OCD as a Failure to Integrate Goal-Directed and Habitual Action Control

Oxford University Press
Publication Type:
Obsessive-compulsive Disorder: Phenomenology, Pathophysiology, and Treatment, 2017
Issue Date:
Full metadata record
Selecting appropriate actions in a complex world is a formidable challenge for any animal. The mammalian brain contains distinct circuits that subserve different strategies to address this challenge; adaptive behavior requires the dynamic and context-appropriate recruitment of these strategies. In particular, relatively inflexible learned behaviors—habits—may be efficient and appropriate in familiar situations in which a set pattern of behaviors has repeatedly been successful in the past, but such inflexibility may be maladaptive in novel or unpredictable circumstances in which more flexible goal-directed actions are required (chapter 16). This chapter summarizes animal work addressing the neural underpinnings of habitual and goal-directed actions and how it may inform our understanding of OCD. Over the past 10 years it has been established that distinct circuits within the cortico-basal ganglia network mediate functionally distinct forms of action control. Specifically, a medial prefrontal cortex-dorsomedial striatal network subserves goal-directed action, whereas a more lateral sensorimotor cortex-dorsolateral striatal network subserves habits (Balleine & Dickinson, 1998; Balleine & O’Doherty, 2010; Yin, Knowlton, & Balleine, 2004, 2005, 2006; Yin, Ostlund, Knowlton, & Balleine, 2005). These systems employ functionally different learning rules and structurally distinct neural mechanisms. Nevertheless, the topography of the behavioral responses controlled by the two systems can often be indistinguishable, and so simply observing or documenting discrete behaviors is often of little use in establishing what is controlling them. For example, an animal may repeatedly press a lever associated with a food reward because lever-pressing has become a habit, but it might equally well press the lever because it wants a specific food and hopes to acquire it by an action that has delivered it in the past. Behavioral pathology, whether in a natural or clinical setting, cannot be expected to map in a simple way onto functional or structural pathology. Experimental evidence, in which circuits are perturbed in a hypothesis-driven way with the consequences for learned behaviors observed, is required to associate abnormal patterns of behavior with changes in the various factors contributing to goal-directed and habitual control and their neural bases. This empirical generalization has implications for the study of various psychiatric conditions, particularly OCD. One of the puzzling features of OCD is that repetitive, ritualistic actions (i.e., compulsions) are carried out by sufferers despite their awareness that these compulsive behaviors are excessive and even ineffective. Conventional accounts of compulsions typically describe them as “safety” or “neutralizing” behaviors and assume that such behavior is performed to reduce the likelihood that an unwanted, or feared, consequence will take place. From this perspective, compulsions are similar to goal-directed actions – that is, intentional responses or choices that, even if ineffective, are motivated by a desire to avoid an unwanted outcome. An alternative view of compulsive behavior has been advanced on the basis of its apparent similarity to the loss of goal-directed control that occurs with habit formation (Gillan & Robbins, 2014; Graybiel & Rauch, 2000; Robbins, Gillan, Smith, de Wit, & Ersche, 2012). Unlike goal-directed actions, the performance of habitual responses does not depend on their specific consequences or outcomes (or the avoidance thereof), but only on their past repetition in the same situation; that is, habits are driven by antecedent stimuli rather than their consequences (Hull, 1943). On this view, compulsions are triggered automatically by environmental stimuli, regardless of their consequences. Such automaticity may help to explain why patients have so little control over compulsions, even if they have good insight. Both of these views find support in the fact that several of the structures (or their homologues) controlling goal-directed and habitual actions have been implicated in OCD in humans and in OCD-like behavior in rodents. In particular, metabolic and structural changes in a corticostriatal circuit—including the orbitofrontal and anterior cingulate cortices and the caudate nucleus—have been associated with OCD, and even identified as an “OCD circuit” (Graybiel et al., 2000; Maia, Cooney, & Peterson, 2008; Saxena, Brody, Schwartz, & Baxter, 1998) (see chapter 21). Activity in this circuit is increased during symptoms, and a decrease in this activity has been reported to accompany successful treatment (Adler et al., 2000; Baxter et al., 1992; Mataix-Cols et al., 2003, 2004; Rauch et al., 1994; Schwartz et al., 1996; Swedo et al., 1992; Whiteside, Port, & Abramowitz, 2004). This chapter briefly reviews the circuits supporting goal-directed and habitual actions as they have been defined in rodents, with a particular focus on structures that have either been directly implicated in mediating OCD-like behavior in animals, or that have had their homologies identified as relevant to OCD in humans (see chapter 20). In particular, it focuses on recent evidence from our laboratory concerning the medial portion of the OFC (mOFC) and its role in the recall of currently unobservable action outcomes. Finally, it considers how a recently described hierarchical model that we have developed in our laboratory (Dezfouli & Balleine, 2012) could potentially reconcile conflicting accounts about the relative contributions of goal-directed and habitual actions to OCD and OCD-like behavior (Robbins et al., 2012; Piantadosi & Ahmari, 2015).
Please use this identifier to cite or link to this item: