**3.5 Crossmodal interplay in transitory cortical regions bridging primary sensory areas**

The main focus of the present review is on multimodal integration in primary cortical areas. Nonetheless, while considering low-level crossmodal interplay, it is relevant to mention findings related to multimodal convergence and integration in transitional zones lying between primary areas. Visual and somatosensory inputs converge and interact in a graded multisensory zone forming a narrow strip within the associative parietal cortex (APC) of rodent. Previous findings analyzing current source density [53] (rat), or using calcium imaging [54] (mouse) or

**145**

*after acute UVN [modified from Facchini et al. (in press)].*

**Figure 1.**

*Interplay between Primary Cortical Areas and Crossmodal Plasticity*

*Immediate effects of unilateral vestibular neurectomy on hindpaw cutaneous representation in S1. (A) Typical receptive fields (RFs) recorded in the S1 cortex and located on hindpaw glabrous skin surfaces of an intact rat (CTRL, left panel) and after double mapping, on ipsilesional (ipsi) and contralesional (contra) hindpaws one hour after unilateral vestibular neurectomy (UVN) (1H, right panel). In green: small size RFs covering less than 10% of the total skin surface of the paw; in purple: medium size RFs >10% and <40% of the paw surface; in yellow: large size RFs including more than 40% of the paw surface. (B) Distribution of plantar cutaneous RF recorded in CTRL and UVN rats. The height of each area of the stacked histogram represents the mean proportion of RFs falling into each category (green: small; purple: medium; yellow: large RFs). (C) Representative electrophysiological cortical maps obtained from an intact rat (CTRL: left panel) and from two rats in which ipsilesional and contralesional hindpaw maps were obtained starting 1 hour after UVN (1H, right panel). The map remodeling was accounted for by the expansion of plantar cutaneous RF illustrated in A-B. Note the drastic dedifferentiation of the somatotopic maps. Simple areas correspond to neurons with RF located on the ventral or dorsal aspect of individual fingers, or encompassing palmar pads. Mixed areas correspond to neurons displaying enlarged RFs extending beyond the somatotopic regions observed in prelesion hindlimb maps, i.e. RFs on 2 or more skin territories of the hindpaw. (D) Stacked histogram showing the relative mean areas of the different map regions, normalized with respect to the total hindlimb area. These relative areas are color-coded (green: simple area; purple: mixed area). \*\*: P < 0.01; \*\*\*: P < 0.001 (comparisons with control values); (Kruskal Wallis analysis and Dunnett's post-hoc test). Vertical bars illustrate standard errors of the means (SEM). Hairy skin RFs and representational zones were not altered* 

*DOI: http://dx.doi.org/10.5772/intechopen.95450*

*Interplay between Primary Cortical Areas and Crossmodal Plasticity DOI: http://dx.doi.org/10.5772/intechopen.95450*

#### **Figure 1.**

*Connectivity and Functional Specialization in the Brain*

driving cutaneous responses in S1.

posturo-locomotor adjustments.

**sensory areas**

congruent with the functional role of these inputs in posturo-locomotor regulation. Accordingly, we evaluated the immediate effects of a complete unilateral vestibular neurectomy on the response properties of S1 neurons in the hindpaw cutaneous representations [45]. We found that the acute deafferentation immediately induces a bilateral expansion of the cutaneous receptive fields that exclusively concerned those located on the plantar skin surfaces. A corrolary effect consisted of a dedifferentiation of the topographic organization of the cortical maps representing these surfaces (**Figure 1**). However, this somatotopy disruption was relatively less pronounced for the representation of ipsilesional hindpaw, consistently with the contralateral predominance of vestibulo-thalamic projections [46, 47] (cat). The rapid deafferentation-induced expansion of cutaneous receptive field indicates that in intact animals, vestibular inputs exert a suppressive effect onto synaptic inputs

It is well documented that cortical maps are dynamically reshaped through ongoing adjustments in the balance of excitatory and inhibitory influences on their constituent neurons. Hence, it is very likely that the receptive field enlargement induced by the vestibular loss results from a disinhibition process, possibly via thalamo-cortical inputs on S1 inhibitory interneurons or direct cortico-cortical connections*.* It has long been argued that intracortical inhibition plays a key role in controlling the spatial selectivity of cortical neurons through segregation of broad sets of converging synaptic inputs. Consistently, studies have reported a substantial enlargement of the cutaneous receptive field of somatosensory cortical neurons when GABA-mediated local inhibition was antagonized by an intracortical bicuculine injection [48–50] (cat; racoon), whereas injection of baclofen, a selective agonist for the GABAA receptors, induced a shrinkage of these receptive fields [51] (rat). A release of afferent-driven intracortical tonic inhibition results in an enhanced effectiveness of convergent cutaneous inputs. Therefore, this could be a most likely mechanism for rapid unmasking of previously subthreshold afferent connections reflected by the rapid expansion of cutaneous receptive field of S1 neurons following the loss of vestibular inputs. Our results extend previous findings, already mentioned in the present review, showing that auditory inputs to V1 decrease visually induced activity (mouse), while acute hearing loss releases the inhibitory effects of A1 neurons on visually elicited responses in V1 and leads to a concomitant increase in V1 activation [52] (mouse). As also previously noted, auditory stimulation sharpens the orientation selectivity of V1 neurons [22] (mouse). Collectively, the available evidence supports the view that, in normal conditions, cross modal modulation between primary cortical cortices may act to improve the tuning of neuronal response properties in these areas. In our study, the postlesion expansion of cutaneous RF was selectively located on the hindpaw plantar skin surfaces. Hence, we hypothesize that, in normal conditions, the vestibular influences on the S1 cortex could improve tactile acuity during perceptually guided

