**3. Estrogen and cognition**

The term cognition indicates the totality of human information processing, including psychomotor skills, pattern recognition, attention, language, learning and memory, problem solving, abstract reasoning or higher-order intellectual functioning.

In female mammals, including rodents and non-human primates, estrogen effects on nonreproductive behaviors include, besides anxiety and depressive-like behaviors, cognitive behaviors (Spencer et al, 2008; Walf & Frye, 2006) When administered to ovariectomized (OVX) rats, estradiol decreases anxiety and depressant behavior in laboratory tests (Walf & Frye, 2006). The effects of estrogen on cognition depend on the type of task performed and on the brain regions involved. For instance, while estradiol impairs performance on striatum-dependent tasks in female rats (Davis et al, 2005; Korol, 2004) it improves performance on prefrontal cortical-dependent learning in female rats (Luine, 2008) rhesus monkeys (Hao et al, 2007; Rapp et al, 2003) and both young adult and post-menopausal women (Berman, 1997) . It also enhances performance on HF-dependent tasks in female mice (Li et al, 2004; Xu & Zhang, 2006) , rats (Daniel et al, 1997; Sandstrom & Williams, 2004) and rhesus monkeys (Lacreuse et al, 2002; Rapp et al, 2003). Findings showing improved performance after estradiol infusion directly into the HF, but not other cerebral areas (Zurkovsky et al, 2006), provides behavioral evidence that the estradiol enhancement of HFdependent tasks indeed represents a specific effect on HF function. However, estrogen's roles on cognitive function may result from the sum of interacting influences on numerous cerebral regions, including striatum, HF, basal forebrain, and prefrontal cortex (PFC).

#### **3.1 Estrogen in learning and memory**

Signaling pathways and gene expression regulated by estrogen include activation of CREB, GABA-A receptors, NMDA receptors, glutamic acid decarboxylase (GAD), ChAT, and synaptic and spine-associated proteins (Frick et al, 2002; McEwen et al, 2001; Rudick & Woolley , 2003).

Studies in knockout mice using the selective estrogen receptor modulators suggest that ERα and ERβ contribute differently to memory mechanisms (Rhodes & Frye, 2006; Rissman et al, 2002). Several studies have shown estrogen regulation of ERα (Hart et al, 2001; 2007) Moreover, it was reported that selective ERβ agonists increased levels of key synaptic proteins in vivo in the HF, and these effects were absent in ERβ knockout mice or after treatment with an ERα agonist. ERβ agonists also induced morphological changes in HF neurons, such as an enhanced density of mushroom-type spines. Most importantly, estrogen or ERβ agonists improved performance in some HF-dependent memory tasks (Liu et al, 2008). Therefore, these results confirm the role of ERβ in memory, but cross-talk between ERα and ERβ receptors cannot be excluded.

It was also evidenced that rapid improvements in cognition could be mediated by membrane associated estrogen receptors activating mitogen-activated protein kinase

are markedly decreased in Alzheimer's disease (AD) (Gibbs & Aggarwal, 1998). The neuroprotective action of estrogen could also be exerted through a modulator effect on molecules involved in apoptosis (Pike, 1999) and its antioxidant action. The potential for the numerous mechanisms of action of estrogen to affect the structure and function of cerebral areas that subserve several cognitive functions provides biological plausibility for the

The term cognition indicates the totality of human information processing, including psychomotor skills, pattern recognition, attention, language, learning and memory, problem

In female mammals, including rodents and non-human primates, estrogen effects on nonreproductive behaviors include, besides anxiety and depressive-like behaviors, cognitive behaviors (Spencer et al, 2008; Walf & Frye, 2006) When administered to ovariectomized (OVX) rats, estradiol decreases anxiety and depressant behavior in laboratory tests (Walf & Frye, 2006). The effects of estrogen on cognition depend on the type of task performed and on the brain regions involved. For instance, while estradiol impairs performance on striatum-dependent tasks in female rats (Davis et al, 2005; Korol, 2004) it improves performance on prefrontal cortical-dependent learning in female rats (Luine, 2008) rhesus monkeys (Hao et al, 2007; Rapp et al, 2003) and both young adult and post-menopausal women (Berman, 1997) . It also enhances performance on HF-dependent tasks in female mice (Li et al, 2004; Xu & Zhang, 2006) , rats (Daniel et al, 1997; Sandstrom & Williams, 2004) and rhesus monkeys (Lacreuse et al, 2002; Rapp et al, 2003). Findings showing improved performance after estradiol infusion directly into the HF, but not other cerebral areas (Zurkovsky et al, 2006), provides behavioral evidence that the estradiol enhancement of HFdependent tasks indeed represents a specific effect on HF function. However, estrogen's roles on cognitive function may result from the sum of interacting influences on numerous cerebral regions, including striatum, HF, basal forebrain, and prefrontal cortex (PFC).

Signaling pathways and gene expression regulated by estrogen include activation of CREB, GABA-A receptors, NMDA receptors, glutamic acid decarboxylase (GAD), ChAT, and synaptic and spine-associated proteins (Frick et al, 2002; McEwen et al, 2001; Rudick &

Studies in knockout mice using the selective estrogen receptor modulators suggest that ERα and ERβ contribute differently to memory mechanisms (Rhodes & Frye, 2006; Rissman et al, 2002). Several studies have shown estrogen regulation of ERα (Hart et al, 2001; 2007) Moreover, it was reported that selective ERβ agonists increased levels of key synaptic proteins in vivo in the HF, and these effects were absent in ERβ knockout mice or after treatment with an ERα agonist. ERβ agonists also induced morphological changes in HF neurons, such as an enhanced density of mushroom-type spines. Most importantly, estrogen or ERβ agonists improved performance in some HF-dependent memory tasks (Liu et al, 2008). Therefore, these results confirm the role of ERβ in memory, but cross-talk between

It was also evidenced that rapid improvements in cognition could be mediated by membrane associated estrogen receptors activating mitogen-activated protein kinase

hypothesis that estrogen could protect cognitive functions in aging women.

solving, abstract reasoning or higher-order intellectual functioning.

