Prefrontal cortex executive processes affected by stress in health and disease

Abstract

Prefrontal cortical executive functions comprise a number of cognitive capabilities necessary for goal directed behavior and adaptation to a changing environment. Executive dysfunction that leads to maladaptive behavior and is a symptom of psychiatric pathology can be instigated or exacerbated by stress. In this review we survey research addressing the impact of stress on executive function, with specific focus on working memory, attention, response inhibition, and cognitive flexibility. We then consider the neurochemical pathways underlying these cognitive capabilities and, where known, how stress alters them. Finally, we review work exploring potential pharmacological and non-pharmacological approaches that can ameliorate deficits in executive function. Both preclinical and clinical literature indicates that chronic stress negatively affects executive function. Although some of the circuitry and neurochemical processes underlying executive function have been characterized, a great deal is still unknown regarding how stress affects these processes. Additional work focusing on this question is needed in order to make progress on developing interventions that ameliorate executive dysfunction.

1. Introduction

Executive function comprises several top-down cognitive processes that are necessary for everyday adaptive behaviors, such as the need to pay attention and concentrate, plan a course of action, adapt to unforeseen events, or control impulsive behaviors that are not appropriate to the situation (Barnes et al 2011, Diamond 2013, Leh et al 2010, Logue & Gould 2014, Robbins & Arnsten 2009). These cognitive processes, which include working memory, attention, response inhibition and cognitive flexibility, require effort and conscious engagement, are acquired in the course of development, and decline with age. Throughout life, executive function can be challenged by intense or prolonged stress, and the dysregulation of these processes can reduce the quality of life and daily performance of otherwise healthy individuals. Furthermore, it has long been recognized that both acute stressful life events and chronic stressors are strong risk factors for the development of mental illnesses such as mood disorders (Beck 2008, Kessler 1997), anxiety disorders and addictive disorders (Kessler et al 1997). Executive dysfunction is a common symptom of many of these psychiatric conditions that include depression, generalized anxiety disorder (GAD), obsessive-compulsive disorder (OCD), attention-deficit hyperactivity disorder (ADHD), post-traumatic stress disorder (PTSD), and addictive behavior (Baler & Volkow 2006, Carvalho et al 2014, Ferreri et al 2011, Polak et al 2012). Given the consistent evidence discussed below that stress compromises executive abilities in healthy individuals and in animal models, it seems reasonable to infer that stress contributes to the executive dysfunction seen in these conditions. Indeed, an inverse correlation is found between resilience (or the capacity to respond positively to adverse situations) and indices of anxiety and depression (Holden et al 2012, Wingo et al 2010). However, we also must acknowledge that in some cases, executive impairment may be a direct consequence of an underlying disease process independent of stress. In addition, it should also be recognized that executive dysfunction in itself can produce a stressful experience that can then exacerbate psychiatric illness secondarily, as has been shown with depression (Jaeger et al 2006).

A wealth of evidence derived from lesion and inactivation studies supports the idea that prefrontal cortical (PFC) regions (i.e., anterior cingulate, prelimbic, infralimbic, and orbitofrontal cortex in rodents, and Brodmann areas 24b, 32 and 25, as well as the orbital cortex in humans) are essential for optimal executive control (Baier et al 2010, Barbey et al 2013, Colvin et al 2001, Drevets et al 2008, Levens et al 2014, Muller et al 2002). However, the PFC does not work in isolation, and a distributed network of connectivity with other regions such as the hippocampus, amygdala, striatum, and posterior parietal cortex is essential for modulating several aspects of executive function (Holmes & Wellman 2009). Indeed, human fMRI studies of large-scale interconnectivity networks identified the central executive network (CEN) as one of three core neurocognitive networks. Neurocognitive networks are intrinsically coupled brain areas that are systematically co-activated during higher order cognitive tasks. The CEN nodes (the dorso-lateral PFC and the lateral posterior parietal cortex) show strong co-activation during processes requiring executive control, such as working memory and decision-making tasks in goal-oriented behavior (Menon 2011). Thus, it is important to remember that the behavioral outcomes of such processes stem from the coordinated activation and/or inactivation of several brain circuits integrating PFC top-down control. In this review we will focus primarily on the effects of stress on the prefrontal cortex, a major component of executive processes.

We will review the effects of stress on several examples of executive function (working memory, attention, response inhibition and cognitive flexibility), in both humans and animal models. We will then discuss neurotransmitter systems that have been shown to respond to stress and are targets of current therapeutics, namely, the glutamate, GABA, dopamine, norepinephrine and serotonin systems. We will also discuss molecular pathways that have potential as novel therapeutic targets and emergent non-pharmacological approaches to improve executive function in humans.

1.1. Overall impact of stress on executive function

The stress response is a highly conserved process essential for survival under conditions of environmental challenge (McEwen et al 2015, McKlveen et al 2015). Thus, the response to acute stress (i.e. to a temporary challenge to the organism homeostasis, (McEwen 2004, Selye 1973)) rapidly mobilizes the autonomic and neuroendocrine systems, producing a nearly instantaneous release of catecholamines and HPA axis hormones (CRF, ACTH and glucocorticoids), which alter several physiological functions, such as cardiovascular capacity, metabolic resource allocation, and immune activation in order to effectively respond to a threat. Acute catecholamine effects are short-lived, disappearing within an hour; in contrast, glucocorticoid effects can be both rapid (with onset within minutes after the stimulus) and long-lasting. The long-term effects develop over the course of several hours, and comprise transcriptional effects of activated glucocorticoid receptors (Henckens et al 2010, Henckens et al 2011). The acute stress response also has a strong impact on cognitive function. Acute stress in humans has been shown to activate saliency networks centered around the amygdala, cingulate cortex, hypothalamus, insula, striatum, and locus coeruleus, and is responsible for enhancing sensory gain and environmental scanning, resulting in better performance (Cousijn et al 2010, Oei et al 2012, van Marle et al 2009, van Marle et al 2010). Conversely, processes underlying working memory, problem solving and cognitive flexibility are negatively affected by acute stress (Oei et al 2006, Plessow et al 2011, Plessow et al 2012, Schoofs et al 2008, Schoofs et al 2009, Steinhauser et al 2007). Similar immediate and detrimental effects of acute stress on cognitive flexibility and working memory have also been shown in rodents (Butts et al 2011, George et al 2015, Thai et al 2013). Together, these results are consistent with an adaptive strategy that, for the short term, ensures allocation of resources to cognitive functions that increase sensory hypervigilance, scanning attention, and rapid (but more rigid) behavioral responses, at the expense of high order cognitive engagement. These effects require the actions of catecholamines as well as the rapid effects of glucocorticoids.

The adaptive value of the stress response relies on the rapid resolution of the acute effects through negative feedback mechanisms (Herman et al 2016, Hill & Tasker 2012). Thus, about 1 hour after exposure to a stressful stimulus, when catecholamine levels are low but glucocorticoid levels are still elevated, the neurocognitive processes associated with the salience network weaken (Henckens et al 2012, Henckens et al 2010), while working memory and the ability to perform cognitive tasks improve, along with reduced anxiety behavior. In the aftermath of an acute stress exposure, executive processes improve and these effects have been shown to correlate with genomic corticosteroid actions (Henckens et al 2011, Het & Wolf 2007, Maheu et al 2005, Oei et al 2009, Putman et al 2007).

However, when the physiological response to stress fails to return to homeostasis after acute activation, or with prolonged and/or intense stress exposure, (conditions that are the hallmarks of chronic stress) the consequences for the organism, both physiological and cognitive, can be detrimental. This is evidenced in non-clinical human cohorts where chronic stress levels which elicited subjective cognitive complaint, i.e., perceived difficulties with concentration, memory and decision making, were associated with poor performance in tests of attentional shifting and working memory (Stenfors et al 2013). In another study of healthy individuals, chronic stress biased decision-making strategies towards habitual responding (Soares et al 2012). Patients with mood and anxiety disorders also display deficits in executive function. In particular, individuals with major depressive disorder show impairments in working memory (Landro et al 2001, Rose & Ebmeier 2006) and cognitive flexibility on the Wisconsin card sorting test (Merriam et al 1999). Significant deficits in cognitive flexibility were also found in individuals that developed obsessive-compulsive symptoms after a traumatic stress exposure (Borges et al 2011). Paralleling human studies, chronic stress impairs working memory (Arnsten et al 2012, Barsegyan et al 2010, Mizoguchi et al 2000) as well as cognitive flexibility (Birrell & Brown 2000, Bondi et al 2008, Cerqueira et al 2005a, Floresco et al 1997, Lapiz-Bluhm et al 2009, Liston et al 2006) in animal models.

Deficits in executive function following chronic stress are accompanied by morphological changes in the prefrontal cortex. In rodents, repeated stress and chronic corticosterone administration cause reduction in apical dendrites, debranching of pyramidal neurons, and dendritic spine loss in the medial prefrontal cortex (mPFC) (Cerqueira et al 2005a, Cerqueira et al 2007a, Cerqueira et al 2005b, Cerqueira et al 2007b, Cook & Wellman 2004, Dias-Ferreira et al 2009, Liston et al 2006, Michelsen et al 2007, Radley et al 2006, Radley et al 2004, Silva-Gomez et al 2003). Interestingly, in other brain regions, such as amygdala, orbitofrontal cortex (OFC), and putamen, chronic stress increases dendritic elaboration and produces structural hypertrophy (Dias-Ferreira et al 2009, Liston et al 2006). The differential effects of stress on neurons of the mPFC and the OFC align with data suggesting opposing firing properties of mPFC and OFC neurons (Moghaddam & Homayoun 2008).

