Abstract

Compulsive behavior, which MeSH classifies under impulsive behavior, is repetitive, rigid action continued despite being no longer rewarding and often in the face of harm. Cognitive psychology explains it through two complementary accounts: the habit account, which casts compulsion as over-reliance on the stimulus–response habit system shown by insensitivity to outcome devaluation in obsessive-compulsive disorder, and the distress-reduction account, which casts it as negatively reinforced avoidance maintained because it terminates anxiety. Both converge on weakened top-down prefrontal control over cortico-striatal loops, and both motivate exposure-based treatment. Contemporary dimensional models place compulsivity and impulsivity at opposite poles of a single under- to over-controlled axis cutting across diagnoses from OCD to addiction. Three interactive demonstrations trace outcome devaluation, the anxiety–compulsion reinforcement cycle, and the impulsivity–compulsivity spectrum.

Keywords: compulsivity, habit, goal-directed control, negative reinforcement, obsessive-compulsive disorder

A person washes their hands until the skin cracks, returns a seventh time to check a locked door, or keeps gambling long after the money and the pleasure are gone. What unites these acts is not their content but their grammar: they are repeated, they are hard to resist, and they continue even as the actor recognises them as excessive or senseless. The puzzle compulsive behavior poses to cognitive psychology is precisely this persistence-against-interest. A rational agent repeats an action because it pays; a compulsive one repeats it when it has stopped paying, and sometimes because stopping has become intolerable (Luigjes et al., 2019).

For most of the twentieth century compulsion was described clinically — as a symptom of obsessive-compulsive disorder, to be catalogued and treated — rather than explained mechanistically. The last two decades have changed that. By borrowing the experimental distinction between goal-directed and habitual action from animal learning theory, and by mapping both onto cortico-striatal circuitry, researchers have turned compulsivity from a descriptive label into a measurable disturbance of the control of action (Gillan et al., 2016; Robbins et al., 2024).

Key Takeaways
  • Compulsive behavior is repetitive action that persists despite being no longer rewarding — the defining contrast with ordinary goal-directed behavior.
  • The habit account explains it as over-reliance on the stimulus–response habit system, shown by insensitivity to outcome devaluation in OCD.
  • The distress-reduction account explains it as negative reinforcement: the compulsion is maintained because it relieves anxiety, which is why exposure and response prevention works.
  • Both accounts converge on weakened top-down prefrontal control over cortico-striatal loops linking orbitofrontal cortex, caudate, and thalamus.
  • Dimensional models place compulsivity and impulsivity at opposite poles of one under- to over-controlled dimension, yet the two frequently co-occur because they share a failure of cognitive control.

What Compulsive Behavior Is

Compulsive behavior is most usefully defined by a contrast with the ordinary logic of reward. Normal instrumental action is goal-directed: we perform it because we expect it to produce an outcome we currently value, and we stop when the outcome loses its value or the action stops delivering it. Compulsive action breaks this contingency. It is performed repetitively and rigidly, it is experienced as difficult to control, and — the diagnostic core — it continues even when it no longer produces reward and may actively produce harm (Luigjes et al., 2019). A recent synthesis distilled four recurring criteria from across the literature: repetition, the subjective quality of being driven or irresistible, the absence of a rational relationship to the stated goal, and persistence despite negative consequences.

This definition deliberately spans more than obsessive-compulsive disorder. The same functional signature appears in the checking and washing rituals of OCD, in the compulsive drug-seeking of addiction, in pathological gambling, and in disorders of eating and skin-picking. That breadth is the reason compulsivity has become a transdiagnostic construct rather than the property of a single diagnosis (Robbins et al., 2012; Fineberg et al., 2010). What differs across these conditions is the content of the act and its initial motivation; what they share is the mechanism by which an act, once adaptive or at least rewarding, becomes decoupled from its consequences and runs on regardless.

Figure 1

Goal-Directed Versus Compulsive Control of Action

Goal-directed versus compulsive control of action Two loops. In the goal-directed loop an action is evaluated against its outcome and stops when the outcome is devalued. In the compulsive loop the action is triggered directly by a stimulus and persists after the outcome is devalued. Goal-directed action → outcome valued? if devalued → stop Compulsive stimulus → action persists although outcome devalued the same act, two kinds of control
Note. The behavioural test that separates the two loops is outcome devaluation: a goal-directed action drops immediately when its outcome is made worthless, whereas a habitual or compulsive action, triggered directly by the eliciting stimulus, continues. Demo 1 runs this test interactively.