**3. Estrogen and cognition** 

**3.1 Estrogen in learning and memory** 

ERα and ERβ receptors cannot be excluded.

Woolley , 2003).

(MAPK) signalling pathways in specific neural sites (Bryant et al, 2006). For example, estrogen enhances performance in tasks such as inhibitory avoidance (IA), object recognition and placement within 4h of treatment; a post-training paradigm evidenced that these effects are due to the facilitatory action of estrogen on memory (Frye et al, 2007; Luine, 2008; Rhodes & Frye, 2006; Walf & Frye, 2006). Previous memory studies hypothesized that newly-acquired informations are transferred to long-term memory over time, and seminal work by McGaugh and co-workers has shown that consolidation takes place within 1–2 h post-training (McGaugh, 2000). In addition, the impairment or improvement of the consolidation process due to drugs or hormones can occur if they are given within this time, but not later. Estrogen-related enhancement of consolidation utilizing post-training paradigms have been shown in some memory tasks, such as Morris water maze (MWM), IA, object recognition and object placement (Frye et al, 2007; Luine et al, 2008; Rhodes & Frye, 2006; Walf & Frye, 2006). Administration of the powerful estrogen agonist, diethylstilbestrol, either immediately before or immediately after the presentation of objects, increased discrimination between previously viewed and never viewed items in the recognition trial. Therefore, temporal relations between hormonal application and performance enhancements are in agreement with memory improvement.

Estrogen not only modulates memory formation and maintenance processes in some contexts, but also biases the learning strategy utilized to solve a task, thus changing what and how information is learned, and therefore not only how much is learned, i.e., the strength of the memory (Gasbarri et al, 2009, Pompili et al, 2010).

Rats with high estrogen levels utilize place or allocentric strategies rather successfully, outperforming hormone-deprived rats on tasks requiring the configuration and use of extramaze cues for successful completion. On the contrary, rats with low estrogen levels tend to use response or egocentric strategies on tasks where the use of a directional turn, e.g., left or right, is required for acquisition (Korol, 2004). Taking into account the actions of estrogen across a large range of neural systems, its modulation on cognition could be exerted by altering the relative involvement of specific memory systems, acting much like a conductor, orchestrating the dynamics, timing and coordination of multiple cognitive strategies during learning (McGaugh, 2000) . Influences on neurotransmitters, such as acetylcholine (ACh), regulating other processes, like inhibitory tone and excitability, reflect one of the mechanisms by which estrogen may orchestrate learning and memory. In fact, the ACh system is also activated by estrogen in cerebral areas that are important for memory, such as the basal forebrain and its ACh-containing projections to the HF and frontal cortex (Gibbs et al, 2004; Luine, 2008).

Even though gonadal hormones influence cognition, these hormone-induced changes are not large (Luine, 2008), and they are reported especially when function is compromised by aging or lesions (Gulinello et al, 2006; Scharfman et al, 2007) however, they do not improve all the different aspects of cognition such as, for example, acquisition during memory processes (Dohanich, 2002; Luine, 2007).

Rodents have been evaluated in different tasks, utilizing several kinds of mazes, and they rely on diverse reinforcements or contingencies (positive food rewards or aversive electric shocks) for the learning phase, and the tasks measure different kinds of memory, such as spatial memory, which requires the establishment of relationships between distant cues in the environment and the reinforcement site (Gasbarri et al, 2009). Other tasks use visual memory, based on visual associations. Nonetheless, many studies show positive effects of

Estrogen Influences on Cognition 169

2003). ERT relieves several menopausal symptoms, but whether its benefits include protection of cognitive functions is still controversial (LeBlanc et al, 2007; Tivis et al, 2001). In recent years, considerable progress has been made towards specifying the neural

Data from OVX rats treated with estrogen, compared to OVX untreated controls, showed improvements in performance of some tasks, including those require spatial WM, such as the RM (Daniel et al, 1997; Fader et al, 1999) and a 2-choice WM task (O'Neal et al, 1996) and impairments in spatial reference memory tests, such as the MWM (Warren & Juraska, 1997). Estrogen replacement therapy (ERT) enhances spatial WM performance both on MWM and RM (Bimonte & Denenberg, 1999; Fader et al, 1999), confirming previous evidence that estrogen selectively improves performance on tasks depending on WM (Daniel et al, 1997; O'Neal et al, 1996). In fact, estrogen treatment improved WM performance during maze acquisition, without affecting reference memory performance; scopolamine treatment impaired WM, but not reference memory, while estrogen prevented the impairment of WM by scopolamine. A recent paper reported substantial sex differences in the effects of gonadectomy and hormone replacement on spatial working and reference memory in male and female rats (Gibbs & Jognson, 2008). An interesting direction of this field is the idea that

Furthermore, ERT in both physiologically low and moderate doses improved the capability of ovariectomized rats to handle increasing amounts of WM information, when the demand on an animal's WM system was restricted to one to four elements of information (Bimonte & Denenberg, 1999). However, when the demand on the WM system was increased to six elements of information, ERT in physiologically moderate doses provided the maximum