**3.5 Crossmodal interplay in transitory cortical regions bridging primary** 

The main focus of the present review is on multimodal integration in primary cortical areas. Nonetheless, while considering low-level crossmodal interplay, it is relevant to mention findings related to multimodal convergence and integration in transitional zones lying between primary areas. Visual and somatosensory inputs converge and interact in a graded multisensory zone forming a narrow strip within the associative parietal cortex (APC) of rodent. Previous findings analyzing current source density [53] (rat), or using calcium imaging [54] (mouse) or

**144**

*Immediate effects of unilateral vestibular neurectomy on hindpaw cutaneous representation in S1. (A) Typical receptive fields (RFs) recorded in the S1 cortex and located on hindpaw glabrous skin surfaces of an intact rat (CTRL, left panel) and after double mapping, on ipsilesional (ipsi) and contralesional (contra) hindpaws one hour after unilateral vestibular neurectomy (UVN) (1H, right panel). In green: small size RFs covering less than 10% of the total skin surface of the paw; in purple: medium size RFs >10% and <40% of the paw surface; in yellow: large size RFs including more than 40% of the paw surface. (B) Distribution of plantar cutaneous RF recorded in CTRL and UVN rats. The height of each area of the stacked histogram represents the mean proportion of RFs falling into each category (green: small; purple: medium; yellow: large RFs). (C) Representative electrophysiological cortical maps obtained from an intact rat (CTRL: left panel) and from two rats in which ipsilesional and contralesional hindpaw maps were obtained starting 1 hour after UVN (1H, right panel). The map remodeling was accounted for by the expansion of plantar cutaneous RF illustrated in A-B. Note the drastic dedifferentiation of the somatotopic maps. Simple areas correspond to neurons with RF located on the ventral or dorsal aspect of individual fingers, or encompassing palmar pads. Mixed areas correspond to neurons displaying enlarged RFs extending beyond the somatotopic regions observed in prelesion hindlimb maps, i.e. RFs on 2 or more skin territories of the hindpaw. (D) Stacked histogram showing the relative mean areas of the different map regions, normalized with respect to the total hindlimb area. These relative areas are color-coded (green: simple area; purple: mixed area). \*\*: P < 0.01; \*\*\*: P < 0.001 (comparisons with control values); (Kruskal Wallis analysis and Dunnett's post-hoc test). Vertical bars illustrate standard errors of the means (SEM). Hairy skin RFs and representational zones were not altered after acute UVN [modified from Facchini et al. (in press)].*

voltage-sensitive dye imaging (VSDI) described this region as heteromodal [55] (rat). In addition, a gradual merging of modalities from the borders of the primary cortices to the middle of the APC strip has been reported [54] (mouse). Using optical imaging combined with laminar electrophysiological recordings, it was observed that both inputs elicited similar response patterns in this cortical zone [53] (rat) However, current source density analysis of event-related potentials revealed a supra-additive interaction of subthreshold activity when the somatosensory response preceded the visual response, whereas a sub-linear summation was induced by reversing the stimulus order. This finding suggests an asymmetry in the excitation-inhibition balance mediated by the underlying connectivity network, that may be consistent with the observation that visual responses were located deeper than somatosensory responses. The laminar pattern of these visual-somatosensory interactions and the fact that they vanished upon GABAergic silencing of local post-synaptic activity suggest their intracortical origin.