Types of Compulsive Behavior

In the Medical Subject Headings vocabulary, Compulsive Behavior (D003192) is filed as a narrower descriptor of impulsive behavior (D007175), and MeSH records one narrower descriptor beneath it in turn. MeSH is an indexing classification built for retrieving the biomedical literature, not a theory of how compulsion is organised; its hierarchy reflects historical indexing practice rather than the mechanistic distinctions the experimental literature draws, and the two ways of carving the domain are orthogonal and should not be conflated.

Narrower MeSH descriptor What it denotes
Behavior, Addictive The compulsive, relapsing pursuit of a substance or activity despite adverse consequences; MeSH places addictive behavior as a kind of compulsive behavior, reflecting the shared transition from reward-driven to habitual, consequence-insensitive action.

Addictive Behavior does not yet have its own article on this site, so the term is listed here but not linked. The placement is itself a theoretical claim worth noting: by nesting addiction under compulsion, MeSH encodes the influential view that chronic drug-seeking is the end state of a slide from actions to habits to compulsion, in which behaviour initially controlled by its rewarding outcome is progressively taken over by stimulus–response habits and finally becomes compulsive (Everitt & Robbins, 2005).

Two Accounts of Compulsion

Cognitive psychology offers two accounts of how an action becomes compulsive. They are not rivals so much as descriptions of the same disturbance from the standpoints of learning and of motivation, and both locate the fault in the balance between a controlled, goal-sensitive system and an automatic one, implemented across the cortico-striatal loops linking orbitofrontal cortex, caudate nucleus, and thalamus (Graybiel & Rauch, 2000; Pauls et al., 2014).

The habit account: when goals give way to habits

Instrumental behaviour is controlled by two dissociable systems. A goal-directed system selects actions by their expected consequences and the current value of those consequences; a habit system selects actions that past reinforcement has stamped in, triggered directly by the eliciting stimulus and indifferent to the current worth of the outcome. The experimental assay that tells them apart is outcome devaluation: after training, the outcome is made worthless (a food is fed to satiety, or paired with illness), and the question is whether the animal or person stops performing the action. A goal-directed responder stops; a habitual one keeps going (Everitt & Robbins, 2005).

The habit account of compulsion holds that compulsive behavior is an over-reliance on this habit system. The decisive evidence is that patients with obsessive-compulsive disorder show reduced sensitivity to outcome devaluation in the laboratory — they go on making a trained response after its outcome has been devalued, and this excessive habit formation is accompanied by hyperactivity in the caudate (Gillan et al., 2011; Gillan & Robbins, 2014). On this view the clinical compulsion is not driven primarily by the obsession it is rationalised as serving; rather, an over-strong habit is executed automatically, and the obsessive thought is in part a post-hoc explanation for an urge the person cannot otherwise account for. The same over-reliance on habit at the expense of goal-directed control has since been found across compulsive disorders, which is the empirical backbone of the transdiagnostic view (Gillan et al., 2016).

The outcome-devaluation test

A trained responder blends two systems. A goal-directed system presses only while the outcome is valued; a habit system presses automatically, whatever the outcome is now worth. Set how far the responder relies on habit, then devalue the outcome (feed to satiety). A goal-directed agent stops; a habitual one keeps going — and that residual responding is the laboratory signature of compulsion.

Reliance on the habit system (w)0.30
0.00.51.0respondingbefore1.00after (not devalued)1.00
With habit weight w = 0.30 this responder is mixed control. After the outcome is devalued, responding falls to 0.30 — only the habitual share persists, so the devaluation reveals exactly how much of the behaviour had slipped out of goal-directed control.

The distress-reduction account: negative reinforcement

The second account starts from motivation rather than learning, and it is the dominant clinical model of OCD. Here a compulsion is maintained by negative reinforcement: an intrusive thought (the obsession) provokes anxiety, the compulsive act (washing, checking) terminates that anxiety, and the relief reinforces the act, making it more likely next time (Stein, 2002). The tragedy of the cycle is that the relief is short-lived and the long-run effect is to entrench the obsession, because the person never discovers that the feared outcome would not have occurred anyway.