Detrimental effects of chronic stress on the morphology and activation patterns in the mPFC have also been documented in humans. In studies employing non-clinical cohorts, individuals exposed to chronic stress show a shift toward automated response patterns during decision-making tasks that correlate with atrophy of the medial prefrontal cortex and the caudate (Soares et al 2012). Conversely, the putamen presented an increase in volume and dendritic arborization following chronic stress (Soares et al 2012). These data complement the preclinical studies reported above (Dias-Ferreira et al 2009), where rats exposed to chronic stress displayed atrophy of the associative network (dorsomedial striatum-prefrontal circuitry) and hypertrophy of the sensorimotor network responsible for automated responses (dorsolateral striatal-prefrontal circuitry). These structural changes in rats were also accompanied by increased biases toward habit behaviors. Combined with the human studies, this work demonstrates stress produces structural and functinoal changes in fronto-striatal circuitry that reduces the ability of an individual to shift from automated behavior to goal-directed behavior and may be maladaptive in situations of change.

Changes in volume, morphology and activation of the prefrontal cortex are also evident in subjects with mood disorders. For example, several prefrontal cortex regions including the anterior cingulate, the orbital cortex and the ventrolateral PFC have decreased grey matter volume in individuals with major depression and bipolar disorder (Drevets et al 2008, Lyoo et al 2004). Anomalies in the recruitment of prefrontal circuitry and patterns of activation in major depression have also been shown (Johnstone et al 2007), with decreases in activation of the dorsolateral PFC (DLPFC) (Drevets 1998) and hyperactivation of the OFC (Biver et al 1994, Drevets et al 1992). Although other factors beside stress may produce structural and functional changes in pathology, the fact that similar alterations are reported in preclinical and non-clinical stress studies suggests that stress contributes to these effects also in pathological states. In summary, the above literature documents a substantial impact of stress on the structure and function of prefrontal cortical neurons that correlate with behavioral impairments in executive function.

2. Methods

This study is a review investigating the effects of stress on working memory, attention, response inhibition, reversal learning and set shifting. For the sections covering the effects of stress on each executive function we conducted two types of search. The first was centered on the effects of stress on working memory, attention, etc., in animal models. The second search was focused on human populations. We sought articles illustrating the deficits in executive function correlating with stress exposure in non-clinical human cohorts as well as deficits observed in disorders for which stress has been shown to play an etiological or exacerbating role (Depression, PTSD, OCD, ADHD). For the “Neurochemical mechanisms and pharmacological target” sections, the combinatorial search included the terms “stress” “working memory” etc., and “dopamine” or “glutamate”, or “transcranial magnetic stimulation” etc. The search engines used were Google Scholar and the electronic database Medline (PubMed). In the latter, searches were done with MeSH (Medical Subject Headings) and were limited to the English language from 1960 until 2017. We excluded articles where the deficits in executive functions were confounded by neurological conditions or physical trauma. Finally, we cross-referenced publications within review articles for additional citations.

3. Working Memory

Working memory is defined as the temporary storage of information used to perform a variety of cognitive tasks (Baddeley 1992). The role of the PFC in working memory, particularly in tasks with a delay component, is supported experimentally in non-human primate and rodent studies (Brito & Brito 1990). Specifically, lesion studies indicate that the prelimbic cortex of the ventral medial PFC is critical to working memory in rodents (Brito & Brito 1990), and similarly lesions of the homologous lateral PFC demonstrate its importance for working memory in humans (Muller et al 2002). Similarly, the involvement of the prefrontal cortex (PFC) in working memory in humans is evidenced by increased PFC activity in healthy individuals performing working memory tasks (D’Esposito et al 1995), and working memory impairments in patients with frontal lobe damage (as reviewed in (Stuss & Benson 1984).

3.1. Assessments of working memory

In animals, T-maze alternation tasks and object recognition tasks with variable lengths of delay are used to test spatial and non-spatial working memory, respectively (Kinnavane et al 2015, Lalonde 2002, Warburton & Brown 2015). These tasks rely on the intrinsic propensity of the animal to explore novel locations or objects when given a choice between a previously experienced situation and a novel one. Deficits in working memory will diminish preference for the novel location/object, as the animal does not retain information of previously encountered situations. In humans, a commonly used test for working memory is the n-back paradigm (Jonides et al 1997, Kirchner 1958, Pelegrina et al 2015). In this test the subject is presented with a series of stimuli and is asked to indicate when a current stimulus matches one from n steps earlier.

3.2. Effects of stress on working memory

The effects of stress on working memory appear to follow an inverted “U” response in animal models. While acute moderate stress has been shown to have positive effects on working memory in rats (Yuen et al 2009), more intense or prolonged stressors impair working memory in tasks such as spatial delayed alternation (Shansky et al 2006) and spontaneous delayed non-matching-to sample task (Morrow et al 2000). Likewise, repeated corticosterone administration impairs temporal order recognition in rats (Yuen et al 2012), and in rhesus monkeys, loud noise stress impairs delayed-response performance (Arnsten & Goldman-Rakic 1998).

Paralleling the preclinical literature, the effects of stress on working memory in humans can be either positive or negative, depending on intensity and duration of the stressor. For instance, exogenous hydrocorticosteroid administration in the afternoon, near the trough of glucocorticoid circadian rhythm, improves working memory in healthy human subjects (Lupien et al 2002). However, working memory is impaired when the same dose is administered in the morning, at the time of glucocorticoid peak (Lupien et al 1999). In general, prolonged or high intensity stressors worsen performance in working memory tests. For example, social stress causes a robust deficit in working memory as measured by a modified reading span task as well as in the n-back paradigm (Luethi et al 2008, Schoofs et al 2008). Working memory deficits in adults, correlated with childhood poverty, could be attributed to chronic stress experienced in early life (Evans & Schamberg 2009). Finally, working memory deficits are present in a wide range of stress-related psychiatric illnesses, such as PTSD (Veltmeyer et al 2006), depression (Rose & Ebmeier 2006), substance dependence (Bechara & Martin 2004), schizophrenia (Stone et al 1998), and ADHD (Martinussen et al 2005).

3.3. Neurochemical mechanisms and pharmacological targets

Preclinical studies examining the mechanisms underlying stress-induced impairments in working memory as well as current standard treatments for psychiatric illness have informed the testing of various therapies to ameliorate working memory deficits.

Glutamate

The importance of excitatory glutamate signaling to optimal working memory has been demonstrated in healthy subjects both preclinically and clinically, with antagonists of N-methyl-D-aspartate glutamate receptors (NMDA-Rs) resulting in working memory impairments in rodents (Pontecorvo et al 1991), non-human primates (Baron & Wenger 2001, Frederick et al 1995, Roberts et al 2010), and humans (Ghoneim et al 1985, Krystal et al 1994, Oye et al 1992). In rats, the beneficial effect of acute glucocorticoid administration on working memory is accompanied by increased expression of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPA-Rs) and NMDA-Rs as well as increased AMPA-R activity in the PFC (Yuen et al 2009). In contrast, the deleterious effects of repeated corticosterone exposure correlate with decreased glutamate receptor expression and AMPA-R activity (Yuen et al 2012).

GABA

Targeting the primary inhibitory neurotransmitter γ-aminobutyric acid (GABA) has drawn attention from schizophrenia researchers largely due to the hypothesized role that GABAergic cortical interneurons play in the disease. One study comparing healthy volunteers to patients with schizophrenia and concurrent working memory impairment revealed that lorazepam, a positive modulator at the GABA_A receptor benzodiazepine site, impaired working memory in healthy controls, with even more impairment in the patient group. Conversely, the GABA_A receptor antagonist flumazenil significantly improved working memory in the patient group but impaired working memory in healthy controls (Menzies et al 2007). These data suggest that GABA-related “overinhibition” may underlie working memory impairment in schizophrenia, and that targeting this system may offer a strategy for treating such impairments.

Dopamine and norepinephrine

In animal models, it has been found that stress-induced working memory impairments are accompanied by increased prefrontal dopamine turnover and release, similar to the prefrontal dopamine dysfunction seen in schizophrenia. These behavioral and biochemical effects of stress can be mimicked by the anxiogenic compound FG7142, a GABA_A receptor inverse agonist. The working memory deficits caused by this pharmacological stressor are prevented by the GABA_A receptor antagonist flumazenil (RO15-1788), as well as the dopamine receptor antagonists haloperidol, clozapine, and SCH23390, in both monkeys and rats (Murphy et al 1996). Despite the evidence for a hyperdopaminergic mechanism underlying stress effects on working memory, and similarly in psychiatric illness, Mizoguchi and colleagues argue that after a prolonged period of chronic behavioral stress (four weeks, as compared with one week of water immersion/restraint stress), rats actually show decreased prefrontal dopamine transmission that accompanied impaired working memory in a delayed-alternation task. This group went on to show that working memory can be restored with intra-PFC administration of low doses of the D₁ receptor agonist SKF 81297 (Mizoguchi et al 2000). Therefore, similar to corticosterone, an optimal range of dopamine transmission is required for optimal working memory.

FG7142 also mimics the increased norepinephrine release in the PFC that accompanies stress, implicating a role for norepinephrine in the detrimental effects on PFC-mediated working memory. Although propranolol, a β-adrenergic receptor antagonist, did not prevent the FG7142-induced working memory deficit (Murphy et al 1996), the α-1 receptor antagonist urapidil and the α₂ autoreceptor agonists clonidine and guanfacine did restore working memory after pharmacological stress in rats (Birnbaum et al 1999, Birnbaum et al 2000). Indeed, the atypical antipsychotic risperidone, which produces dopaminergic, noradrenergic, and serotonergic antagonism, improves working memory in schizophrenia and bipolar patients (Harvey et al 2007, Harvey et al 2005). Additionally, adjunct administration of guanfacine to schizophrenia patients being treated with risperidone produced mild improvements in working memory as compared to treatment with risperidone alone (Friedman et al 2001). Long-term administration of clonidine to patients with schizophrenia also improved working memory (Fields et al 1988).