Moreover, it was reported that estrogen can prevent deficits in spatial WM induced by neurotoxin treatments aimed to mimic the pathology of early AD (Hruska & Dohanich, 2007). Cholinergic and HF systems are closely related to learning and memory processes (Hasselmo, 2006), and it can be predicted that estrogen has its most profound effect on HFdependent cognitive functions such as learning and memory. In fact, estrogen enhances ACh function and the synthesis of ACh in basal forebrain and the Ach neurons projecting to the HF and cortex (Hasselmo, 2006), (Singh et al, 1994), and mediates dendritic spine density in the hippocampal CA1 region (Li et al, 2004; Wallace et al, 2007). The HF and adjacent anatomically related cortex play a crucial role in the explicit encoding and consolidation of verbal and nonverbal information into short-term memory, in humans (Squire, 2004). It has been speculated that estrogen activity in HF might underlie the effects of ERT on memory in postmenopausal women (Maki & Resnick, 2000; Maki, 2005). Estrogen receptors, as well as estradiol-concentrating neurons, were detected in the HF and entorhinal cortex of rodents (Prange-Kiel & Rune, 2006). Circulating estrogens have quantifiable effects on neurotransmitter activities in HF where, for example, a low estrogen state increases serotonin (5-HT) transporter activity in the HF, despite an apparent reduction in 5-HT transporter density (Bertrand, 2005); moreover, a regulation of NMDA and GABA receptors has also been reported (Jelks, 2007; Rudick & Woolley, 2001) Estradiol administration in OVX rats produces increased ChAT activity and high-affinity choline uptake in CA1 field (Singh et al, 1994). Even though research has mainly focused on the medial temporal lobe areas, they do not represent the only neuroanatomical regions involved in human memory. In fact, the PFC mediates a number of cognitive processes contributing to memory function, particularly WM which is strongly related to the PFC in both humans and nonhuman

mechanisms underlying WM in humans (Baddeley, 1998; Repovs & Baddeley , 2006).

estrogens may influence learning strategy, independent from memory.

benefit, even beyond that of intact females.

estradiol on cognition (Dohanich, 2002) . Spatial memory, which is dependent on the HF, has been extensively evaluated using the radial arm maze (RM) and MWM; studies conducted in OVX subjects show enhancements in performance during the acquisition (Dohanich, 2002; Luine et al. 1998) but, after learning how to solve the task (reference memory), estradiol no longer enhances performance (Fader et al, 1999; Luine et al. 1998) and could even impair spatial memory, although the data are not conclusive (Dohanichn 2002). Consistent estrogen-related improvements are reported in studies utilizing spatial tasks for the evaluation of working memory (WM) (Daniel & Dohanich, 2001; Luine, 2008; Sandstrom & Williams, 2001, 2004; Scharfman et al, 2007) defined as the ability to retain information in the face of potentially interfering distraction, in order to guide behavior and make a response (Baddeley, 1992, 1998).

Results of studies, assessing hormonal effects on learning and memory, evidence the importance that context and or experience can have on performance, and these considerations may account, at least in part, for some inconsistency in the literature. Therefore, it is hypothesized that stress experienced during task performance may interfere with estrogen enhancements of some spatial tasks (Englemann et al, 2006; Frick et al, 2004). In addition, extensive handling, housing conditions, or environmental enrichment can also mitigate hormonal effects on other spatial tasks (Gresack et al, 2007; Rubinow, et al 2004) Taking into account that cognition represents a complex, multidimensional set of higherorder functions that are sub-served by specific, yet inter-related, cerebral areas, the intervention of other stimuli on the effects of estrogen is not unexpected. It is interesting to note that more recent research, evidencing consistent estrogen-related improvements of memory, use tasks evaluating working or short-term memory, tap into higher order memory or executive function, and also rely on cortical integration with HF fields (Ennaceur, et al 1997; Mumby et al, 2002). In addition, subjects are not exposed to stressful circumstances or negative reinforcers during the task. Therefore, recognition memory tests, where subjects have to discriminate between familiar and unfamiliar objects or objects in familiar or unfamiliar locations, appear to be quite consistently improved by estrogen and its agonists in OVX rats ( Luine, 2008) or mice (Fernandez & Frick, 2004; Li et al, 2004).

In agreement with OVX models, pro-estrous rats evidenced better recognition memory compared to rats in a different phase of the estrous cycle (Frye et al 2007; Walf & Frye, 2006) and mice show better spatial memory in pro-estrous (Frick et al, 2001). However, rats in proestrous phase are often impaired during acquisition (Bowman et al, 2001; Frye, 1995; Warren & Juraska, 1997). Other researchers did not show modifications over the cycle (Berry et al, 1997; Stackman et al, 1997) this inconsistency could be explaining taking into account that they evaluated reference memory, which seems to be insensitive to hormones after acquisition.

#### **3.1.1 Estrogen and working memory**

As evidenced by research assessing performances across the estrous and menstrual cycles, ovarian hormones affect cognition and neural substrates subserving learning and memory, including WM, in both rodents (Craig &Murphy , 2007; Daniel, 2006; Warren & Juraska, 1997) and humans (Bimonte & Denenberg, 1999; Janowski et al, 2000). The decrease of estrogen following ovariectomy or menopause enhances the risk of diseases, such as osteoporosis and vasomotor dysfunction (Timins, 2004; Warren & Halpert, 2004), but could also be involved in the development of cognitive impairments (Markou, 2007; Sherwin,

estradiol on cognition (Dohanich, 2002) . Spatial memory, which is dependent on the HF, has been extensively evaluated using the radial arm maze (RM) and MWM; studies conducted in OVX subjects show enhancements in performance during the acquisition (Dohanich, 2002; Luine et al. 1998) but, after learning how to solve the task (reference memory), estradiol no longer enhances performance (Fader et al, 1999; Luine et al. 1998) and could even impair spatial memory, although the data are not conclusive (Dohanichn 2002). Consistent estrogen-related improvements are reported in studies utilizing spatial tasks for the evaluation of working memory (WM) (Daniel & Dohanich, 2001; Luine, 2008; Sandstrom & Williams, 2001, 2004; Scharfman et al, 2007) defined as the ability to retain information in the face of potentially interfering distraction, in order to guide behavior and make a