In a recent study, we investigated the neural processing of visual and somatosensory motion cues in individual neurons of the APC [56] (rat). The animals were exposed to moving visual gratings presented in different directions, with various motion speeds, and to air puffs deflecting bilaterally all the whiskers in the antero-posterior (backward) or postero-anterior (forward) directions. When delivered simultaneously, visual and tactile stimuli could be either in the same or opposite direction (congruent or incongruent). We used both voltage-sensitive dye imaging to identify the cortical zone of convergence of tactile and visual afferents, and single-unit recordings to investigate the uni- and bimodal processing of these inputs. We showed the convergence of visual and tactile information, both in layer 2/3 as revealed by VSDI, and in layer 4, as demonstrated by the single-unit recordings. Both whisker deflections and visual moving gratings evoked neural responses in the APC, with similar magnitudes, reflecting the convergence of equally weighted visual and somatosensory information (**Figure 2**). The majority of recorded cells were bimodal with about 50% exhibiting a directional congruence for the stimulus orientations tested, which strongly points to a potential role of the APC in heteromodal sensory integration. A machine learning approach revealed that the integration of the visual-tactile motion stimuli relies predominantly on the bimodal population, as performing decoding on the unimodal neurons did not yield accuracies above chance. In addition, we found that visual neurons in APC selectively respond to the direction (about 50%) and speed (about 30%) of visual grating motion, while somatosensory neurons display a direction selectivity for whisker stimulation (about 60%). Like in the study mentioned above [53], a temporal dissociation was observed between somatosensory and visual responses, both in the supragranular and granular layers, as the somatosensory stimulations evoked earlier responses than did the visual stimulations. This finding underscores the importance of timing in multimodal integration, and is consistent with the view that whiskers information predominantly relates to fast changing contacts with objects or congeners, while vision mainly provides information about the physical and social environment that likely facilitates the interpretation of somatosensory information. It is plausible that APC is designed as a hub in which multisensory motion information is integrated to contribute to elaborating in higher-order areas a supramodal percept guiding purposeful behavior. Interestingly, these animal studies are consistent with human investigations showing the existence of a multisensory homunculus posterior to S1, along the postcentral sulcus, that overlaps the most anterior retinotopic map with a topographic alignment of tactile and visual representations [57, 58]. The authors proposed that these multisensory topographically organized maps may play a pivotal role in perception and cognition related to peripersonal space.

**147**

**Figure 2.**

*Interplay between Primary Cortical Areas and Crossmodal Plasticity*

*Convergence of visual and somatosensory inputs in the associative parietal cortex (APC). (A) Example of cortical activation dynamics evoked by somatosensory (upper row) or visual (lower row) stimulation revealed by voltage-sensitive-dye imaging. The latency to the somatosensory stimulation (60 – 90 ms) is shorter than to the visual stimulation (150 – 350 ms). The value in ms indicate the time after stimulation onset. (B) Example of mean DF/F over time. Time course of responses to the unimodal stimuli, in V1, S1 and APC with a 3D representation of selectivity indices. For each pixel of the acquisition window the colormap depicts the level of selectivity. The low selectivity belt (yellow) is characterized by comparable levels of activation ("ROI in low selectivity belt (APC)" plot), while high selectivity indices (blue) are observed in S1 and V1. (C) Spatial distribution of the recorded neurons corresponding to the visual, somatosensory and bimodal conditions for a representative animal. The heights of the histograms represent the proportions of neurons recorded at the corresponding cortical sites. The proportion of direction selective cells in the neuronal populations is indicated (backward, green; forward, yellow; not selective, gray). Examples of a recorded unit significantly responding only to the backward visual stimulus (pink line), a unit responding to both air puff directions, with a larger spiking probability to the forward direction and a bimodal neuron that responds to all 4 conditions. The inset presents the unit's waveform. The spiking probabilities to each condition are represented as a function of time from stimulus onset. Note the latency shift existing in neuronal response to visual and somatosensory conditions in the neurons recorded in APC. (D) Proportions of visual (pink), somatosensory (yellow), bimodal (blue) and non-responsive (gray) neurons recorded in the APC (N = 914) (modified from Caron-Guyon et al. [56]).*

*DOI: http://dx.doi.org/10.5772/intechopen.95450*

#### *Interplay between Primary Cortical Areas and Crossmodal Plasticity DOI: http://dx.doi.org/10.5772/intechopen.95450*

#### **Figure 2.**

*Connectivity and Functional Specialization in the Brain*

local post-synaptic activity suggest their intracortical origin.

voltage-sensitive dye imaging (VSDI) described this region as heteromodal [55] (rat). In addition, a gradual merging of modalities from the borders of the primary cortices to the middle of the APC strip has been reported [54] (mouse). Using optical imaging combined with laminar electrophysiological recordings, it was observed that both inputs elicited similar response patterns in this cortical zone [53] (rat) However, current source density analysis of event-related potentials revealed a supra-additive interaction of subthreshold activity when the somatosensory response preceded the visual response, whereas a sub-linear summation was induced by reversing the stimulus order. This finding suggests an asymmetry in the excitation-inhibition balance mediated by the underlying connectivity network, that may be consistent with the observation that visual responses were located deeper than somatosensory responses. The laminar pattern of these visual-somatosensory interactions and the fact that they vanished upon GABAergic silencing of