Given these observations, it is perhaps paradoxical then that stimulants like methylphenidate and amphetamine, which increase prefrontal catecholamines, have commonly been used to treat cognitive dysfunction such as working memory impairments in patients with ADHD (Arnsten 2006, Pietrzak et al 2006). Administration of stimulants to unmedicated schizophrenic patients can enhance positive symptoms (Angrist et al 1980, van Kammen et al 1982), which is likely due to subcortical D₂ receptor activation. Conversely, low doses administered to stably medicated patients can improve cognitive performance in working memory tasks, likely via prefrontal D₁ receptor activation (Barch & Carter 2005, Goldberg et al 1991), which would agree with the previously discussed preclinical data demonstrating therapeutic effects of the D₁ receptor agonist SKF81297. Furthermore, modafinil, an atypical dopamine transporter inhibitor, reverses chronic stress-induced impairments in a spontaneous alternation task in mice and improves working memory in non-stressed animals (Pierard et al 2006). Similarly, modafinil enhances working memory in healthy individuals as well as individuals with a wide range of psychiatric illnesses, and this improvement is likely due to modafinil’s catecholaminergic actions (Kalechstein et al 2010, Kalechstein et al 2013, Minzenberg & Carter 2008).

Serotonin

With respect to the role of serotonin, while the selective serotonin reuptake inhibitor (SSRI) fluoxetine did not appear to ameliorate the FG7142-induced working memory impairment in rats (Murphy et al 1996), systemic administration of fluoxetine did reverse a predator stress-induced deficit in working memory, while the benzodiazepine diazepam was not effective (Hage et al 2004). Use of SSRIs is not typically associated with improved cognitive functions such as working memory in depressed patients (see Amado-Boccara et al., 1995, but see also (Zobel et al 2004). However, vortioxetine, a multimodal serotonin modulator used to treat depression, has gained attention recently for its beneficial effect on cognition. These cognitive effects seem not to be mediated by its activity as a serotonin transporter inhibitor, but rather depend on its direct actions on serotonin receptors, namely 5-HT₃, 5-HT₇ and 5-HT_1D receptor antagonism, 5-HT_1B receptor partial agonism, and 5-HT_1A receptor agonism. Indeed, vortioxetine reverses working memory impairments caused by serotonin depletion, effects attributed to its 5HT_1A agonism (du Jardin et al 2014), although a recent clinical trial of vortioxetine in depressed patients did not reveal a significant improvement in working memory in the n-back task despite the many other cognitive improvements that were seen (Mahableshwarkar et al 2015).

3.4. Non-pharmacological treatments

Behavioral training

Behavioral therapies have been used clinically to counter the negative effects of stress on working memory. Clinical evidence demonstrates that training patients with disorders, such as ADHD, specifically on visuospatial and verbal working memory tasks via a computerized program (i.e., CogMed), can sometimes improve their working memory performance (as reviewed in (Rapport et al 2013). Preclinical lesion studies show a similar improvement after several training sessions on a T-maze alternation task with short inter-trial intervals. However, rats with PFC lesions never recovered working memory with longer delays even after several training sessions (Brito & Brito 1990). It is possible that such training increases activity of the PFC, and thus can restore function if the brain region is hypoactive (as is thought to be the case in ADHD and other psychiatric illnesses) rather than completely lesioned. Indeed, increased prefrontal activity, as measured by functional magnetic resonance imaging, has been reported in healthy adults after working memory training (Olesen et al 2004). Furthermore, positron emission tomography (PET) data reveals increased D1 receptor density in the PFC after working memory training (McNab et al 2009), again highlighting a role of dopamine in optimal working memory performance.

Mindfulness

Mindfulness is a type of attentional focus characterized by awareness of and attention to the present moment, both internally and externally, allowing for negative thoughts to arise and be acknowledged without judgment or reaction (Bishop et al 2004). Mindfulness-Based Cognitive Behavioral Therapy (MBCBT) is a form of cognitive behavioral therapy that utilizes the awareness underlying the mindfulness technique (Teasdale 1999), and has been shown to prevent relapse in depression. In combination with relaxation techniques, it produces changes in brain and immune function (Davidson et al 2003). Military personnel with a high practice time of mindfulness training during the high-stress pre-deployment period demonstrated increases in working memory capacity, while those in the military control group or those with low practice time demonstrated decreases in working memory. These increases or decreases were correlated with reports of positive or negative affect, respectively (Jha et al 2010). Additionally, expressive writing, in particular about a negative stressful experience, produced higher increases in working memory capacity in college freshmen compared to those writing about a trivial topic; these gains were associated with decreases in intrusive and avoidant thinking (Klein & Boals 2001).

Transcranial magnetic stimulation

In transcranial magnetic stimulation (TMS), high intensity magnetic fields are used to depolarize neurons in a particular area of the brain over a course of several sessions; this can be used to either decrease (with low frequency) or increase (with high frequency) excitability in that region (Luber & Lisanby 2014). In support of the notion that increasing PFC activity can improve working memory, several clinical studies have investigated the therapeutic potential of TMS in the dorsolateral PFC (DLPFC) (as reviewed in (Brunoni & Vanderhasselt 2014). For instance, one session of TMS of the DLPFC was sufficient to improve working memory on the n-back task fifteen minutes later in antidepressant-free patients diagnosed with depression (Oliveira et al 2013). Repetitive TMS targeting the DLPFC in medicated schizophrenia patients has also demonstrated improvements in working memory (e.g. (Barr et al 2013). While stimulation of other brain regions can also improve working memory (e.g. the parietal cortex, cerebellum, and temporal cortex), it is still unclear how the impact on working memory may differ between healthy adults and patients with psychiatric illness (Brunoni & Vanderhasselt 2014).

Vagal Nerve Stimulation

Vagal nerve stimulation (VNS) is another paradigm used for illnesses like treatment-resistant depression, and is thought to function via altering cortical activity and/or by changing the activity of neurotransmitters such as norepinephrine, serotonin, glutamate, and GABA. In VNS, an electrode surrounds the left vagus nerve and communicates with a pulse generator to send pulses of a programmed width, frequency, current, and cycle to the vagus nerve (Schachter & Saper 1998). One study examining the effects of VNS on cognitive function found a significant improvement in working memory in patients with treatment-resistant depression after ten weeks of VNS (Sackeim et al 2001).

4. Attention

Attention is defined as the readiness to detect rare or unpredictable stimuli over an extended time period (Sarter et al 2001). Attention is another executive function of the PFC that is affected by an individual’s emotional and motivational state (Goetz et al 2008) and is often impaired by stress in both humans and in animal models. Lesions of the mPFC, as well as compromised connectivity of the inferior PFC to striatal, cerebellar, and parietal regions, result in deficits in attention in rats and humans (Maddux & Holland 2011, Muir et al 1996), (Arnsten 2009, Arnsten & Rubia 2012, Aron & Poldrack 2005).

4.1. Assessments of attention

Within animal models, variants of the 5 choice serial reaction time task (5-CSRTT) can be used to measure both attention and response inhibition. For this task, rats are placed in a chamber containing multiple nose-poke holes. Stimulus lights are arranged above each hole, and the rat must pay attention to which stimulus light is lit and respond by nose-poking the correct hole. When the rat performs within the time that the light is on, they receive a food reward. The attention portion of the task measures the number of omissions that occur, in which the rat does not respond to the stimulus light (Robbins 2002).

Several tests are used to measure attention in humans, e.g., the Visual Attention Task, Selective Attention Task, Test of Variables of Attention and the Sustained Attention to Response Task (Hanania & Smith 2010). In general, in these tests the subject responds by pressing a button when a target object is presented. The subject may have to pay specific attention to some of the object’s attributes, like color or shape, (for example, in a sequence of objects only respond to red circles) or may have to recognize one specific object among a series of distractors. Another test of attention is the Stroop ask. The Stroop task evaluates selective attention by determining an individual’s ability to overcome Stroop interference (Washburn 2016) (MacLeod 1992). The classic example of the Stroop task consists of presenting a subject with words that spell a color but are of a different color (for example the word “green” colored in blue), thus forcing the subject direct his or her attention to the color of the word.

4.2. Effects of stress on attention

Several studies have shown different modalities of attention to be compromised by stress. Selective attention, or the ability to ignore irrelevant cues, worsens in animals after they receive an inescapable footshock (Minor et al 1984). Rats exposed to prenatal stress exhibit impaired accuracy in the 5-CSRT (Wilson et al 2012). Stress can result in attention deficits also in humans. For example, prenatal stress in the form of malnutrition has linked a predisposition to attention problems (Liu & Raine 2006). Attention deficits are also linked to times of acute stress. Students tested on their ability to pay attention to details performed more poorly during exam periods in a semester than during non-exam periods (Vedhara et al 2000). Interestingly, short-term stress and emotional state can affect the redirection of attention. A study performed on college students found that students in the negative stressor condition (where stress was induced by engaging in a competitive computer task) were quicker to switch their attention from negative words to positive or neutral words, and participants who had higher scores on the Beck Depression Inventory were slower to redirect their attention from negative to positive words (Ellenbogen et al 2002). Another study in humans found that in high-stress situations, attention to irrelevant stimuli as measured by the Stroop and Garner task is decreased when compared to low-stress situations (Chajut & Algom 2003). Further, in one study assessing attentional bias towards threatening information, participants who received experimentally-induced stress paid more attention to threatening words as measured by the attention deployment task (Mogg et al 1990). Finally, combat veterans who have experienced stressful trauma and exhibit PTSD have decreased attention during cognitive tasks as compared to newly recruited national guardsmen (Uddo et al 1993). Taken together, these studies suggest that the effects of stress on attention may depend on the amount of time one is subjected to stress (i.e. chronic, acute, and experimentally-induced brief stress) and the emotional valence of the stimuli presented during the task.

4.3. Neurochemical mechanisms and pharmacological targets

Glutamate

Not much is known about the role of the glutamate system in regulating attention in conditions of stress. In non-stressed animals both competitive NMDA-R antagonists, such as (R)-CPP (Mirjana et al 2004) and non-competitive NMDA-R antagonists such as dizocilpine, and phencyclidine (Higgins et al 2003, Le Pen et al 2003) injected into the mPFC produce pronounced attentional performance deficits as measured by increase in omissions and latency for correct detection in the 5-CSRTT. Recent work has shown that the mGluR2/3 antagonist LY341495 prevents the (R)-CPP-associated deficits in attention (Pozzi et al 2011), suggesting a specific role of glutamate signaling at metabotropic glutamate receptors, potentially involving CREB signaling (Paine et al 2009, Pozzi et al 2011) in modulating attentional behaviors. In support of this, in a genome-wide study of children with ADHD and matched controls, variations in copy number within genes interacting with metabotropic glutamate receptors were enriched in 10% of ADHD cases (Elia et al 2011). Thus, this system may represent a potential target for future therapeutics. Whether such mechanisms are also playing a part in stress-induced attentional deficits is a question worthy of future testing.

Dopamine

Multiple studies have explored the role of D₁ and D₂ receptors within the PFC in regulating attention in rats. Activating D1 receptors can facilitate attention but excessive D1 activation via stimulant medication can impair mPFC function (Chen et al 2014). The effects of D₁/D₂ agonists vary between prefrontal cortical subregions. For example, direct infusion of D₁ agonists into the mPFC increased attention and accuracy, but only in more challenging tasks (Chudasama & Robbins 2004, Passetti et al 2003). However, D₁ agonist infusion into the OFC had no effect on attention, but D₂ agonist infusion in the same area decreased attention. In humans, dysregulation of the dopamine system can result in attentional impairments that may be caused by a deficiency in forming reward-seeking connections and thus inhibit goal-directed attention (Sagvolden et al 1998). As dopamine functions to regulate reward-seeking behavior, impaired dopamine function in the brain can result in deficits in reward-related memory formation (Sagvolden et al 2005). Additionally, genetic analyses in humans have shown that individuals with attention deficits have higher expression of DAT (Krause et al 2000). Thus, drugs such as methylphenidate that function by increasing extracellular dopamine through DAT inhibition are routinely used to target attention deficits, and these effects are context-dependent (Volkow et al 2005).

Norepinephrine

Norepinephrine transport inhibitors increase attention in both human and animal models (Chamberlain et al 2007, Michelson et al 2003, Navarra et al 2008). Conversely, depleting norepinephrine in the PFC decreases attention (Carli et al 1983, Milstein et al 2007). These effects seem to be mediated by α-adrenergic receptors, in slightly different manners. The α₁-adrenoceptor antagonist prazosin and the αβ₁/β/α₂ adrenergic receptor antagonist propranolol produced deficits in attention, the α₂-adrenoceptor antagonist atipamezole improved continued attention (Bari & Robbins 2013b). This is in contrast to studies showing deleterious effects of other α₂ receptor antagonists, such as idazoxan and yohimbine, on attention (Arnsten & Li 2005, Rowe et al 1996). These discrepancies have been attributed to differences in behavioral tests, as well as the lower selectivity of the latter compounds, which may produce non-specific effects. Activation of α₂ autoreceptors improved attention in some studies (Smith & Nutt 1996), but not in others (Fernando et al 2012). Thus, the specific role of α₂ receptor in the regulation of attention still needs to be fully clarified.

A few studies have investigated the role of norepinephrine on attentional tests in chronically stressed animals. In adult rats exposed to variable prenatal stress, deficits in attentional responses were found in a variant of the 5-CSRTT, and these deficits were corrected in a dose-dependent manner by the norepinephrine reuptake inhibitor atomoxetine (Wilson et al 2012). Other studies have shown that auditory attention impaired by chronic restraint stress is alleviated by reboxetine, another norepinephrine reuptake inhibitor (Perez-Valenzuela et al 2016).

Serotonin

The serotonergic system has long been implicated in attentional performance (Harrison et al 1997, Hiraide et al 2013). The involvement of various serotonin receptors in regulating attention has been extensively studied. For example, mPFC infusion of the 5-HT_2A/2C receptor antagonist ketanserin had no effect on attention (Passetti et al 2003), while M100907, a selective 5-HT_2A antagonist, improved performance by increasing accuracy. The 5-HT_2A/2C agonist DOI reduced performance by negatively affecting attention (Koskinen et al 2000). By contrast, a selective 5-HT_1A agonist 8-OH-DPAT had beneficial effects, increasing attention (Winstanley et al 2003). In a human study of elderly subjects with generalized anxiety, citalopram reduced attentional performance compared to placebo; this effect strongly correlated with high-transcription variants of 5-HT_2A and 5-HT_1B receptors (Lenze et al 2013). Together, the above data suggest that serotonin signaling through the 5-HT_1A receptor and concomitant inhibition of 5-HT_2A receptor in the PFC would be beneficial to attention.

4.4. Non-pharmacological treatments

Non-pharmacological treatments to treat attentional deficits are often administered in adjunct with pharmacotherapy. The molecular mechanisms underlying how these therapies work are still being explored and may uncover new therapeutic targets.

Cognitive behavioral therapy

CBT, alone and in combination with pharmacotherapy, has been successfully used to ameliorate core symptoms of adult ADHD, such as attentional deficits (Safren et al 2005, Weiss et al 2012). Additionally, twelve sessions of CBT reduced abnormal functional connectivity in the frontoparietal network and in the superior parietal gyrus in ADHD subjects (Wang et al 2016). Another form of behavioral therapy, mindfulness-based stress reduction (MBSR), improves subjects’ ability to direct attention and detect changes in stimuli through mindfulness training (Jha et al 2007).

Transcranial magnetic stimulation

While TMS treats multiple cognitive deficits, it specifically improves attention when targeted to the left DLPFC. Within depressed patients, one session of TMS targeted to the left DLPFC increased activity within the region and increased attention in an affective go-no-go task (Bermpohl et al 2006).

5. Response Inhibition

Response inhibition (the lack of which results in impulsivity) is another executive function of the PFC that is detrimentally affected by stress in both animal models and human disorders. Impulsivity can be viewed as the tendency to act prematurely, without foresight (Dalley et al 2011), and has been defined as “actions which are poorly conceived, prematurely expressed, unduly risky or inappropriate to the situation and that often result in undesirable consequences,” (Chamberlain & Sahakian 2007). Impulsivity spans multiple cognitive domains and comprises multiple types of behavior: the failure to adequately collect and evaluate information before reaching decisions (reflection impulsivity), the tendency to opt for smaller and immediate rewards over larger delayed rewards (impulsive choice), and the inability to suppress prepotent motor responses (response inhibition) (Chamberlain & Sahakian 2007, Dalley et al 2011, Fineberg et al 2010). In this review we will focus on response inhibition.

OFC and infralimbic (IL) mPFC have been shown to mediate response inhibition, as lesions of these areas produce premature responding in the 5-CSRTT and SSRT (Chudasama et al 2003, Eagle et al 2008a, Eagle & Robbins 2003).

5.1. Assessments of response inhibition

Within animal models, several different behavioral tasks are used to measure response inhibition. First, variants of the 5 choice serial reaction time task (5-CSRTT) can be used to measure response inhibition. This portion of the task measures the number of premature responses, or when the animal responds before a stimulus light is lit. Another behavioral task used to evaluate response inhibition is the Stop-Signal Reaction Time Test (SSRTT). This test first developed for human subjects (Logan et al 1984, Votruba & Langenecker 2013) has been subsequently adapted with minor modifications for rats (Eagle et al 2008a, Eagle et al 2008b). In this test, subjects are trained to respond to a visual cue by pressing a lever or button on the left (for cue A) or on the right (for cue B). A number of trials (generally 75%) are recorded under these conditions (Go-trials). In 25% of trials a stop signal, generally a tone, is given concomitantly or after presentation of the visual cue and the subject is required to stop responding (stop or No-Go trial). The parameters measured in this test are the accuracy of response, the proportion of successful stops, the reaction time on Go-trials, and the stop signal reaction time.

5.2. Effects of stress on response inhibition

As with other cognitive functions of the PFC, response inhibition is affected differently by acute or chronic stressors. Local microinjections of increasing doses of corticotrophin releasing factor (CRF) to simulate acute stress had no effect on response inhibition in 5-CRSTT (Ohmura et al 2009). However, chronic restraint stress reduced rats’ ability to inhibit their responses (Mika et al 2012). Chronic exposure to cortisol levels that simulated prolonged stress response caused reduced response inhibition also in non-human primates (Lyons et al 2000). In human studies, acute stress enhanced response inhibition and this effect was blocked by the mineralocorticoid receptor antagonist spironolactone (Schwabe et al 2013). Conversely, prolonged stress worsens response inhibition in humans. Both adult and adolescent volunteers that were in a self-reported “high stress” state performed worse in an impulse inhibition task than individuals in a “low stress” state (Rahdar & Galvan 2014). Response inhibition is also compromised in psychopathology. For example, patients with PTSD made more errors during the NoGo phase of the SSRTT than controls. Higher levels of PTSD were correlated with even higher error rates (Swick et al 2012). Individuals with borderline personality disorder also presented increased impulsivity following exposure to stress (Krause-Utz et al 2013). Both human and animal studies indicate that reduced response inhibition emerge from compromised functionality of the inferior PFC and its connections to striatal, cerebellar, and parietal regions (Arnsten 2009, Arnsten & Rubia 2012, Aron & Poldrack 2005).

5.3. Neurochemical mechanisms and pharmacological targets

Glutamate

The role of the glutamate system in regulating response inhibition in conditions of stress is not very well documented. However, some studies have explored the role of NMDA receptors within the PFC in non-stressed animals. For example, (R)-CPP produced increased premature responding, indicating deficits in response inhibition (Mirjana et al 2004). Further studies revealed that the IL portion of the mPFC may be the site responsible for the effects of (R)-CPP on response inhibition (Murphy et al 2005); this was confirmed in other studies using the NMDA-R antagonist MK801 (Benn & Robinson 2014).

Glutamate is most likely being modulated by other neurotransmitters, as administration of the serotonin 5-HT_2A receptor antagonist M100907 prevents increases in premature responding induced by MK-801 (Higgins et al 2003) and (R)-CPP (Ceglia et al 2004). Additionally, both NMDA-R antagonism and 5-HT_2A receptor activation result in increased glutamate release (Aghajanian & Marek 2000, Moghaddam et al 1997). Thus, it has been proposed that excess glutamate in the IL cortex, acting via non-NMDA receptors, may be responsible for increases in premature responding.

Dopamine

Although some studies have reported altered dopamine transmission in association with impaired pre-pulse inhibition in rats exposed to early life stress (Heidbreder et al 2000), the majority of studies investigating the role of dopamine in response inhibition involve non-stressed cohorts. The role of dopaminergic neurotransmission in response inhibition is complex. Multiple studies have explored the role of D₁ and D₂ receptors within the PFC in regulating response inhibition in rats. Drugs of abuse that increase dopamine in the synapse, such as amphetamine, decrease response inhibition (Robbins 2002), however direct agonism at D₁ or D₂ receptors within the OFC per se does not increase impulsivity (Winstanley et al 2010) indicating that other neurotransmitter systems or other brain regions are involved in this behavior. Indeed, antagonism of dopamine receptors demonstrates the complexity of the system, as blocking either D₁ or D₂ in the mPFC or OFC had no effect on the 5-CRSTT (Granon et al 2000, Winstanley et al 2010), or the SST (Bari et al 2011, Bari & Robbins 2013b) (Bari & Robbins 2013b). However, blocking D2 receptors in the dorsal striatum negatively affects SST performance (Eagle et al 2011). From these data it has been suggested that dopaminergic transmission within the striatum underlies the automated motor response of the behavior (Bari & Robbins 2013a). The role for D2 receptors in response inhibition has also been observed in human subjects where different polymorphisms of this gene correlate with differential response times on SSRT (Hamidovic et al 2009). Additionally, the role of striatal D2 receptors have been linked to impulsivity and addictive behavior (Volkow et al 2007).

Norepinephrine

Although the role of norepinephrine in impulsive behavior following stress has not been investigated so far, several studies have shown its importance in modulating this behavior in non-stress conditions. For example, atomoxetine, a relatively selective norepinephrine uptake inhibitor, improves response inhibition in both the 5-CRSTT and the stop-signal task (Robinson et al 2008). Conversely, guanfacine, an agonist of α₂-adrenergic autoreceptors, which reduces ascending norepinephrine release, worsens response inhibition in rats (Bari et al 2009, Bari et al 2011) while the antagonist atipamezole improves performance (Bari & Robbins 2013b). However, guanfacine did not improve impulsivity in healthy volunteers (Muller et al 2005) and other α₂ receptor antagonists, such as idazoxan and yohimbine, have been reported to decrease impulse control (Arnsten & Li 2005, Rowe et al 1996). The α₁-adrenoceptor antagonist prazosin and the αβ₁/β/α₂ adrenergic receptor antagonist propranolol do not affect response inhibition on their own, but propranolol improves inhibitory control impaired by yohimbine, suggesting that norepinephrine may act at multiple receptors to bring about its effects on impulse control (Adams et al 2017). The mechanism by which norepinephrine improves response inhibition is not currently known, but it may reflect the role of this neurotransmitter in regulating signal-to-noise ratio in PFC input and reorienting attention to salient stimuli in the environment (Woodward et al 1991) (Aston-Jones & Cohen 2005).

Serotonin

The serotonergic system has long been implicated in the regulation of behavioral inhibition (Evenden 1999, Linnoila et al 1983). Microinjections of fluoxetine into the OFC increase response inhibition in rats already showing impulsive tendencies, but have no effect in low impulsivity rats. This finding indicates direct effects of SSRIs within the PFC of animals already exhibiting a deficit, but do not enhance response inhibition in normal animals (Darna et al 2015). However, this seems specific to fluoxetine, as citalopram, another SSRI, does not correct deficits in response inhibition in either human or animal studies (Chamberlain et al 2006, Clark et al 2005, Eagle et al 2008a, Eagle et al 2009). (Bari Robbins 2009) This phenomenon may be linked to the recent evidence that the serotonin transporter genotype influences the effects of SSRIs on response inhibition in humans (Fischer et al 2015).

The involvement of various serotonin receptors in regulating response inhibition has been extensively studied, and has revealed a complex picture. For example, mPFC infusion of ketanserin increased response inhibition (Passetti et al 2003), while M100907 improved performance by increasing response control. DOI reduced performance by negatively affecting response inhibition (Koskinen et al 2000). By contrast, 8-OH-DPAT had no effect on impulse inhibition (Winstanley et al 2003). Together, the above data suggest that serotonin signaling through the 5-HT_1A receptor and concomitant inhibition of 5-HT_2A receptor in the PFC reduces impulsivity.

Interestingly, 5-HT_2A antagonism is also able to reverse the effects of NMDA-R blockade. Injecting M100907 subcutaneously reverses CPP-induced decreases in response inhibition (Mirjana et al 2004). 5-HT_1A antagonism with 8-OH-DPAT within the PFC was also able to block CPP-induced elevations in extracellular glutamate (Calcagno et al 2006). These observations suggest that serotonin’s effects on response inhibition may be through modulation of the glutamate system.

Multiple Neurotransmitter Systems

While many of the studies above have focused on each of these neurotransmitter systems individually, it is overly simplistic to believe that a simple therapeutic approach can be discovered without looking at the system as a whole. Under basal conditions, the monoaminergic systems modulate excitatory and inhibitory transmission and this convergence in signaling often underlies the multiple and overlaying effects observed using even highly selective pharmacotherapies. In a complementary way, the effectiveness of some pharmacological treatments depends on their multi-systems actions. For example, methylphenidate (MPH, or Ritalin), a drug often used in the treatment of ADHD, interacts with multiple neurotransmitter systems for its therapeutic treatment, including dopamine, norepinephrine, and acetylcholine (Robbins 2002). MPH is also effective in increasing response inhibition in rats (Eagle et al 2007, Navarra et al 2008) where MPH effects on response inhibition can be blocked with a combination of adrenoceptor antagonists and dopamine antagonists, but are not affected if only a single antagonist is administered (Arnsten & Dudley 2005), suggesting an interplay of the two neurotransmitter systems.

Additionally, atypical antipsychotics, such as sertindole and clozapine, that act at both dopamine and serotonin receptors prevent CPP-induced deficits in response inhibition (Carli et al 2011). Likewise, both drugs are able to suppress the elevated glutamate release that is associated with both NMDA-R antagonism and behavioral deficits (Carli et al 2011). These drug studies suggest that the serotonin and dopamine systems have a synergistic effect on the glutamate system, specifically in regards to NMDA receptor hypofunction.

5.4. Non-pharmacological treatments

Cognitive behavioral therapy and mindfulness

CBT, alone and in combination with pharmacotherapy, has been successfully used to reduce impulsivity and disinhibition in ADD (Safren et al 2005, Weiss et al 2012). Recent studies have also demonstrated beneficial effects of mindfulness-based stress reduction (MBSR) training on behavioral inhibition. Compared to non-meditators, healthy volunteers who had followed a meditation training course for 6–8 weeks performed better on a Stroop task (Allen et al 2012, Moore & Malinowski 2009, Teper & Inzlicht 2013), the Hayling task (Heeren et al 2009), and response inhibition tasks (Sahdra et al 2011).

Vagal nerve stimulation

Finally, another noninvasive treatment for impulse control in current use is VNS. In comparing patients before and after 10 weeks of VNS, the treatment significantly improved response inhibition, as well as other cognitive symptoms (Sackeim et al 2001). VNS increases firing rates of 5-HT and norepinephrine neurons (Dorr & Debonnel 2006), which may increase levels of 5-HT and norepinephrine leading to increased NMDA-R function within the mPFC, and subsequent beneficial effects on response inhibition measures (Carli et al 2011, Carli & Invernizzi 2014, Darna et al 2015, Mirjana et al 2004).