Results of studies, assessing hormonal effects on learning and memory, evidence the importance that context and or experience can have on performance, and these considerations may account, at least in part, for some inconsistency in the literature. Therefore, it is hypothesized that stress experienced during task performance may interfere with estrogen enhancements of some spatial tasks (Englemann et al, 2006; Frick et al, 2004). In addition, extensive handling, housing conditions, or environmental enrichment can also mitigate hormonal effects on other spatial tasks (Gresack et al, 2007; Rubinow, et al 2004) Taking into account that cognition represents a complex, multidimensional set of higherorder functions that are sub-served by specific, yet inter-related, cerebral areas, the intervention of other stimuli on the effects of estrogen is not unexpected. It is interesting to note that more recent research, evidencing consistent estrogen-related improvements of memory, use tasks evaluating working or short-term memory, tap into higher order memory or executive function, and also rely on cortical integration with HF fields (Ennaceur, et al 1997; Mumby et al, 2002). In addition, subjects are not exposed to stressful circumstances or negative reinforcers during the task. Therefore, recognition memory tests, where subjects have to discriminate between familiar and unfamiliar objects or objects in familiar or unfamiliar locations, appear to be quite consistently improved by estrogen and its agonists in OVX rats ( Luine, 2008) or mice (Fernandez & Frick, 2004; Li et al, 2004). In agreement with OVX models, pro-estrous rats evidenced better recognition memory compared to rats in a different phase of the estrous cycle (Frye et al 2007; Walf & Frye, 2006) and mice show better spatial memory in pro-estrous (Frick et al, 2001). However, rats in proestrous phase are often impaired during acquisition (Bowman et al, 2001; Frye, 1995; Warren & Juraska, 1997). Other researchers did not show modifications over the cycle (Berry et al, 1997; Stackman et al, 1997) this inconsistency could be explaining taking into account that they evaluated reference memory, which seems to be insensitive to hormones after

As evidenced by research assessing performances across the estrous and menstrual cycles, ovarian hormones affect cognition and neural substrates subserving learning and memory, including WM, in both rodents (Craig &Murphy , 2007; Daniel, 2006; Warren & Juraska, 1997) and humans (Bimonte & Denenberg, 1999; Janowski et al, 2000). The decrease of estrogen following ovariectomy or menopause enhances the risk of diseases, such as osteoporosis and vasomotor dysfunction (Timins, 2004; Warren & Halpert, 2004), but could also be involved in the development of cognitive impairments (Markou, 2007; Sherwin,

response (Baddeley, 1992, 1998).

acquisition.

**3.1.1 Estrogen and working memory** 

2003). ERT relieves several menopausal symptoms, but whether its benefits include protection of cognitive functions is still controversial (LeBlanc et al, 2007; Tivis et al, 2001).

In recent years, considerable progress has been made towards specifying the neural mechanisms underlying WM in humans (Baddeley, 1998; Repovs & Baddeley , 2006).

Data from OVX rats treated with estrogen, compared to OVX untreated controls, showed improvements in performance of some tasks, including those require spatial WM, such as the RM (Daniel et al, 1997; Fader et al, 1999) and a 2-choice WM task (O'Neal et al, 1996) and impairments in spatial reference memory tests, such as the MWM (Warren & Juraska, 1997). Estrogen replacement therapy (ERT) enhances spatial WM performance both on MWM and RM (Bimonte & Denenberg, 1999; Fader et al, 1999), confirming previous evidence that estrogen selectively improves performance on tasks depending on WM (Daniel et al, 1997; O'Neal et al, 1996). In fact, estrogen treatment improved WM performance during maze acquisition, without affecting reference memory performance; scopolamine treatment impaired WM, but not reference memory, while estrogen prevented the impairment of WM by scopolamine. A recent paper reported substantial sex differences in the effects of gonadectomy and hormone replacement on spatial working and reference memory in male and female rats (Gibbs & Jognson, 2008). An interesting direction of this field is the idea that estrogens may influence learning strategy, independent from memory.

Furthermore, ERT in both physiologically low and moderate doses improved the capability of ovariectomized rats to handle increasing amounts of WM information, when the demand on an animal's WM system was restricted to one to four elements of information (Bimonte & Denenberg, 1999). However, when the demand on the WM system was increased to six elements of information, ERT in physiologically moderate doses provided the maximum benefit, even beyond that of intact females.

Moreover, it was reported that estrogen can prevent deficits in spatial WM induced by neurotoxin treatments aimed to mimic the pathology of early AD (Hruska & Dohanich, 2007). Cholinergic and HF systems are closely related to learning and memory processes (Hasselmo, 2006), and it can be predicted that estrogen has its most profound effect on HFdependent cognitive functions such as learning and memory. In fact, estrogen enhances ACh function and the synthesis of ACh in basal forebrain and the Ach neurons projecting to the HF and cortex (Hasselmo, 2006), (Singh et al, 1994), and mediates dendritic spine density in the hippocampal CA1 region (Li et al, 2004; Wallace et al, 2007). The HF and adjacent anatomically related cortex play a crucial role in the explicit encoding and consolidation of verbal and nonverbal information into short-term memory, in humans (Squire, 2004). It has been speculated that estrogen activity in HF might underlie the effects of ERT on memory in postmenopausal women (Maki & Resnick, 2000; Maki, 2005). Estrogen receptors, as well as estradiol-concentrating neurons, were detected in the HF and entorhinal cortex of rodents (Prange-Kiel & Rune, 2006). Circulating estrogens have quantifiable effects on neurotransmitter activities in HF where, for example, a low estrogen state increases serotonin (5-HT) transporter activity in the HF, despite an apparent reduction in 5-HT transporter density (Bertrand, 2005); moreover, a regulation of NMDA and GABA receptors has also been reported (Jelks, 2007; Rudick & Woolley, 2001) Estradiol administration in OVX rats produces increased ChAT activity and high-affinity choline uptake in CA1 field (Singh et al, 1994). Even though research has mainly focused on the medial temporal lobe areas, they do not represent the only neuroanatomical regions involved in human memory. In fact, the PFC mediates a number of cognitive processes contributing to memory function, particularly WM which is strongly related to the PFC in both humans and nonhuman

Estrogen Influences on Cognition 171

order to verify the hypothesis that the WM system is responsive to estrogen in women, Maki et al. (Maki, 2005) designed a study evaluating, in a group of postmenopausal women, two measures, one verbal and one spatial, which strongly recruit the WM system (Digit Ordering, Spatial WM task). Their findings confirmed the hypothesis that estrogen is active