This account is not merely descriptive; it generates the most effective psychological treatment for OCD, exposure and response prevention (ERP). The patient is exposed to the anxiety-provoking trigger and prevented from performing the compulsion, and across repeated exposures the anxiety habituates — it rises less steeply and falls faster — until the compulsion is no longer needed to end it. The worked example below quantifies that habituation curve, and Demo 2 drives the cycle with and without response prevention.

The anxiety–compulsion cycle and ERP habituation

A trigger drives subjective distress (SUDS, 0–100) to a first-exposure peak of 80. Under response prevention the compulsion is withheld, the exposure runs to completion, and the peak decays across exposures as Pn = 80(1 − h)n. Turn prevention off and the compulsion is performed each time, aborting the exposure at its peak so distress never habituates.

Habituation fraction per exposure (h)0.25
020406080100half of peakpeak distress (SUDS)exposure number (n)012345678
With prevention and h = 0.25, peak distress runs 80 → 60 → 45 → 33.75… across exposures 0–3. Distress first halves at n = 2.41 (so by the third exposure). Each withheld compulsion teaches that the anxiety subsides on its own, breaking the negative-reinforcement contingency.

The impulsivity–compulsivity dimension

Although MeSH files compulsive behavior under impulsive behavior, the experimental literature increasingly treats the two as opposite poles of a single dimension of behavioural control that runs from under-regulated to over-regulated action (Robbins et al., 2012; Dalley et al., 2011). At the impulsive pole, action is premature and driven by the pull of immediate reward; at the compulsive pole, action is rigid, repetitive, and driven by the avoidance of harm. The contrast is summarised as reward-seeking versus harm-avoidance, and as positive versus negative reinforcement.

Yet the poles are not simply unrelated. Impulsivity and compulsivity frequently co-occur in the same individuals and disorders, and a leading hypothesis is that impulsivity is a predisposing endophenotype that can, through repeated reinforcement and the shift to habitual control, give way to compulsivity over the course of a disorder — the same actions-to-habits-to-compulsion trajectory seen in addiction (Fineberg et al., 2010). Both poles, on this account, express a common failure of top-down cognitive control, which is why they share neural substrates and risk factors even as their surface behaviour looks opposite. Demo 3 moves a behaviour along this spectrum and reads off its changing drivers.

One axis, two poles

The dimensional model places impulsivity and compulsivity at opposite ends of a single under- to over-controlled axis. Slide a behaviour from one pole to the other and watch its drivers swap — reward-seeking for harm-avoidance, positive for negative reinforcement — while a shared failure of top-down control stays constant, which is why the "opposite" poles so often co-occur.

Position on the axisimpulsive 0%
ImpulsiveCompulsivepremature · reward-seekingrigid · harm-avoidantpositive reinforcementnegative reinforcement
At the mid-axis (mixed): the dominant driver is harm-avoidance, the action is rigid / cannot stop, and it is maintained by negative (relief) reinforcement (reward-seeking 50% vs harm-avoidance 50%). The shared failure of top-down cognitive control stays at ~60% the whole way across — the common substrate that lets a behaviour drift from one pole toward the other over the course of a disorder.

Worked Example

Consider the anxiety–compulsion cycle of the distress-reduction account, and the habituation that exposure and response prevention is designed to produce. Suppose a patient with a contamination obsession reaches a peak subjective distress of 80 units (on the 0–100 SUDS scale clinicians use) the first time they touch a surface they believe to be contaminated and are prevented from washing.

Habituation across exposures is well modelled as a geometric decay of the peak: on the n-th exposure the peak distress is

Pn = P0 (1 − h)n

where P0 = 80 is the first-exposure peak and h is the habituation fraction per exposure. Take h = 0.25:

- P1 = 80 × (0.75)1 = 60 - P2 = 80 × (0.75)2 = 45 - P3 = 80 × (0.75)3 = 33.75

By the third exposure the peak distress has fallen below half of its original value. We can find the exposure at which distress first halves by setting Pn = 40 and solving for n:

(0.75)n = 0.5 ⟹ n = ln(0.5) / ln(0.75) ≈ 2.41

so distress has more than halved by the third exposure (n = 3). This is the engine of ERP: each time the compulsion is withheld, the patient learns that the anxiety subsides on its own, and the negative-reinforcement contingency that maintained the compulsion is broken. Contrast the opposite policy of performing the compulsion every time: each wash aborts the exposure at its peak, no habituation occurs, P stays at 80, and the cycle is preserved indefinitely. Demo 2 traces both policies across trials, reproducing exactly these numbers.