6. Cognitive flexibility

Cognitive flexibility, or the ability to adapt one’s behavior in response to a changing environment, is often dysregulated in patients with neuropsychiatric disorders such as depression, post-traumatic disorder, eating disorders and schizophrenia (Anisman & Matheson 2005, Austin et al 2001, Leeson et al 2009, Tchanturia et al 2012). For example, in patients suffering from depression, the biased attention for negative information and the maladaptive perseverative cognition and behaviors persist despite a changing environment (Disner et al 2011, Peckham et al 2010). A large body of evidence indicates that stress and anxiety negatively affect several aspects of cognitive flexibility (McKlveen et al 2015, Park & Moghaddam 2017); in this review, we will specifically focus on two cognitive flexibility dimensions: reversal learning and set-shifting.

6.1 Assessments of cognitive flexibility

Since the animal tests for cognitive flexibility are modeled directly on the human tests, we will begin by describing the latter first. Tests used to measure cognitive flexibility in humans and nonhuman primates include the Stroop Test, the Trail Making Test, the Wisconsin Card Sorting Test, and a computerized adaptation of the Wisconsin Card Sorting Test, the interdimensional/extradimentional (ID/ED) set-shifting task of the Cambridge Neuropsychological Test Automated Battery (CANTAB). The Stroop Test measures interference in the reaction time of a task (for example, naming the color “red” when presented with the word “green” written in red ink), and has been used not only as an indicator of selective attention, as mentioned above, but also of broad cognitive flexibility processes involved in planning, decision making and managing environmental interference. The Trail Making Test (Sanchez-Cubillo et al 2009) measures the time employed to connect a series of dots. This test consists of two parts: in the first segment, the subject is asked to quickly connect a set of dots identified by numbers (1,2,3…) in ascending order; in the second segment, the subject is asked to connect alternating number and letters (1, A, 2, B….). The first part of the test measures visual and scanning speed, while the second part is an index of the ability to maintain two trains of thought simultaneously, the ability to generate and modify a plan of action, and general cognitive flexibility. In the Wisconsin Card Sorting Test (WCST) (Berg 1948, Nyhus & Barcelo 2009, Park & Moghaddam 2017), participants have to classify cards according to different criteria (color, number, or shape of symbols on cards). The only feedback given is “correct” or “not correct”. The classification rule changes every 10 trials; therefore, after figuring out the rule, the subject will make a series of mistakes when the rule changes. The test measures the ability of subjects to modify behavior in the face of a change in reward contingency. The ID/ED set-shifting task, used with both humans and non-human primates, is similar but requires selection of line figures and color-filled shapes on a computer screen (Roberts et al 1988, Tyson et al 2004, Weed et al 2008). In rodent models, the attentional set-shifting task (AST; (Birrell & Brown 2000)), an adaptation of the ID/ED test, requires rats or mice to learn to associate a specific cue, within one of multiple stimulus dimensions, with a reward. Once they master a given contingency, the rules are changed, requiring them to shift attention between perceptual dimensions, allowing tests of reversal learning and set-shifting. Detailed methodological descriptions of this test can be found in (Birrell & Brown 2000, Bondi et al 2008, Brown & Tait 2016, Lapiz-Bluhm et al 2008).

6.2. Reversal learning

Reversal learning is a type of discrimination learning. After the acquisition of a rule associating a specific discriminatory stimulus within one sensory dimension with a reward, the reinforcement contingencies are swapped (Izquierdo et al 2017). Thus, in typical paradigms, subjects are trained to discriminate between two stimuli in one sensory dimension visual, olfactory or spatial; only one stimulus is rewarded every time it is chosen. After successful discrimination learning, gauged by meeting a predefined criterion of performance, the rewarded contingency is reversed, such that the previously negative stimulus is now associated with the reward. Subjects are then trained to learn this new rule until they meet the performance criterion. Reversal paradigms have high translational value because they can be used with minor variations in different species, including humans (Fellows & Farah 2003, Hornak et al 2004), monkeys (Clarke et al 2008, Crofts et al 1999, Izquierdo et al 2004, Walker et al 2009), and rodents (Bissonette & Powell 2012, Boulougouris et al 2007, Jentsch & Taylor 2001, Schoenbaum et al 2000). In the various test paradigms, the relationship between stimuli and outcomes can be deterministic (often used in rodent models), or probabilistic (used in monkeys and humans, to reduce the use of simple strategies to predict outcomes) (Costa et al 2015, Dalton et al 2016, Walton et al 2010). This form of learning is mediated by the orbitofrontal cortex (OFC), as evidenced by impairments in reversal learning following lesions of this region in humans (Fellows & Farah 2003, Hornak et al 2004), monkeys (Dias et al 1996), mice (Bissonette et al 2008) and rats (McAlonan & Brown 2003). Other brain regions important for reversal learning are the striatum (Cools et al 2002, Rogers et al 2000), amygdala (Izquierdo et al 2013, Schoenbaum et al 2003, Wassum & Izquierdo 2015) and the hippocampus (Mala et al 2015).

6.2.1. Effects of stress on reversal learning

Chronic stress produces deficits in reversal learning in animal models. For example, two to five weeks of chronic intermittent cold stress results in robust deficits in the reversal learning component of the AST in rats (Danet et al 2010, Donegan et al 2014, Furr et al 2012, Lapiz-Bluhm et al 2009, Patton et al 2016, Wallace et al 2014). Similarly, two weeks of chronic unpredictable stress also produces significant deficits in reversal learning, measured using the AST (Bondi et al 2007, Bondi et al 2010, Jett et al 2015, Jett & Morilak 2013) or using a Morris water maze spatial reversal task (Hill et al 2005, Quan et al 2011, Yu et al 2016). Reversal learning deficits have also been documented in several stress-related psychiatric disorders such as major depressive disorder (where it was accompanied by abnormalities in the OFC) obsessive-compulsive disorder, and generalized anxiety disorder (Drevets 2007, Szabo et al 2013). Moreover, individuals repeatedly exposed to traumatic events, even without a diagnosis of PTSD or anxiety-related disorder, displayed impaired performance in a cue-context reversal learning paradigm (Levy-Gigi & Richter-Levin 2014).

6.2.2. Neurochemical mechanisms and pharmacological targets

Glutamate

Glutamate neurotransmission is necessary for optimal reversal learning. For example, mice hemizygous for the vesicular glutamate transporter 1, which controls the number and size of synaptic vesicles in excitatory synapses, show impaired reversal learning due to increased perseveration in a visual discrimination task (Granseth et al 2015). However, a large number of studies investigating the contributions of NMDA-R have yielded contradictory results. In rats, acute systemic administration of the NMDA-R antagonist MK-801 produced deficits in reversal learning in some studies (Lobellova et al 2013) but not in others (Svoboda et al 2015). Similarly, phencyclidine (PCP), when used at doses to mimic schizophrenia-like symptoms in rats, impaired reversal in some cases (Abdul-Monim et al 2007), but in other studies it was found to have no effect (Brigman et al 2010, Janhunen et al 2015), or to even improve reversal learning (Dix et al 2010, Fellini et al 2014, McAllister et al 2015).

In stressed animals, NMDA-R antagonists appear to have beneficial effects on stress-induced reversal deficits. Thus, amantadine, a low affinity NMDA-R antagonist, has been shown to correct deficits in spatial reversal learning in the Morris water maze in mice that experience a combination of chronic unpredictable stress and isolation feeding (Yu et al 2016). In addition, ketamine, another noncompetitive NMDA-R antagonist, with rapid antidepressant actions (Autry et al 2011, Duman et al 1997, Zarate et al 2006), is able to correct a deficit in reversal learning induced by chronic cold stress in rats (Patton et al 2016). The apparent discrepancies between studies investigating the effects of NMDA-R antagonists in reversal learning may be due to methodological differences. However, it should also be noted that the different compounds used in these studies vary in the other targets they interact with in addition to the NMDA receptor. Amantadine increases dopamine release and adrenergic activity, both of which can facilitate reversal learning (see below). In the case of ketamine, recent evidence suggests that one of its metabolites, norketamime (HNK) is actually responsible for the antidepressant effects of this drug, through activation of glutamate AMPA receptor (GluA-R) (Zanos et al 2016). Thus, it is possible that ketamine’s beneficial effects on reversal learning are due to regulation of AMPA-R transmission. Indeed, recent evidence suggests this may be the case. It is thought that synaptic depression facilitates reversal learning by preventing the interference of previous learning with the encoding of new information necessary to respond to a task change. Indeed, LTD is required for optimal reversal learning: administration of a ionotropic glutamate receptor NMDA subunit 2B (GluN2B) selective antagonist or blockade of GluA-R internalization prevents LTD in the hippocampus, and produces a disruption of spatial reversal learning in the Morris water maze (Dong et al 2013, Duffy et al 2008, Mills et al 2014). In a recent study, the ability of ketamine to correct stress-induced reversal learning deficits was accompanied by decreased field potentials in the OFC evoked by activation of mediodorsothalamic afferents, and both effects required activation of the JAK2/STAT3 signaling pathway in the OFC (Patton et al 2016). JAK2 and STAT3 have been shown to mediate maintenance of LTD also in the hippocampus (Nicolas et al 2012). Although the precise mechanism is not known, JAK2 has recently been shown to modulate the levels of the immediate early gene Arc, which regulates GluA receptor internalization (Chowdhury et al 2006, Patton et al 2016), suggesting that JAK2 facilitates LTD by modulating Arc levels. With these considerations, one possible mechanism for the actions of ketamine on reversal learning is that by activating JAK2, ketamine facilitates conditions of increased GluA-R internalization and LTD within the OFC that favor reversal learning. Supporting and extending this idea, other activators of the JAK2/STAT3 pathway, such as interleukin 6 (IL-6), have been shown to correct the reversal learning deficits induced by cold stress, and blockade of IL-6 signaling or JAK2 inhibition produce deficits in reversal learning (Donegan et al 2014).

In sum, although it is not currently clear whether NMDA-R antagonism is in itself beneficial to reversal learning, the picture that emerges from the above studies is that modulation of glutamatergic transmission leading to LTD in the OFC (or hippocampus) is necessary for optimal performance in reversal learning tasks, thus suggesting potential new avenues for therapeutic development.

Dopamine and norepinephrine

Dopamine is required for several reward-associated learning processes; therefore, the dopamine system may represent a potential point of therapeutic intervention for reversal learning deficits caused by stress (Adamantidis et al 2011, Rossi et al 2013, Schultz 2013). The importance of dopaminergic transmission in reversal behavior has been well documented, especially after extensive training on reversal learning, when the possibility of reversal is “expected” (Costa et al 2015, Klanker et al 2015). There seems to be regional specificity to the effects of dopamine on reversal learning. In particular, dopamine release, specifically in the striatum, is likely to be necessary for optimal performance in reversal learning tasks. Depletion of dopamine in this brain region, but not in the OFC, caused deficits in marmosets (Clarke et al 2011, Clarke et al 2007), and optimal reversal learning in humans was correlated with increased dopaminergic activity only in the striatum (Clatworthy et al 2009).

Targeting dopamine receptors, specifically D₂ receptor signaling (which is increased by repeated social defeat stress (Montagud-Romero et al 2016) could prove effective. Impaired reversal learning following D₂/D₃ antagonists was observed in rats (Boulougouris et al 2009) and monkeys (Lee et al 2007, Smith et al 1999). In healthy humans, the D₂ receptor antagonist sulpiride enhances reward-based reversal learning (van der Schaaf et al 2014), whereas a D₂ receptor agonist bromocriptine impairs probabilistic reversal learning (Mehta et al 2001). Together, these results suggest that down-regulation of D₂R may facilitate reversal learning performance.

Norepinephrine is also known to play a modulatory role in cognitive flexibility. Propanolol (a β-adrenergic receptor antagonist) improved performance in the reversal learning component of the rodent AST (Hecht et al 2014). In addition, acute and chronic administration of the norepinephrine reuptake inhibitor desipramine (DMI) improved reversal learning in naïve rats, in both the AST (Lapiz et al 2007) and a four-position operant discrimination test (Seu & Jentsch 2009, Seu et al 2009). Elevating norepinephrine transmission using either DMI or the α₂-autoreceptor inhibitor atipamezole corrected deficits in reversal learning brought about by chronic unpredictable stress, and this effect was dependent on α₁-receptors (Bondi et al 2010). Additionally, in human studies, by using an anagram test as a measure of cognitive flexibility, healthy volunteers receiving propranolol had enhanced performance in the test, while individuals receiving ephedrine (an activator of α- and β-adrenergic receptors) had impaired performance (Beversdorf et al 1999). Taken together, this evidence suggests norepinephrine transmission, through α receptors but not β receptors, is required for optimal reversal learning.

Serotonin

The role of serotonin in reversal learning has been extensively studied. Systemic administration of agents that deplete serotonin, such as paracholorophenylalanine (PCPA) (a tryptophan hydroxylase inhibitor) or parachloroamphetamine (PCA) creates deficits in reward learning paradigms, including reversal learning, in non-stressed rats (Izquierdo et al 2012, Lapiz-Bluhm et al 2009, Masaki et al 2006). In monkeys, serotonin depletion within the prefrontal cortex using 5,7-dihydroxytryptamine caused reversal learning deficits, but not set-shifting deficits (Clarke et al 2004, Clarke et al 2005). In addition, SSRIs improve reversal learning in spatial reversal tasks in naïve rats (Barlow et al 2015, Brown et al 2012), and correct reversal learning deficits induced by chronic cold stress (Danet et al 2010, Lapiz-Bluhm et al 2009). Functionality of the serotonin transporter (5-HTT) plays an important role, as genetic inactivation of 5-HTT in mice produces deficits in reversal learning (Brigman et al 2010), and depletion of tryptophan in healthy human volunteers homozygous for the long allele of 5-HTT reduced the capacity of the individuals to utilize negative feedback information from punishment and impaired performance on reversal learning. The increased sensitivity of long allele homozygotes to decreased serotonin may be due in part to their accelerated serotonin reuptake compared to short allele carriers (Finger et al 2007).

The importance of the activation of specific serotonin receptors has also been investigated. For example, the 5-HT_2A receptor antagonist MDL 100-907, administered systemically or intra-OFC, produced deficits in reversal learning in naïve rats (Boulougouris et al 2008, Furr et al 2012), whereas the 5-HT_2C receptor antagonist SB 242084 improved reversal learning in naïve rats (Alsio et al 2015, Boulougouris et al 2008, Boulougouris & Robbins 2010). Finally, the multimodal serotonin modulator, vortioxetine is effective in ameliorating reversal learning deficits produced by chronic cold stress (Wallace et al 2014). The effects of vortioxetine are not due solely to its ability to inhibit serotonin reuptake, since the drug was effective in correcting reversal learning deficits induced by serotonin depletion with PCPA (Wallace et al 2014). Rather, it is likely that the beneficial effects of vortioxetine are due to direct agonism at post-synaptic 5-HT_1A and or 5-HT_1B receptors (du Jardin et al 2014). Together, these studies suggest that serotonergic tone in the OFC is crucial for optimal reversal learning across species.

6.2.3. Non-pharmacological treatments

Cognitive Behavioral Therapy

Not many studies have investigated the impact of cognitive behavioral therapy (CBT) on reversal learning performance. However, one recent report has shown positive effects of CBT on reversal learning in OCD patients. In this study, individuals that scored worse on a scale for OCD symptoms (Yale-Brown Obsessive Compulsive Scale, or YBOCS) made more spontaneous errors and more unnecessary strategy changes on a probabilistic reversal learning task, and showed reduced activity in the orbitofrontal cortex and right putamen. After 8–12 weeks of CBT, these individuals improved both their YBOCS and reversal task scores and showed a more stable activation of the pallidum (Freyer et al 2011). Thus, CBT may effectively improve reversal learning deficits seen in psychiatric disorders.

Repetitive Transcranial Magnetic Stimulation

Several studies have examined the relationship between TMS and impaired cognitive functioning (Demirtas-Tatlidede et al 2015). Repetitive TMS improved spatial reversal learning in the Morris water maze in non-stressed rats; these behavioral changes were associated with increased synaptic plasticity and increased expression of BDNF, post-synaptic NMDA-R subunit NB2, and the presynaptic protein synaptophysin in the hippocampus (Shang et al 2016). Theta transcranial stimulation also improved reversal learning in healthy human volunteers (Wischnewski et al 2016).

6.3. Set-shifting

Set-shifting, unlike reversal learning, requires shifting an attentional response to different stimulus attributes across different dimensions, when their relative reinforcement value is changed. As mentioned earlier, the Wisconsin Card Sorting Test (WCST) is often used to investigate set-shifting. Using positron emission tomography (PET) imaging on human subjects, it has been shown that prefrontal cortical activity is associated with set-shifting performance on the WCST (Miller & Cohen 2001, Rogers et al 2000). Thus, disruption of the integrity of the prefrontal cortex, in particular the dorsolateral areas, results in declined performance on set-shifting in the WCST (Manes et al 2002, Miller & Cohen 2001, Owen et al 1991, Stuss et al 2000) (Manes et al 2002, Owen et al 1991, Stuss et al 2000). Similar outcomes are also observed in monkeys and rats (Bissonette et al 2008, Brown & Bowman 2002). Moreover, OFC damage does not impair ED shifting, and lateral or medial PFC damage does not impair reversal learning, suggesting a functional dissociation between those regions.

6.3.1. Effects of Stress on Set-shifting

In animal models, the outcomes of experiments with acute stressors have been mixed, possibly due to varying intensity of the stress used. Thus, 30 min restraint stress did not affect set-shifting acutely (Thai et al 2013), while 15 min of tail pinch produced deficits (Butts et al 2013). Conversely, chronic stress (for example, chronic unpredictable stress (CUS) and repeated restraint stress) has been shown to produce a deficit in ED set-shifting on the AST in rodents (Bondi et al 2008, Fucich et al 2016, Jett & Morilak 2013, Liston et al 2006, Morilak et al 2005), and these behavioral changes correlate with dendritic retraction and debranching in the mPFC (Floresco et al 2008, Liston et al 2006). In humans, four weeks of psychosocial stress has been shown to alter the connectivity of DLPFC as measured by fMRI, and impair attentional set-shifting as measured by a visual discrimination task (Liston et al 2009). In another study, healthy volunteers with high self-perceived chronic stress required more time to complete the set-shifting component of the Trail Making Test (Orem et al 2008).

6.3.2. Neurochemical mechanisms and pharmacological targets

Glutamate

Glutamate transmission is necessary for set-shifting. Intra-mPFC microinjections of antagonists of either AMPA-R (LY293558, NBQX) or NMDA-R (MK80, AP-5) created deficits in set-shifting (Jett et al 2017, Stefani et al 2003, Stefani & Moghaddam 2005); in particular, NMDA-R activation is necessary for the modification of previously learned information (Stefani et al 2003). Conversely, antagonizing the metabotropic glutamate receptor mGluR5 with MPEP did not affect ED set-shifting (Jett et al 2017). Ketamine given systemically at a sub-anesthetic dose 24 hours before testing has been shown to reverse the effects of stress on ED set-shifting (Jett et al 2015). Furthermore, an intraperitoneal injection of ketamine prior to each restraint session of a chronic restraint stress procedure prevented the impairment of ED set-shifting performance (Nikiforuk & Popik 2014).

Norepinephrine

Reduction of norepinephrine in the mPFC, by either noradrenergic deafferentiation or lesions of the dorsal noradrenergic bundle, produced marked set shifting deficits in rats (McGaughy et al 2008, Tait et al 2007). The serotonin and norepinephrine reuptake inhibitor atomoxetine was able to rescue the deficits in ED caused by deafferentiation, presumably by preventing reuptake of norepinephrine released from the residual intact adrenergic terminals (Newman et al 2008). Atomoxetine also improved ED set-shifting in intact adolescent rats (Cain et al 2011). Enhancing noradrenergic neurotransmission acutely, by administration of the α₂-adrenergic autoreceptor antagonist atipamezole, improved performance on the ED task through the activation of the α₁-adrenergic receptor (Bondi et al 2010, Lapiz & Morilak 2006). Furthermore, chronic treatment with desipramine, a selective norepinephrine reuptake blocker, increases tonic noradrenergic transmission in the mPFC and enhances performance on the ED task (Lapiz et al 2007). Chronic desipramine treatment also prevents the detrimental effects of chronic stress on cognitive flexibility when administered beginning one week prior to the start of CUS (Bondi et al 2008). The SNRI milnacipran is prescribed as a treatment for fibromylagia in the United States and as an antidepressant in Europe. Although milnacipran blocks the reuptake of both norepinephrine and serotonin, it has a high preference for the norepinephrine transporter (Briley et al 1996, Mochizuki et al 2002, Moret & Briley 1997, Vaishnavi et al 2004). Chronic milnacipran treatment enhanced cognitive flexibility compromised by chronic stress (Naegeli et al 2013). However, because norepinephrine is released during stress (Nakane et al 1994) and is linked with the activation of the HPA axis (Joels & Baram 2009), the repeated elicitation of acute norepinephrine and glucocorticoid activity by chronic repeated stress may be detrimental and result in dendritic atrophy (Woolley et al 1990). Thus, while enhancing norepinephrine tone acutely in the mPFC can enhance cognition acutely, blocking α₁-adrenergic and β₂-adrenergic receptors prior to each stress session during a chronic unpredictable stress treatment protected against the detrimental effects of chronic stress on set-shifting capability (Jett & Morilak 2013).

The specific association between set-shifting and norepinephrine neurotransmission in humans is less well documented, because the most commonly used pharmaceuticals lack selectivity for the noradrenergic system. In a study of patients diagnosed with Major Depressive Disorder that compared the effects of the SNRI duloxetine with the SSRI escitalopram on a battery of cognitive tests for executive function, both patient groups improved their scores on the Hamilton Depression Rating Scale (HSRS) and on the ID/ED set-shifting test (Herrera-Guzman et al 2010). Another study also reported trends to improved ID/ED performance in individuals with major depression that were administered duloxetine; however, because of the small sample size, statistical significance of the effects compared to controls was achieved only when several measures of executive function were considered together (Greer et al 2014).

Serotonin

Although serotonin depletion has been shown to impair reversal learning but not set-shifting (Clarke et al 2005, Lapiz-Bluhm et al 2009), a role for serotonin signaling in set-shifting is implicated by the fact that SERT knockout mice perform better than control mice on the set-shifting task of AST (Nonkes et al 2012). Moreover, drugs that target serotonin reuptake can enhance cognitive function. Thus, chronic treatment with the selective 5-HT reuptake inhibitor escitalopram has been shown to reverse the effects of CUS on set-shifting, and acute administration of escitalopram reversed a deficit in set-shifting caused by repeated restraint stress (Bondi et al 2008, Nikiforuk & Popik 2011). In depressed individuals, the SSRI sertraline has been shown to enhance attentional and executive cognitive functions (Constant et al 2005).

Specific serotonin receptors appear to be linked to optimal set-shifting. In rats, the negative effects of chronic restraint stress on set-shifting were ameliorated by antagonizing the 5-HT₇ receptor, which is implicated in the cognitive and behavioral manifestation of depression, with amisulpride (Hedlund 2009, Hedlund et al 2005, Leopoldo et al 2011, Nikiforuk & Popik 2013). The 5-HT₆ receptor has also been implicated in the treatment of cognitive symptoms. In PCP-treated rats, the 5-HT₆ receptor antagonist, SB 271046, and sertindole, a second generation antipsychotic that acts as an antagonist on several neurotransmitter receptors, with high affinity for the 5-HT₆ receptor, reduced the PCP-induced deficit in set-shifting (Idris et al 2010, Rodefer et al 2008). Vortioxetine has been shown to improve a range of cognitive impairments in depressed patients (McIntyre et al 2014, Pehrson & Sanchez 2014, Sanchez et al 2015).

Dopamine

Dopamine activity in the mPFC is required for optimal performance in set-shifting tasks. Pharmacological depletion of dopamine in the mPFC of monkeys using 6-OHDA produced deficits in set-shifting (Crofts et al 2001, Robbins & Roberts 2007). In contrast, increasing dopamine activity in the mPFC improved set-shifting in humans (Apud et al 2007). The role played by different dopamine receptors in this behavior is complex, and depends on differential activation of D₁, D_2, and D₄ dopamine receptors (Floresco 2013). For example, the D₁ receptor antagonist SCH23390 infused within the mPFC impaired set-shifting in rats (Haluk & Floresco 2009, Ragozzino 2002), while mPFC infusions of the D₄ antagonist L-745,870 improved set-shifting (Floresco et al 2006). Finally, administration of the D₂/D₃ receptor antagonist sulpiride to human volunteers impaired extradimensional set-shifting (Mehta et al 1999).

6.3.3. Non-pharmacological treatments

Exposure Therapy

Cognitive behavioral therapy, such as exposure therapy, is effective in the reduction of symptoms for some treatment-resistant patients; the combination of pharmacotherapy and psychotherapy can improve symptoms even further (Beck 2005, Hollon et al 2002, Nemeroff et al 2003, Pampallona et al 2004). Prolonged exposure therapy shares some similarities to the process of fear extinction training. During fear extinction training, animals that previously have associated an aversive stimulus to an innocuous stimulus (e.g. a shock and a tone, respectively) eventually exhibit a decreased fear response when exposed to the innocuous stimulus in the absence of the aversive stimulus (Myers & Davis 2007). Recent studies have shown that fear extinction rescues a deficit in set-shifting caused by chronic stress, and that these effects are mediated by neuronal activation and de novo protein synthesis in the IL region of the mPFC (Fucich et al 2016). These findings suggest that behavior-driven activity of the mPFC produces plastic changes in this brain region that can benefit other mPFC-dependent behavioral tasks, including set-shifting.

Mindfulness

A few studies have investigated the impact of Mindfulness-Based Stress Reduction (MBSR) based training on cognitive flexibility. In healthy adults, MBSR did not improve set-shifting on the Trail Making Test (TMT) (Gallant 2016, Heeren et al 2009), or on the dual attention to response task (DART) (Jensen et al 2012). However, MBSR improves set-shifting in older adults as measured by post-test performance on the TMT (Moynihan et al 2013), suggesting that this technique may be a useful tool to improve cognitive set-shifting in compromised or aged individuals.

Electroconvulsive Therapy (ECT)

A recent meta-analysis found that patients tested before and after ECT therapy showed long-term improvements in various measures of executive functioning (Semkovska & McLoughlin 2010). In particular, data from multiple studies demonstrate that ECT improved performance in the Stroop Color-Word test and Trail Making Test Part B, two measures of cognitive flexibility. Although the authors still observed evidence of short-term cognitive deficits, these appeared to be fully recovered and even improved over time. This suggests that ECT can improve deficits in cognitive flexibility associated with psychiatric disorders.

Repetitive Transcranial Magnetic Stimulation

Repetitive TMS has also proven to be beneficial for set-shifting. In patients with concurrent Parkinson’s disease and major depressive disorder, rTMS over the left DLPFC significantly improved performance on the Stroop Test and the Wisconsin Card Sorting Test (Boggio et al 2005).

7. Concluding remarks

The impact of stress on executive function is well documented both in humans and in animal models. The consequences are particularly dreary for aging and clinical populations, where poor executive skills resulting in reduced attention, impulse control, prospective memory, self-awareness, decision making and flexibility can cause further decline and social isolation, or can exacerbate affective and neurological symptoms, as well as other medical illnesses. Compounding this is the fact that available therapies are only partially effective at ameliorating cognitive symptoms. Anatomical lesion studies and pharmacological studies have given us a wealth of insightful knowledge about the brain areas and neurotransmitter systems implicated in executive function. The future challenge is to achieve a more detailed understanding of the circuitries and molecular mechanisms involved in the behavioral manifestations of executive processes in both healthy states and in conditions impaired by stress. Doing so will allow us to identify the systems most sensitive to dysregulation that may therefore be candidate targets for new pharmacological and non-pharmacological interventions.

Highlights.

• Executive functions such as working memory, attention, behavioral inhibition and cognitive flexibility are important cognitive processes mediated in the prefrontal cortex.

• Stress impairs executive function, which can compromise adaptive behavior and lead to psychiatric pathology.

• Several neurotransmitter systems and molecular signaling pathways involved in mediating optimal executive function are altered by chronic stress.

• Understanding how chronic stress affects executive function may provide new therapeutic targets for the treatment of executive impairments in psychiatric disorders.