In agreement with the above findings, evidence exists showing the activation of PFC during the performance of WM tasks (Badre D, Wagner, 2007; Petrides et al, 1993) and decrease of WM with increasing age (Grady & Craik, 2000) . The integrity of both the PFC and its complex neural circuitry, which consolidates input from various modalities via cortical, subcortical, and limbic connections, are critical to intact executive functions, an amalgamation of cognitive processes that includes WM, besides directed attention, response inhibition, dual task coordination, cognitive set switching, and behavioral monitoring. Dopaminergic and serotonergic brain stem afferents to PFC (Jakob & Goldman-Rakic, 1998) modulate the excitability of prefrontal pyramidal neurons. Experimental reduction of prefrontal dopamine in rhesus monkeys and naturally occurring loss of dopaminergic neurons in Parkinson's disease are associated with deficits in WM (Gotham et al, 1988). The dopaminergic D2 receptor agonist bromocriptine improves WM (Luciana et al, 1991) while D2 antagonist raclopride had a minor inhibitory effect (Williams & Goldman-Rakic, 1995). Ovarian steroids are powerful modulators of the dopaminergic neurotransmission. In monkeys ovariectomy reduces, while subsequent estrogen and progesterone replacement restores, the density of axons immunoreactive for tyrosine hydroxylase in the dorsolateral PFC (Kritzer & Kohama, 1998) . Ovariectomy also decreases the density of axons immunoreactive for ChAT and increases the density of fibers immunoreactive for dopamine β–hydroxylase. ERT alone attenuates these effects (Kritzer & Kohama, 1999) estradiol also decreases monoamine oxidase (MAO), involved in the degradation of dopamine (McEwen,

**3.1.1.1 Working memory for emotional facial expressions across the menstrual cycle in** 

Facial expressions represent non-verbal communicative displays that are critical in social cognition, allowing quick transmission of valence information to cospecifics concerning objects or environments (Blair et al, 1999). In particular, humans and non-human primates use facial expressions to communicate their emotional state. This communication can be reflexive, as situations may induce emotions that are spontaneously expressed on the face. In other cases, particularly in humans, facial expressions may consist in volitional signals with the aim of communicating, and not reflecting, the real emotional state of the subject (Ekman, 1993) . Six basic emotions - happiness, sadness, anger, fear, disgust and surprise and their corresponding facial expressions are recognized across different cultures (Ekman & Friesen, 1971). Imaging studies showed that different cerebral areas are activated during the processing of different, distinct emotions (Blair et al, 1999). It was also reported that not only subcortical areas, such as amygdala or basal ganglia, but also cortical areas, mainly PFC, cingulate cortex, and temporal cortices, are essential in emotion processing (Blair et al, 1999; Northoff et al, 2000). Many studies on emotion perception in faces have been focused on the identification of the cerebral regions, whose damage causes emotion perception deficits (Adolphs, 2002) . This facial emotion recognition deficit appears to be, at least in part, related to a more general problem in cognitive functions including the categorisation, discrimination and identification of facial stimuli, as well as deficits in other cognitive

within PFC and it can influence functions dependent on this region, like WM.

2002).

**young women.** 

primates. In humans, WM represents the basis for many cognitive functions, including reasoning, reading comprehension, and mental calculations (Baddeley, 1998). Both non verbal (Owen et al, 1996) and verbal (Petrides et al, 1993); stimuli were utilized in experimental tasks with a relevant WM component. The important role of the PFC in WM was demonstrated after lesion and electrophysiological techniques in monkeys (Funahashi et al, 1993; Petrides, 1995) functional neuroimaging techniques in healthy human volunteers (Jonides et al, 1993; Owen et al, 1996); and localized cortical excisions in human neurological patients (Owen et al, 1995). Taking into consideration that, by definition, WM tasks intrinsically involve both temporary retention of verbal or visual information and its active manipulation, some research have clarified that the requirement for active manipulation during WM tasks specifically recruits activity in dorsolateral PFC (Owen et al, 1995; Petrides, 1995; Postle et al, 1999). By contrast, passive storage processes seem to depend on more posterior brain areas, as evidenced by deficits in the immediate span for spatial or verbal information, in patients with lesion of parietal or perisylvian cortex (Milner, 1971) and by changes in functional cerebral activity in parietal and temporal regions of healthy volunteers, during performance of neuroimaging tasks that emphasize passive storage of information (Postle et al, 1999). Therefore, the dorsolateral PFC, as part of the WM system, plays a critical role in mediating the control processes required for the active manipulation, or selective utilization of items contained in WM.

Several lines of research raised the possibility of estrogen's modulating effect on the PFC (Joffe et al, 2006). In particular, analysis of human brain specimens has revealed that in PFC estradiol concentrations was approximately 2 times higher than in temporal cortex or 7 times higher than in HF, showing that the PFC is a principal target for estrogen in the adult female brain (Bixo et al, 1995). Animal studies reported that estrogen influences the activity of several neurotransmitter systems in the PFC. For example, a 56% reduction in ChAT and a 24% reduction in high affinity choline uptake in the frontal cortex of female rats at 28 weeks post-OVX were found; this effect was prevented or reversed in rats treated with ERT (Singh et al, 1994) Estrogen may also regulate neurotransmission in the PFC of nonhuman primates. Remarkable increases in axons immunoreaction for dopamine β-hydroxylase and 5-HT and reductions in the density of axons immunoreactive for ChAT and tyrosine hydroxylase were observed in the dorsolateral PFC of adult rhesus monkeys, following OVX (Kritzer & Kohama, 1998, 1999) In OVX monkeys treated with estrogen, the density of labeling was similar to hormonally intact controls, suggesting that estrogen plays a role in maintaining cholinergic, noradrenergic, serotonergic, and dopaminergic activity in the PFC. In addition, in humans, neuroimaging studies using positron emission tomography (PET) (Berman et al, 1997) or functional magnetic resonance imaging (fMRI) (Shaywitz et al, 1999) have evidenced systematic differences in patterns of task-induced brain activation in PFC, connected to differences in women's estrogen status (Roberts et al, 1997). A behavioral study conducted on rhesus monkeys showed that menopausal and postmenopausal females, compared to age-matched but premenopausal females, exhibited an impairment of performance on the WM delayed response task, which is commonly used to assess PFC dysfunction in nonhuman primates. Taken together, the neuroendocrine and behavioral data supply evidence to suggest that estrogen is active in the PFC. In such a case, estrogen could modulate cognitive functions mediated by the PFC in women.

Taking into account that the dorsolateral PFC is one of the areas of the frontal cortex where estrogen activity was demonstrated (Maki, 2005), this steroid hormone might contribute to WM function by modulating information processing in the PFC (Duff & Hampson, 2000). In

primates. In humans, WM represents the basis for many cognitive functions, including reasoning, reading comprehension, and mental calculations (Baddeley, 1998). Both non verbal (Owen et al, 1996) and verbal (Petrides et al, 1993); stimuli were utilized in experimental tasks with a relevant WM component. The important role of the PFC in WM was demonstrated after lesion and electrophysiological techniques in monkeys (Funahashi et al, 1993; Petrides, 1995) functional neuroimaging techniques in healthy human volunteers (Jonides et al, 1993; Owen et al, 1996); and localized cortical excisions in human neurological patients (Owen et al, 1995). Taking into consideration that, by definition, WM tasks intrinsically involve both temporary retention of verbal or visual information and its active manipulation, some research have clarified that the requirement for active manipulation during WM tasks specifically recruits activity in dorsolateral PFC (Owen et al, 1995; Petrides, 1995; Postle et al, 1999). By contrast, passive storage processes seem to depend on more posterior brain areas, as evidenced by deficits in the immediate span for spatial or verbal information, in patients with lesion of parietal or perisylvian cortex (Milner, 1971) and by changes in functional cerebral activity in parietal and temporal regions of healthy volunteers, during performance of neuroimaging tasks that emphasize passive storage of information (Postle et al, 1999). Therefore, the dorsolateral PFC, as part of the WM system, plays a critical role in mediating the control processes required for the active manipulation,

Several lines of research raised the possibility of estrogen's modulating effect on the PFC (Joffe et al, 2006). In particular, analysis of human brain specimens has revealed that in PFC estradiol concentrations was approximately 2 times higher than in temporal cortex or 7 times higher than in HF, showing that the PFC is a principal target for estrogen in the adult female brain (Bixo et al, 1995). Animal studies reported that estrogen influences the activity of several neurotransmitter systems in the PFC. For example, a 56% reduction in ChAT and a 24% reduction in high affinity choline uptake in the frontal cortex of female rats at 28 weeks post-OVX were found; this effect was prevented or reversed in rats treated with ERT (Singh et al, 1994) Estrogen may also regulate neurotransmission in the PFC of nonhuman primates. Remarkable increases in axons immunoreaction for dopamine β-hydroxylase and 5-HT and reductions in the density of axons immunoreactive for ChAT and tyrosine hydroxylase were observed in the dorsolateral PFC of adult rhesus monkeys, following OVX (Kritzer & Kohama, 1998, 1999) In OVX monkeys treated with estrogen, the density of labeling was similar to hormonally intact controls, suggesting that estrogen plays a role in maintaining cholinergic, noradrenergic, serotonergic, and dopaminergic activity in the PFC. In addition, in humans, neuroimaging studies using positron emission tomography (PET) (Berman et al, 1997) or functional magnetic resonance imaging (fMRI) (Shaywitz et al, 1999) have evidenced systematic differences in patterns of task-induced brain activation in PFC, connected to differences in women's estrogen status (Roberts et al, 1997). A behavioral study conducted on rhesus monkeys showed that menopausal and postmenopausal females, compared to age-matched but premenopausal females, exhibited an impairment of performance on the WM delayed response task, which is commonly used to assess PFC dysfunction in nonhuman primates. Taken together, the neuroendocrine and behavioral data supply evidence to suggest that estrogen is active in the PFC. In such a case, estrogen

or selective utilization of items contained in WM.

could modulate cognitive functions mediated by the PFC in women.

Taking into account that the dorsolateral PFC is one of the areas of the frontal cortex where estrogen activity was demonstrated (Maki, 2005), this steroid hormone might contribute to WM function by modulating information processing in the PFC (Duff & Hampson, 2000). In order to verify the hypothesis that the WM system is responsive to estrogen in women, Maki et al. (Maki, 2005) designed a study evaluating, in a group of postmenopausal women, two measures, one verbal and one spatial, which strongly recruit the WM system (Digit Ordering, Spatial WM task). Their findings confirmed the hypothesis that estrogen is active within PFC and it can influence functions dependent on this region, like WM.

In agreement with the above findings, evidence exists showing the activation of PFC during the performance of WM tasks (Badre D, Wagner, 2007; Petrides et al, 1993) and decrease of WM with increasing age (Grady & Craik, 2000) . The integrity of both the PFC and its complex neural circuitry, which consolidates input from various modalities via cortical, subcortical, and limbic connections, are critical to intact executive functions, an amalgamation of cognitive processes that includes WM, besides directed attention, response inhibition, dual task coordination, cognitive set switching, and behavioral monitoring. Dopaminergic and serotonergic brain stem afferents to PFC (Jakob & Goldman-Rakic, 1998) modulate the excitability of prefrontal pyramidal neurons. Experimental reduction of prefrontal dopamine in rhesus monkeys and naturally occurring loss of dopaminergic neurons in Parkinson's disease are associated with deficits in WM (Gotham et al, 1988). The dopaminergic D2 receptor agonist bromocriptine improves WM (Luciana et al, 1991) while D2 antagonist raclopride had a minor inhibitory effect (Williams & Goldman-Rakic, 1995). Ovarian steroids are powerful modulators of the dopaminergic neurotransmission. In monkeys ovariectomy reduces, while subsequent estrogen and progesterone replacement restores, the density of axons immunoreactive for tyrosine hydroxylase in the dorsolateral PFC (Kritzer & Kohama, 1998) . Ovariectomy also decreases the density of axons immunoreactive for ChAT and increases the density of fibers immunoreactive for dopamine β–hydroxylase. ERT alone attenuates these effects (Kritzer & Kohama, 1999) estradiol also decreases monoamine oxidase (MAO), involved in the degradation of dopamine (McEwen, 2002).