In a recent study, we investigated the neural processing of visual and somatosensory motion cues in individual neurons of the APC [56] (rat). The animals were exposed to moving visual gratings presented in different directions, with various motion speeds, and to air puffs deflecting bilaterally all the whiskers in the antero-posterior (backward) or postero-anterior (forward) directions. When delivered simultaneously, visual and tactile stimuli could be either in the same or opposite direction (congruent or incongruent). We used both voltage-sensitive dye imaging to identify the cortical zone of convergence of tactile and visual afferents, and single-unit recordings to investigate the uni- and bimodal processing of these inputs. We showed the convergence of visual and tactile information, both in layer 2/3 as revealed by VSDI, and in layer 4, as demonstrated by the single-unit recordings. Both whisker deflections and visual moving gratings evoked neural responses in the APC, with similar magnitudes, reflecting the convergence of equally weighted visual and somatosensory information (**Figure 2**). The majority of recorded cells were bimodal with about 50% exhibiting a directional congruence for the stimulus orientations tested, which strongly points to a potential role of the APC in heteromodal sensory integration. A machine learning approach revealed that the integration of the visual-tactile motion stimuli relies predominantly on the bimodal population, as performing decoding on the unimodal neurons did not yield accuracies above chance. In addition, we found that visual neurons in APC selectively respond to the direction (about 50%) and speed (about 30%) of visual grating motion, while somatosensory neurons display a direction selectivity for whisker stimulation (about 60%). Like in the study mentioned above [53], a temporal dissociation was observed between somatosensory and visual responses, both in the supragranular and granular layers, as the somatosensory stimulations evoked earlier responses than did the visual stimulations. This finding underscores the importance of timing in multimodal integration, and is consistent with the view that whiskers information predominantly relates to fast changing contacts with objects or congeners, while vision mainly provides information about the physical and social environment that likely facilitates the interpretation of somatosensory information. It is plausible that APC is designed as a hub in which multisensory motion information is integrated to contribute to elaborating in higher-order areas a supramodal percept guiding purposeful behavior. Interestingly, these animal studies are consistent with human investigations showing the existence of a multisensory homunculus posterior to S1, along the postcentral sulcus, that overlaps the most anterior retinotopic map with a topographic alignment of tactile and visual representations [57, 58]. The authors proposed that these multisensory topographically organized maps may play a pivotal role in perception and cognition related to

**146**

peripersonal space.

*Convergence of visual and somatosensory inputs in the associative parietal cortex (APC). (A) Example of cortical activation dynamics evoked by somatosensory (upper row) or visual (lower row) stimulation revealed by voltage-sensitive-dye imaging. The latency to the somatosensory stimulation (60 – 90 ms) is shorter than to the visual stimulation (150 – 350 ms). The value in ms indicate the time after stimulation onset. (B) Example of mean DF/F over time. Time course of responses to the unimodal stimuli, in V1, S1 and APC with a 3D representation of selectivity indices. For each pixel of the acquisition window the colormap depicts the level of selectivity. The low selectivity belt (yellow) is characterized by comparable levels of activation ("ROI in low selectivity belt (APC)" plot), while high selectivity indices (blue) are observed in S1 and V1. (C) Spatial distribution of the recorded neurons corresponding to the visual, somatosensory and bimodal conditions for a representative animal. The heights of the histograms represent the proportions of neurons recorded at the corresponding cortical sites. The proportion of direction selective cells in the neuronal populations is indicated (backward, green; forward, yellow; not selective, gray). Examples of a recorded unit significantly responding only to the backward visual stimulus (pink line), a unit responding to both air puff directions, with a larger spiking probability to the forward direction and a bimodal neuron that responds to all 4 conditions. The inset presents the unit's waveform. The spiking probabilities to each condition are represented as a function of time from stimulus onset. Note the latency shift existing in neuronal response to visual and somatosensory conditions in the neurons recorded in APC. (D) Proportions of visual (pink), somatosensory (yellow), bimodal (blue) and non-responsive (gray) neurons recorded in the APC (N = 914) (modified from Caron-Guyon et al. [56]).*

The set of studies reviewed highlights the broad panoply of connectivity patterns and functional interactions between primary areas that underpin a flexible cooperation at an early stage of sensory processing. Tentatively, we propose that early crossmodal interactions in primary areas contribute to refining and sharpening neural response tuning adapted to improving "immediate" perception and eliminating perceptual ambiguity. This perceptual optimization could occur through rapid neurophysiological mechanisms operating in the corticocortical circuitry (e. g., local suppressive inhibition, sub-additive or supra-additive integration, oscillatory entrainment of neuronal networks) and serve automated behavioral responses. According to this view, cognitive influences exerted onto higher-order cortical integration areas, operating through relatively slower mechanisms, would adaptively modulate the early multimodal integration to fulfill a more complex integration processing influenced by attention and motivation so as to adjust perception to a continuously changing behavioral context.