Discussion

The two accounts of compulsion are now usually read as complementary levels of a single story. The distress-reduction model explains the motivation that launches a compulsion and the clinical fact that relieving anxiety treats it; the habit model explains how an initially purposeful act becomes automatic and consequence-insensitive, and why compulsions persist even when the original anxiety has faded and the person wants to stop (Gillan et al., 2016; Robbins et al., 2019). A checking ritual may begin as anxiety-driven and goal-directed, then, through repetition, slide into a stimulus-bound habit that outlives its original rationale. Treatments map onto the levels: ERP attacks the reinforcement contingency, while the habit framing motivates interventions that strengthen goal-directed control or disrupt the triggering stimulus.

The dimensional reframing carries a sharp implication for psychiatry. If compulsivity is a measurable disturbance of the balance between goal-directed and habitual control, and if that disturbance cuts across OCD, addiction, and the eating disorders, then the traditional diagnostic categories carve the problem at the wrong joints (Robbins et al., 2012). A transdiagnostic marker — the degree of over-reliance on habit, or the position of a behaviour on the impulsive–compulsive axis — may predict treatment response better than the diagnosis does. That is a promise the field is still testing, and the mechanistic heterogeneity within each disorder is the main obstacle.

Current Directions

Three fronts are active. The first is compulsivity as a computational phenotype: rather than scoring a single devaluation test, researchers fit reinforcement-learning models that separate a model-based (goal-directed) from a model-free (habitual) controller, and estimate each person's reliance on the two. Reduced model-based control has emerged as a candidate transdiagnostic marker of compulsivity, measurable online at large scale. The second is the neural circuit work consolidating the orbitofrontal–striatal account into a mechanistic model of how compulsive disorders arise and how circuit-level interventions — including deep brain stimulation of the striatum — might correct them (Robbins et al., 2024). The third returns to first principles: because compulsivity is still defined differently across subfields, a concerted effort is under way to pin down the construct operationally so that findings in OCD, addiction, and animal models can be compared on the same measurements (Luigjes et al., 2019). Whether a single, mechanistically grounded measure of compulsivity can be agreed — and shown to predict who becomes ill and who recovers — is the open question the next decade must answer.

Common Misconceptions

Compulsions are just strong habits.
A habit is consequence-insensitive but harmless; a compulsion is a habit that persists despite harm and is experienced as difficult to resist. Excessive habit formation is one mechanism of compulsion, not a synonym for it (Gillan & Robbins, 2014).
People perform compulsions because they enjoy them.
Compulsions are typically maintained by relief, not pleasure — negative reinforcement, the termination of anxiety, rather than positive reward. This is why they continue after any pleasure is gone (Stein, 2002).
Compulsivity is the same thing as impulsivity.
On the dominant dimensional model they are opposite poles — under-controlled, reward-driven action versus over-controlled, harm-avoidant action — even though they share a failure of top-down control and often co-occur (Robbins et al., 2012).
Compulsive behavior means obsessive-compulsive disorder.
OCD is one home of compulsion, but the same functional signature appears in addiction, gambling, and eating and body-focused disorders; compulsivity is transdiagnostic (Fineberg et al., 2010).

Glossary

Compulsion versus compulsivity.
A compulsion is a specific repetitive act; compulsivity is the underlying trait-like tendency, measurable as over-reliance on habitual control, that disposes to such acts across disorders.
Cortico-striatal loops.
Parallel circuits connecting regions of cortex (notably orbitofrontal cortex) through the striatum (caudate, putamen) and thalamus back to cortex; their dysregulation is the shared neural substrate of compulsive disorders.
Endophenotype.
A heritable, quantifiable trait lying between genes and a clinical diagnosis; impulsivity is proposed as a predisposing endophenotype that can give way to compulsivity over the course of a disorder.
Exposure and response prevention (ERP).
The first-line behavioural therapy for OCD: the patient is exposed to the trigger and prevented from performing the compulsion, allowing anxiety to habituate and breaking the negative-reinforcement cycle.
Goal-directed control.
Action selection governed by the expected value of the outcome; it is sensitive to outcome devaluation and is the system compulsion is said to under-use.
Habit.
A stimulus–response association, stamped in by past reinforcement, that is triggered automatically by the eliciting stimulus and is insensitive to the current value of the outcome.
Habituation.
The decline of a response, here anxiety, on repeated exposure to the same stimulus; the learning process ERP exploits.
Impulsivity.
A tendency toward premature, poorly considered action driven by the pull of immediate reward; on the dimensional model it is the under-controlled pole opposite compulsivity.
Model-based versus model-free control.
Two reinforcement-learning controllers: model-based control plans using an internal model of action–outcome structure (goal-directed), while model-free control repeats previously reinforced actions (habitual). Reduced model-based control is a candidate marker of compulsivity.
Negative reinforcement.
Strengthening of a behaviour by the removal of an aversive state; the relief of anxiety that maintains a compulsion is the canonical example.
Obsessive-compulsive disorder (OCD).
A disorder characterised by intrusive obsessions and the compulsions performed to neutralise them; the prototypical clinical home of compulsive behavior.
Outcome devaluation.
The experimental test that distinguishes goal-directed from habitual action by making the outcome worthless and seeing whether responding stops; compulsive responders show reduced sensitivity to it.
Subjective units of distress (SUDS).
A 0–100 self-report scale of momentary anxiety used in exposure therapy to track how distress rises and habituates across repeated exposures.
Transdiagnostic.
Describing a process, such as compulsivity, that is elevated across many distinct psychiatric diagnoses rather than being specific to one.