#### **3.1.1.1 Working memory for emotional facial expressions across the menstrual cycle in young women.**

Facial expressions represent non-verbal communicative displays that are critical in social cognition, allowing quick transmission of valence information to cospecifics concerning objects or environments (Blair et al, 1999). In particular, humans and non-human primates use facial expressions to communicate their emotional state. This communication can be reflexive, as situations may induce emotions that are spontaneously expressed on the face. In other cases, particularly in humans, facial expressions may consist in volitional signals with the aim of communicating, and not reflecting, the real emotional state of the subject (Ekman, 1993) . Six basic emotions - happiness, sadness, anger, fear, disgust and surprise and their corresponding facial expressions are recognized across different cultures (Ekman & Friesen, 1971). Imaging studies showed that different cerebral areas are activated during the processing of different, distinct emotions (Blair et al, 1999). It was also reported that not only subcortical areas, such as amygdala or basal ganglia, but also cortical areas, mainly PFC, cingulate cortex, and temporal cortices, are essential in emotion processing (Blair et al, 1999; Northoff et al, 2000). Many studies on emotion perception in faces have been focused on the identification of the cerebral regions, whose damage causes emotion perception deficits (Adolphs, 2002) . This facial emotion recognition deficit appears to be, at least in part, related to a more general problem in cognitive functions including the categorisation, discrimination and identification of facial stimuli, as well as deficits in other cognitive

Estrogen Influences on Cognition 173

Seven adult capuchin monkeys were tested with a computer system and touch screen. Geometric figures (control) and the co-specific faces pictures were used as stimuli. The subjects obtained a similar performance to positive, negative and neutral pictures. However, the monkeys performed above the upper confidence limits around chance to all kinds of stimulus showing that they are able to learn the tests using emotional faces. Furthermore, the capuchin monkeys had much better performance when using geometric figures

On a whole, our results show that capuchin monkeys were able to perform this new WM task, thus indicating the possible usefulness of applying the paradigm utilized in this study

One of the most interesting research fields in women's health of the last decade includes the growing appreciation that estrogen plays relevant neurotrophic and neuroprotective roles during adulthood. This amplifies the relevance of the potential impact of the prolonged post-menopausal hypoestrogenic state on learning and memory processes and the potential increased vulnerability of ageing women to brain injury and neurodegenerative diseases. The longer female life expectancy has implied that nowadays women live one-third of their lives beyond ending of their ovarian function, increasing the need for new therapeutic strategies to facilitate successful aging (defined as low probability of disease), high cognitive and physical abilities, and active engagement in life. Taking into account that changes in the ageing nervous system are subtle, they could be reversed and cognitive performance may

The ematic concentration of estrogens decreases with age and the post-menopause low values of estrogens are often followed by an acceleration of the age effects on cognition. Cognitive decline during aging affect memory abilities, attention, and speed of information

Even though several cognitive functions seem to be unaltered in normal aging, age-related impairments are mainly evident in tasks implying free or cued recall or WM (Small et al, 1999). Although verbal memory has been reported to be the cognitive function most deeply affected with increasing age (Marquis et al, 2002; Rabbitt & Lowe, 2000) other cognitive domains such as attention (Stankov, 1988) visual perception, and verbal fluency (Ashman, 1999) are also influenced. Thus, the attempt to delay or prevent the cognitive impairment occurring with normal aging is an important goal to protect the quality of life for women during the latter one third of their lifespan. Because ERs are present in both the HF and frontal lobes which subserve verbal memory, WM and retrieval, we can hypothesize that estrogen might play an important protective role against the decline in these cognitive functions, occurring with normal aging. Therefore, researchers have tried to verify if the estrogen administration to women at the beginning or during menopause would protect

During the past few decades, data from basic neuroscience and from animal and human studies have suggested that ERT given to postmenopausal women might protect against specific cognitive declines occurring with normal aging. On the other hand, the numerous inconsistencies in this body of evidence point to the possibility that there are contingencies which modify the supposed neuroprotective effects of ERT on cognitive aging (Sherwin &

against cognitive impairments that normally take place with increasing age.

to investigate emotional memory in non-human primates (Abreu et al, 2006).

compared with the co-specific pictures.

**4. Estrogen and the aging brain** 

be improved by pharmacological treatments.

processing (Sherwin & Henry, 2008).

Henry, 2008).

processes, such as WM, which are impaired in the psychiatric and neurological damages (Addington & Addington, 1998; Kee et al, 1998).

Physiological fluctuations in ovarian hormones across the menstrual cycle allow for noninvasive studies of the effects of estrogen on cognition in young women and underlie a reliable pattern of cognitive change across the menstrual cycle (Maki et al, 2002).

The cognitive performance in a WM task for emotional facial expressions, using the six basic emotions (Ekman &Friesen, 1971) as stimuli in the DMTS, was evaluated in young women in the different phases of the menstrual cycle, in order to point out possible differences related to the physiological hormonal fluctuations (Gasbarri et al, 2008, 2009). Our findings suggest that high levels of estradiol in the follicular phase could have a negative effect on delayed matching-to-sample WM task, using stimuli with emotional valence. Moreover, in the follicular phase, compared to the menstrual phase, the percent of errors was significantly higher for the emotional facial expressions of sadness and disgust (Gasbarri et al, 2008, 2009) The evaluation of the response times (time employed to answer) for each facial expression with emotional valence showed a significant difference between follicular and luteal in reference to the emotional facial expression of sadness (Gasbarri et al, 2008, 2009). Our results show that high levels of estradiol in the follicular phase could impair the performance of WM. However, this effect is specific to selective facial expressions suggesting that, across the phases of the menstrual cycle, in which conception risk is high, women could give less importance to the recognition of the emotional facial expressions of sadness and disgust. This study is in agreement with research conducted on non-human primates, showing that fluctuations of ovarian hormones across the menstrual cycle influence a variety of social and cognitive behaviors. For example, female rhesus monkeys exhibit heightened interest for males and enhanced agonistic interactions with other females during periods of high estrogen level (Lacreuse et al, 2007).

Moreover, our data could also represent a useful tool for investigating emotional disturbances linked to menstrual cycle phases and menopause in women.

#### **3.1.1.2 Working memory for emotional facial expressions in capuchin monkeys**

Non-human primates represent important and relevant models for the study of emotional face processing, because they share several cognitive and physiological characteristics with humans. The behavioral evidence includes similarities in innate action patterns such as body movements and communication signals, as well as highly flexible behavioral tactics and clever problem-solving strategies (Preuschoft, 2000) . The capuchin monkey (*Cebus apella*) has been the focus of various researches due to its behavioral similarities with apes. Moreover, capuchins exhibit a rich repertoire of facial expressions and body postures, which convey an array of messages to co-specifics about their internal state (Fragaszy et al, 2004); furthermore, they display tool-using capacities, and can readily solve the WM tasks, such as DNMS and concurrent discrimination learning task (Resende et al, 2003; Tavares & Tomaz, 2002).

Capuchin monkeys have well-developed facial musculature mobility, which allows considerable expressive variability, and they also have excellent visual acuity for discerning signals by others. However, most of the visual signals of capuchin monkeys are accompanied by vocalizations and associated context. In general, movement and body expression are important to understand emotional valence.

In a previous study we developed a pool of 384 pictures of capuchin monkey (*Cebus apella*) faces, classified according to emotional valence (positive/ pleasant, negative/unpleasant and neutral/indifferent), to examine whether WM can benefit from the emotional content of visual stimuli in a delayed non-matching to sample task (DNMTS) (Abreu et al, 2006).

processes, such as WM, which are impaired in the psychiatric and neurological damages

Physiological fluctuations in ovarian hormones across the menstrual cycle allow for noninvasive studies of the effects of estrogen on cognition in young women and underlie a

The cognitive performance in a WM task for emotional facial expressions, using the six basic emotions (Ekman &Friesen, 1971) as stimuli in the DMTS, was evaluated in young women in the different phases of the menstrual cycle, in order to point out possible differences related to the physiological hormonal fluctuations (Gasbarri et al, 2008, 2009). Our findings suggest that high levels of estradiol in the follicular phase could have a negative effect on delayed matching-to-sample WM task, using stimuli with emotional valence. Moreover, in the follicular phase, compared to the menstrual phase, the percent of errors was significantly higher for the emotional facial expressions of sadness and disgust (Gasbarri et al, 2008, 2009) The evaluation of the response times (time employed to answer) for each facial expression with emotional valence showed a significant difference between follicular and luteal in reference to the emotional facial expression of sadness (Gasbarri et al, 2008, 2009). Our results show that high levels of estradiol in the follicular phase could impair the performance of WM. However, this effect is specific to selective facial expressions suggesting that, across the phases of the menstrual cycle, in which conception risk is high, women could give less importance to the recognition of the emotional facial expressions of sadness and disgust. This study is in agreement with research conducted on non-human primates, showing that fluctuations of ovarian hormones across the menstrual cycle influence a variety of social and cognitive behaviors. For example, female rhesus monkeys exhibit heightened interest for males and enhanced agonistic interactions with other females

Moreover, our data could also represent a useful tool for investigating emotional

Non-human primates represent important and relevant models for the study of emotional face processing, because they share several cognitive and physiological characteristics with humans. The behavioral evidence includes similarities in innate action patterns such as body movements and communication signals, as well as highly flexible behavioral tactics and clever problem-solving strategies (Preuschoft, 2000) . The capuchin monkey (*Cebus apella*) has been the focus of various researches due to its behavioral similarities with apes. Moreover, capuchins exhibit a rich repertoire of facial expressions and body postures, which convey an array of messages to co-specifics about their internal state (Fragaszy et al, 2004); furthermore, they display tool-using capacities, and can readily solve the WM tasks, such as DNMS and

reliable pattern of cognitive change across the menstrual cycle (Maki et al, 2002).

(Addington & Addington, 1998; Kee et al, 1998).

during periods of high estrogen level (Lacreuse et al, 2007).

expression are important to understand emotional valence.

disturbances linked to menstrual cycle phases and menopause in women.

**3.1.1.2 Working memory for emotional facial expressions in capuchin monkeys** 

concurrent discrimination learning task (Resende et al, 2003; Tavares & Tomaz, 2002).

Capuchin monkeys have well-developed facial musculature mobility, which allows considerable expressive variability, and they also have excellent visual acuity for discerning signals by others. However, most of the visual signals of capuchin monkeys are accompanied by vocalizations and associated context. In general, movement and body

In a previous study we developed a pool of 384 pictures of capuchin monkey (*Cebus apella*) faces, classified according to emotional valence (positive/ pleasant, negative/unpleasant and neutral/indifferent), to examine whether WM can benefit from the emotional content of visual stimuli in a delayed non-matching to sample task (DNMTS) (Abreu et al, 2006).

Seven adult capuchin monkeys were tested with a computer system and touch screen. Geometric figures (control) and the co-specific faces pictures were used as stimuli. The subjects obtained a similar performance to positive, negative and neutral pictures. However, the monkeys performed above the upper confidence limits around chance to all kinds of stimulus showing that they are able to learn the tests using emotional faces. Furthermore, the capuchin monkeys had much better performance when using geometric figures compared with the co-specific pictures.

On a whole, our results show that capuchin monkeys were able to perform this new WM task, thus indicating the possible usefulness of applying the paradigm utilized in this study to investigate emotional memory in non-human primates (Abreu et al, 2006).
