn 62: 214-20. [45] Sergent J (1982) Theoretical and methodological consequences of variations in exposure

[46] Sergent J (1982b) The cerebral balance of power: confrontation or cooperation? J Exp

[47] Sergent J (1983) Role of the input in visual hemispheric asymmetries. Psychol Bull 93:

[48] Sergent J, Hellige JB (1986) Role of input factors in visual-field asymmetries. Brain Cogn

which spatial frequencies are task relevant. Brain Cogn 20: 308-14.

duration in visual laterality studies. Percept Psychophys 31: 451-61.

selective attention. Brain Res Cogn Brain Res 9: 85-90.

Neurosci 2: 1019-25.

Neurophysiol 85: 134-45.

Neurosci 22: 2768-80.

Brain Cogn 53: 158-61.

Psychophys 48: 297-306.

gratings. Percept Psychophys 50: 15-8.

Psychol Hum Percept Perfor 8: 253-72.

5: 200-22.

481-512.

5: 174-99.

35.

17: 2713-24.


[32] Riesenhuber M, Poggio T (1999) Hierarchical models of object recognition in cortex. Nat Neurosci 2: 1019-25.

78 Visual Cortex – Current Status and Perspectives

400

74-81.

1-10.

34: 72-107.

1027-38.

98: 2077-2082.

[15] Van Essen DC, DeYoe EA (1995) Concurrent processing in the primate visual cortex. In: Gazzaniga M, editor. The cognitive Neurosciences. Cambridge: Bradford Book. pp. 383-

[16] Henriksson L, Raninen A, Nasanen R, Hyvarinen L, Vanni S (2007) Training-induced cortical representation of a hemianopic hemifield. J Neurol Neurosurg Psychiatry 78:

[17] Sasaki Y, Hadjikhani N, Fischl B, Liu AK, Marret S, Dale AM, Tootell RBH (2001) Local and global attention are mapped retinotopically in human occipital cortex. Psychology

[18] Singh KD, Smith AT, Greenlee MW (2000) Spatiotemporal frequency and direction sensitivities of human visual areas measured using fMRI. Neuroimage 12: 550-64. [19] Iidaka T, Yamashita K, Kashikura K, Yonekura Y (2004) Spatial frequency of visual image modulates neural responses in the temporo-occipital lobe. An investigation with

[20] Peyrin C, Baciu M, Segebarth C, Marendaz C (2004) Cerebral regions and hemispheric specialization for processing spatial frequencies during natural scene recognition. An

[21] Peyrin C, Chokron S, Guyader N, Gout O, Moret J, Marendaz C (2006) Neural correlates of spatial frequency processing: A neuropsychological approach. Brain Res 1073-1074:

[22] Kinchla RA, Wolfe JM (1979) The order of visual processing: "Top-down," "bottom-up",

[23] Navon D (1977) Forest before trees: the precedence of global features in visual

[24] Badcock JC, Whitworth FA, Badcock DR, Lovegrove WJ (1990) Low frequency filtering

[25] Lamb MR, Yund EW (1993) The role of spatial frequency in the processing of

[26] Schulman G, Sulivan M, Gisch K, Sadoka W (1986) The role of spatial frequency channels in the perception of local and global structure. Perception 15: 259-273. [27] Oliva A, Schyns PG (1997) Coarse blobs or fine edges? Evidence that information diagnosticity changes the perception of complex visual stimuli. Cognitive Psychology

[28] Parker DM, Lishman JR, Hughes J (1996) Role of coarse and fine spatial information in

[29] Schyns PG, Oliva A. (1997) Flexible, diagnosticity-driven, rather than fixed, perceptually determined scale selection in scene and face recognition. Perception 26:

[30] Schyns PG, Oliva A. (1999) Dr. Angry and Mr. Smile: when categorization flexibly modifies the perception of faces in rapid visual presentations. Cognition 69: 243-65. [31] Biederman I (1995) Visual object recognition. In: Kosslyn SF, Osherson DN, editors. An

face and object processing. J Exp Psychol Hum Percept Perform 22: 1448-66.

invitation to cognitive science. Cambridge, MA: MIT Press. pp. 121–165.

and the processing of local-global stimuli. Perception 19:, 617-629.

hierarchically organized stimuli. P Percept Psychophys 54: 773-84.

event-related fMRI. Brain Res Cogn Brain Res 18: 196-204.

event-related fMRI study. Neuroimage 23: 698-707.

or "middle-out". Percept Psychophys 25: 225-31.

perception. Cognitive Psychology 9: 353-383.


[49] Lamb MR, Robertson LC, Knight RT (1990) Component mechanisms underlying the processing of hierarchically organized patterns: inferences from patients with unilateral cortical lesions. J Exp Psychol Learn Mem Cogn 16: 471-83.

On the Specific Role of the Occipital Cortex in Scene Perception 81

[65] Yamaguchi S, Yamagata S, Kobayashi S (2000) Cerebral asymmetry of the "top-down"

[66] Palmer S (1993) Modern theories of gestalt perception. In: Humphreys GW, editor.

[67] McGlone J, Kertesz A (1973) Sex differences in cerebral processing of visuospatial tasks.

[68] Voyer D (1996) On the magnitude of laterality effects and sex differences in functional

[69] Cousin E, Peyrin C, Baciu M (2006) Hemispheric predominance assessment of phonology and semantics: a divided visual field experiment. Brain Cogn 61: 298-304. [70] Peyrin C, Schwartz S, Seghier M, Michel C, Landis T, Vuilleumier P (2005) Hemispheric specialization of human inferior temporal cortex during coarse-to-fine and fine-to-

[71] Lamb MR, Robertson LC (1990) The effect of visual angle on global and local reaction times depends on the set of visual angles presented. Percept Psychophys 47: 489-96. [72] Martin M (1979) Hemispheric specialization for local and global processing.

[73] Curcio CA, Allen KA (1990) Topography of ganglion cells in human retina. J Comp

[74] Curcio CA, Sloan KR, Kalina RE, Hendrickson AE (1990) Human photoreceptor

[75] Holmes G (1918) Disturbances of Vision by Cerebral Lesions. Br J Ophthalmol 2: 353-84. [76] Horton JC, Hoyt WF (1991) The representation of the visual field in human striate

[77] Engel SA, Glover GH, Wandell BA (1997) Retinotopic organization in human visual

[78] Engel SA, Rumelhart DE, Wandell BA, Lee AT, Glover GH, Chichilnisky EJ, Shadlen

[79] Everson RM, Prashanth AK, Gabbay M, Knight BW, Sirovich L, Kaplan E (1998) Representation of spatial frequency and orientation in the visual cortex. Proc Natl Acad

[80] Issa NP, Trepel C, Stryker MP (2000) Spatial frequency maps in cat visual cortex. J

[81] Foster KH, Gaska JP, Nagler M, Pollen DA (1985) Spatial and temporal frequency selectivity of neurones in visual cortical areas V1 and V2 of the macaque monkey. J

[82] Gegenfurtner KR, Kiper DC, Levitt JB (1997) Functional properties of neurons in

[83] Tootell RB, Silverman MS, Hamilton SL, Switkes E, De Valois RL (1988) Functional anatomy of macaque striate cortex. V. Spatial frequency. J Neurosci 8: 1610-24. [84] Xu X, Anderson TJ, Casagrande VA (2007) How do functional maps in primary visual

cortex. A revision of the classic Holmes map. Arch Ophthalmol 109: 816-24.

cortex and the spatial precision of functional MRI. Cereb Cortex 7: 181-92.

MN (1994) fMRI of human visual cortex. Nature 369: 525.

macaque area V3. J Neurophysiol 77: 1906-23.

cortex vary with eccentricity? J Comp Neurol 501: 741-55.

allocation of attention to global and local features. J Neurosci 20: RC72.

coarse analysis of natural visual scenes. Neuroimage 28: 464-73.

Understanding Vision. Blackwell: Oxford. pp. 39-70.

Cortex 9: 313-20.

lateralities. Laterality 1: 51-83.

Neuropsychologia 17: 33-40.

topography. J Comp Neurol 292: 497-523.

Neurol 300: 5-25.

Sci U S A 95: 8334-8.

Neurosci 20: 8504-14.

Physiol 365: 331-63.


[65] Yamaguchi S, Yamagata S, Kobayashi S (2000) Cerebral asymmetry of the "top-down" allocation of attention to global and local features. J Neurosci 20: RC72.

80 Visual Cortex – Current Status and Perspectives

[49] Lamb MR, Robertson LC, Knight RT (1990) Component mechanisms underlying the processing of hierarchically organized patterns: inferences from patients with unilateral

[50] Robertson LC, Lamb MR (1991) Neuropsychological contributions to theories of

[51] Robertson LC, Lamb MR, Knight RT (1988) Effects of lesions of temporal-parietal junction on perceptual and attentional processing in humans. J Neurosci 8: 3757-69. [52] Buchsbaum MS, Buchsbaum BR, Chokron S, Tang C, Wei TC, Byne W (2006) Thalamocortical circuits: fMRI assessment of the pulvinar and medial dorsal nucleus in

[53] Fink GR, Halligan PW, Marshall JC, Frith CD, Frackowiak RS, Dolan RJ (1996) Where in the brain does visual attention select the forest and the trees? Nature 382: 626-8. [54] Fink GR, Halligan PW, Marshall JC, Frith CD, Frackowiak RS, Dolan RJ (1997) Neural mechanisms involved in the processing of global and local aspects of hierarchically