Key Researchers

Naomi A. Fineberg

(University of Hertfordshire and Hertfordshire Partnership University NHS Foundation Trust). A leading clinical researcher on obsessive-compulsive and related disorders who helped establish the compulsive–impulsive spectrum as a framework spanning animal models and human endophenotypes. Faculty

Claire M. Gillan

(School of Psychology and Trinity College Institute of Neuroscience, Trinity College Dublin). Provided the key laboratory evidence that OCD involves excessive habit formation and reduced goal-directed control, and pioneered large-scale online computational phenotyping of compulsivity. Faculty · Google Scholar

Ann M. Graybiel

(McGovern Institute for Brain Research, Massachusetts Institute of Technology). Mapped the basal-ganglia circuitry of habit learning and its relevance to the repetitive behaviours of obsessive-compulsive disorder. ORCID · Wikipedia

Trevor W. Robbins

(Department of Psychology, University of Cambridge). With Everitt, framed the influential actions to habits to compulsion model of how reward-driven behaviour becomes compulsive, and leads the dimensional account of impulsivity and compulsivity in psychiatry. ORCID · Wikipedia

Barbara J. Sahakian

(Department of Psychiatry, University of Cambridge). Developed neurocognitive tests of compulsivity and impulsivity and their application to OCD, addiction, and treatment, central to the endophenotype programme. Faculty · Wikipedia

Dan J. Stein

(Department of Psychiatry and Mental Health, University of Cape Town; 1962–2025). A prolific clinical scientist of obsessive-compulsive and anxiety disorders whose work shaped the modern understanding and classification of OCD. Wikipedia

Frequently Asked Questions

What makes a behavior compulsive rather than just a strong habit?

A habit is automatic but harmless and can be dropped once it stops paying. A behavior is compulsive when it is repetitive, feels difficult to resist, and persists despite no longer being rewarding and even causing harm. Over-reliance on habit is one mechanism of compulsion, but the defining feature is persistence against the person's own interest.

Are compulsions driven by pleasure?

Usually not. Most compulsions are maintained by relief rather than reward: they terminate anxiety or distress, which negatively reinforces them. This is why they continue long after any pleasure has gone, and why treatments that let anxiety subside on its own are effective.

How is compulsive behavior different from impulsive behavior?

On the dominant dimensional model they sit at opposite poles of one control dimension: impulsive action is premature and reward-seeking, compulsive action is rigid and harm-avoidant. They look opposite but share an underlying failure of top-down cognitive control, which is why the same person can show both.

What is outcome devaluation and why does it matter?

It is the laboratory test that separates goal-directed from habitual action: after an outcome is made worthless, a goal-directed responder stops but a habitual one continues. People with OCD show reduced sensitivity to devaluation, which is the main evidence that compulsion involves excessive habitual control.

Why does exposure and response prevention work?

Because it breaks the negative-reinforcement cycle. By exposing the patient to the trigger and preventing the compulsion, the therapy lets anxiety habituate, falling on its own across repeated exposures, so the person learns the compulsion is unnecessary and the feared outcome does not occur.

Is addiction a compulsive behavior?

Increasingly it is modelled as one. The actions-to-habits-to-compulsion account holds that drug-seeking begins as goal-directed and reward-driven, becomes habitual with repetition, and finally becomes compulsive (consequence-insensitive and hard to stop), which is why MeSH files addictive behavior under compulsive behavior.

Is compulsivity found only in OCD?

No. The same functional signature (repetitive, driven, consequence-insensitive action) appears in addiction, pathological gambling, and disorders of eating and skin-picking. This breadth is why compulsivity is now studied as a transdiagnostic construct rather than a symptom of one disorder.

Can compulsive behavior be measured objectively?

Partly. Beyond clinical interview, researchers use outcome-devaluation tasks and reinforcement-learning models that estimate a person's reliance on goal-directed versus habitual control. Reduced goal-directed (model-based) control is a candidate objective, transdiagnostic marker of compulsivity, though no single measure is yet agreed.

References

Dalley, J. W., Everitt, B. J., & Robbins, T. W. (2011). Impulsivity, compulsivity, and top-down cognitive control. Neuron, 69(4), 680–694. https://doi.org/10.1016/j.neuron.2011.01.020

Everitt, B. J., & Robbins, T. W. (2005). Neural systems of reinforcement for drug addiction: From actions to habits to compulsion. Nature Neuroscience, 8(11), 1481–1489. https://doi.org/10.1038/nn1579

Fineberg, N. A., Potenza, M. N., Chamberlain, S. R., Berlin, H. A., Menzies, L., Bechara, A., Sahakian, B. J., Robbins, T. W., Bullmore, E. T., & Hollander, E. (2010). Probing compulsive and impulsive behaviors, from animal models to endophenotypes: A narrative review. Neuropsychopharmacology, 35(3), 591–604. https://doi.org/10.1038/npp.2009.185

Gillan, C. M., Papmeyer, M., Morein-Zamir, S., Sahakian, B. J., Fineberg, N. A., Robbins, T. W., & de Wit, S. (2011). Disruption in the balance between goal-directed behavior and habit learning in obsessive-compulsive disorder. American Journal of Psychiatry, 168(7), 718–726. https://doi.org/10.1176/appi.ajp.2011.10071062

Gillan, C. M., Robbins, T. W., Sahakian, B. J., van den Heuvel, O. A., & van Wingen, G. (2016). The role of habit in compulsivity. European Neuropsychopharmacology, 26(5), 828–840. https://doi.org/10.1016/j.euroneuro.2015.12.033

Gillan, C. M., & Robbins, T. W. (2014). Goal-directed learning and obsessive-compulsive disorder. Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1655), 20130475. https://doi.org/10.1098/rstb.2013.0475

Graybiel, A. M., & Rauch, S. L. (2000). Toward a neurobiology of obsessive-compulsive disorder. Neuron, 28(2), 343–347. https://doi.org/10.1016/s0896-6273(00)00113-6

Luigjes, J., Lorenzetti, V., de Haan, S., Youssef, G. J., Murawski, C., Sjoerds, Z., van den Brink, W., Denys, D., Fontenelle, L. F., & Yücel, M. (2019). Defining compulsive behavior. Neuropsychology Review, 29(1), 4–13. https://doi.org/10.1007/s11065-019-09404-9

Pauls, D. L., Abramovitch, A., Rauch, S. L., & Geller, D. A. (2014). Obsessive-compulsive disorder: An integrative genetic and neurobiological perspective. Nature Reviews Neuroscience, 15(6), 410–424. https://doi.org/10.1038/nrn3746

Robbins, T. W., Gillan, C. M., Smith, D. G., de Wit, S., & Ersche, K. D. (2012). Neurocognitive endophenotypes of impulsivity and compulsivity: Towards dimensional psychiatry. Trends in Cognitive Sciences, 16(1), 81–91. https://doi.org/10.1016/j.tics.2011.11.009

Robbins, T. W., Vaghi, M. M., & Banca, P. (2019). Obsessive-compulsive disorder: Puzzles and prospects. Neuron, 102(1), 27–47. https://doi.org/10.1016/j.neuron.2019.01.046

Robbins, T. W., Banca, P., & Belin, D. (2024). From compulsivity to compulsion: The neural basis of compulsive disorders. Nature Reviews Neuroscience, 25(5), 313–333. https://doi.org/10.1038/s41583-024-00807-z

Stein, D. J. (2002). Obsessive-compulsive disorder. The Lancet, 360(9330), 397–405. https://doi.org/10.1016/S0140-6736(02)09620-4