Perseveration on a reversal-learning task correlates with rates of self-directed behavior in nonhuman primates

Abstract

In humans and several nonhuman animals, repetitive behavior is associated with deficits on executive function tasks involving response inhibition. We tested for this relationship in nonhuman primates by correlating rates of normative behavior to performance on a reversal-learning task in which animals were required to inhibit a previously learned rule. We focused on rates of self-directed behavior (scratch, autogroom, self touch and manipulation) because these responses are known indicators of arousal or anxiety in primates, however, we also examined rates of other categories of behavior (e.g., locomotion). Behavior rates were obtained from 14 animals representing three nonhuman primate species (Macaca silenus, Saimiri sciureus, Cebus apella) living in separate social groups. The same animals were tested on a reversal-learning task in which they were presented with a black and a grey square on a touch screen and were trained to touch the black square. Once animals learned to select the black square, reward contingencies were reversed and animals were rewarded for selecting the grey square. Performance on the reversal-learning task was positively correlated to self-directed behavior in that animals that exhibited higher rates of self-directed behavior required more trials to achieve reversal. Reversal learning was not correlated to rates of any other category of behavior. Results indicate that rates of behavior associated with anxiety and arousal provide an indicator of executive function in nonhuman primates. The relationship suggests continuity between nonhuman primates and humans in the link between executive functioning and repetitive behavior.

Introduction

In humans, control of reversal learning and other executive functions (EF) involving response inhibition and set-shifting, appear to be localized in various regions of the frontal lobes: the orbitofrontal (OFC) and prefrontal cortices (especially the ventral medial PFC), the dorsolateral prefrontal cortex, and the anterior cingulate cortex (ACC) [1], [2], [3], [4], [5]. These structures project to regions of the limbic system and parts of the basal ganglia, forming multiple cortico-striatal-thalamo-cortical “loops” [6]. For example, OFC and ACC are richly connected to the amygdala [7], which controls affective states such as fear and anxiety. The amygdala and orbital prefrontal cortex comprise a neural pathway that governs a variety of adaptive responses and decision-making processes [7]. Additionally, these corticolimbic regions are linked to various portions of the striatum, which are involved in habit formation, procedural memory, and reversal learning [8], [9], [10], [11], [12]. The prefrontal cortices serve to inhibit prepotent or previously learned motor habits that are governed by the striatum. When functioning normally, these so-called cortical-striatal-thalamo-cortical loops inhibit inappropriate behavior that was once reinforced and facilitate the flexibility to learn more appropriate responses [6], [7], [13]. When not functioning normally, these neural systems or loops are associated with disorders such as obsessive–compulsive disorder (OCD), Tourette syndrome, and schizophrenia that involve disinhibition and perseverative behavior as well as anxiety [7], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25]. The inability to inhibit behavior produces deficits on tasks such as reversal learning, that require inhibition of previously learned responses [11], [13], [24], [26], [27], [28], [29].

Monkeys and rats (Rattus norvegicus) appear to have circuitry similar to humans with highly specific projections from the prefrontal cortex to the striatum [30], [31], [32]. The striatum in monkeys and rats has been linked to the propagation of repetitive behavior [33], [34]. Imaging work has shown that the prefrontal cortex is active during reversal-learning and rule-switching tasks in nonhuman primates [35]; lesions to the prefrontal cortex in monkeys and rats produce decrements in reversal learning and other rule-changing tasks [7], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46]. Lesions to the medial [17] and dorsomedial striatum [47], [48] have also been linked to interruptions on reversal-learning performance in monkeys and rats (respectively), that are similar to those observed when the OFC is lesioned, highlighting the rich connections between orbitofrontal, limbic and striatal regions, as well as their relative roles in emotion, learning, and behavioral inhibition.

Garner [49] and Garner et al. [50] have emphasized that similar system of neural structures is responsible for the link between repetitive behavior and response inhibition in nonhuman animals. Captive animals can exhibit stereotypic, repetitive pacing, rocking, chewing, sucking, hair (or feather) pulling, self-biting and a variety of other abnormal responses [51], [52], [53] that may be caused by several factors including stressful experimental situations, impoverished environments, and social isolation, such as early separation from the mother in primates [52], [53], [54], [55], [56]. The occurrence of such repetitive behavior in animals is linked to deficits in the inhibition of learned responses (Garner, see review [51]). For example, bears (Ursus thibetanus and Helarctos malayanus) that engaged in higher levels of repetitive stereotypic behavior also perseverated longer with unrewarded responses during extinction on a spatial learning task [57]. Birds (Parus caeruleus and P. palustris) exhibiting more stereotypic repetitive behavior also showed more perseveration during extinction of a spatial learning task [58]. Among rodents, stereotypy in bank voles (Clethrionomys glareolus) was positively correlated to the time to extinguish a learned response [59].

The relationship between rates of behavior and executive function has not been studied thoroughly in nonhuman primates. Gluck and Sackett [60] showed that rhesus monkeys raised in social isolation, and, thus, likely to exhibit severe stereotypies [61], were more resistant to extinction on a lever-pressing task than socially-reared animals. Lutz et al. [55] showed that rhesus monkeys that did not inhibit self-injurious behavior were also more resistant to extinction than other monkeys. When amphetamine was administered to common marmosets (Callithrix jacchus) stereotyped behavior increased [62]. In a related study, [63] amphetamine disrupted reversal learning, but not other forms of learning. We further investigated the relationship between rates of behavior and reversal learning in nonhuman primates.

We focused on variable rates of normative behavior because scratching, self-touching, and other repetitive motor behavior are normal phenomena throughout typical development [64], [65]. Many typically developing children, for example, exhibit repetitive stereotyped behavior [65], [66], [67] and repetitive body-focused behavior such as fingernail biting, skin picking, lip chewing, and hair-twirling are prevalent in nonclinical samples [68], [69]. Further, repetitive behavior has been linked to deficits in inhibitory executive function in typically developing children [11]. Normal children less than six years of age who were reported to engage in a relatively high degree of repetitive behavior performed more poorly on response inhibition tasks than children who exhibited less compulsive behavior [11], [70]. Similarly, scores on an obsessive–compulsive inventory collected on a non-clinical sample were positively correlated with perseveration on a rule-changing task [71]. In a nonhuman normative example, piglets (Sus domestica) identified as using less flexible suckling positions as piglets made more errors in a reversal-learning task as adults than piglets identified as using more variable suckling positions [72].

In this exploratory study, our strategy was to correlate response inhibition on a rule-changing task to rates of a range of normative behavior. We employed a reversal-learning task in which an animal was presented with a black and a grey square on a touch screen and was first trained to select the black square. After acquisition, reward contingencies were reversed and monkeys were rewarded for touching the grey square. We recorded the number of trials necessary for the monkeys to switch to the new rule. We selected this reversal-learning task to maximize the number of subjects by utilizing a fairly simple learning task that many animals might be able to perform. The task was also quite similar to those often used to study human reversal learning.

For normative behavior, we focused on a class of behavior referred to as displacement activities or self-directed behavior (SDB) in nonhuman primate studies. SDB include responses such as scratching and self-touching that numerous studies of nonhuman primates have shown to be correlated with emotional arousal and anxiety [73], [74], [75], [76], [77], [78], [79], [80], [81]. Further, pharmacological studies have shown that nonhuman primates administered anti-anxiety medication typically prescribed to humans, reduce rates of SDB without changing rates of other classes of behavior [82], [83]. Humans also appear to engage in more SDB in anxiety promoting situations. For example, people with alexithymic traits, in which they have difficulty describing their emotions, increase rates of SDB when asked to describe their feelings during interviews [84], [85], [86]. Increases in SDB were also observed at the point patients began speaking about anxiety-inducing topics during interviews [87]. Since the relationship between behavioral patterns (especially repetitive behavior) and executive function (such as reversal learning) is also associated with anxiety states, suggesting shared neural circuitry linking pre- and orbito-frontal, limbic and striatal regions, we predicted that a relationship between reversal learning and normative behavior would include SDB.

We also collected frequencies of other behavioral responses to determine if any relationships observed were specific to SDB, the responses known to be indicative of arousal or anxiety in primates. However, other types of behavior indicative of arousal or repetition might also be correlated to reversal-learning performance. For example, general locomotion was examined (e.g., walking, running, climbing) as a subcategory of potentially repetitive behavior. Other general classes of behavior examined were vocalization, social behavior, and “other” self-stimulating behavior, responses that might be considered forms of SDB but otherwise have not been typically included in studies investigating SDB in primates (e.g., rubbing hands together).

We tested whether nonhuman primates that tended to perseverate on the reversal-learning task exhibited higher rates of self-directed behavior and other types of behavior. We also tested the link between rates of behavior and performance on the acquisition task (learning to touch the black square) because the relationship between behavior and cognitive function is reported to be specific to inhibitory executive function tasks such as reversal learning. If so, we would expect a relationship between behavior rates and reversal learning, but not between behavior rates and the discriminative task learned during acquisition. Results would provide a direct test of the relationship between behavior and inhibitory executive function in nonhuman primates. Results might also suggest continuity between normative behavior and the more extreme types of stereotypic behavior sometimes observed in animals.