[55] Fink R, Marshall JC, Halligan PW, Dolan RJ (2000) Neuronal activity in early visual areas during global and local processing: a comment on Heinze, Hinrichs, Scholz,

[56] Han S, Weaver JA, Murray SO, Kang X, Yund EW, Woods DL (2002) Hemispheric asymmetry in global/local processing: Effects of stimulus position and spatial

[57] Heinze HJ, Hinrichs H, Scholz M, Burchert W, Mangun GR (1998) Neural mechanisms of global and local processing. A combined PET and ERP study. J Cogn Neurosci 10: 485-98. [58] Kenemans JL, Baas JM, Mangun GR, Lijffijt M, Verbaten MN (2000) On the processing of spatial frequencies as revealed by evoked-potential source modeling. Clin

[59] Lux S, Marshall JC, Ritzl A, Weiss PH, Pietrzyk U, Shah NJ, Zilles K, Fink GR (2004) A functional magnetic resonance imaging study of local/global processing with stimulus

[60] Mangun GR, Heinze HJ, Scholz M, Hinrichs H (2000) Neural activity in early visual areas during global and local processing: a reply to Fink, Marshall, Halligan and Dolan.

[61] Martinez A, DiRusso F, Anllo-Vento L, Sereno MI, Buxton RB, Hillyard SA (2001) Putting spatial attention on the map: timing and localization of stimulus selection

[62] Martinez A, Moses P, Frank L, Buxton R, Wong E, Stiles J (1997) Hemispheric asymmetries in global and local processing: evidence from fMRI. Neuroreport 8: 1685-9. [63] Weissman DH, Woldorff MG (2005) Hemispheric asymmetries for different components of global/local attention occur in distinct temporo-parietal loci. Cereb

[64] Wilkinson DT, Halligan PW, Marshall JC, Buchel C, Dolan RJ (2001) Switching between the forest and the trees: brain systems involved in local/global changed-level

presentation in the peripheral visual hemifields. Neuroscience 124: 113-20.

processes in striate and extrastriate visual areas. Vision Res 41: 1437-57.

cortical lesions. J Exp Psychol Learn Mem Cogn 16: 471-83.

part/whole organization. Cogn Psychol 23: 299-330.

normal volunteers. Neurosci Lett 404: 282-7.

organized visual stimuli. Brain 120: 1779-91.

Burchert Mangun. J Cogn Neurosci 12: 355-6.

frequency. Neuroimage 17: 1290-1299.

Neurophysiol 111: 1113-23.

J Cogn Neuroscie 12: 357-359.

judgments. Neuroimage 13: 56-67.

Cortex 15: 870-6.


[85] Brefczynski JA, DeYoe EA (1999) A physiological correlate of the 'spotlight' of visual attention. Nat Neurosci 2: 370-4.

**Chapter 4** 

© 2012 Buckthought and Mendola, licensee InTech. This is an open access chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is

© 2012 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution,

properly cited.

networks activated in each of these different domains.

**Neural Mechanisms for Binocular Depth,** 

The purpose of this chapter is to present a review of recent functional neuroimaging (fMRI) studies of binocular vision, including binocular depth and rivalry, as well as a review of studies of perceptual multistability. As such, we will first emphasize the binocular aspects of binocular rivalry, while later emphasizing the rivalrous aspects. The interrelationship of binocular depth and rivalry, as well as multistability, will be described with reference to fMRI studies and single-unit recording studies in animals. These studies have provided provocative new evidence that the neural substrates for depth and rivalry, as well as other forms of multistability are remarkably similar. We will also describe our own research findings from two recent experiments, in which we performed (1) a direct comparison between binocular rivalry and depth, and (2) a direct comparison between binocular rivalry and monocular rivalry, a related form of bistability [1,2]. Our studies are unique in using both matched stimulation and comparable tasks, overcoming a limitation in the interpretation of many previous studies. As a result, these experiments are particularly relevant in delineating some of the global similarities and differences in the cortical

Binocular depth perception arises as a consequence of the slightly displaced point of view of the two eyes. The horizontal displacement of image features in the two eyes (i.e. binocular disparities) makes it possible to reconstruct the depth relationships in the visual world. Binocular matching of local features in the retinal images may be used to obtain estimates of the absolute disparity (and distance) of objects or surfaces, as well as the relative disparity (or relative distances) between different objects. An example of an image with binocular depth is shown in Figure 1a. If the left and right images are cross-fused, the image appears

and reproduction in any medium, provided the original work is properly cited.

**Rivalry and Multistability** 

Athena Buckthought and Janine D. Mendola

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/48446

**1. Introduction** 

**2. Binocular depth** 

