Compulsive Behaviors and Deep Brain Stimulation: The Science of Cravings, OCD, and Behavioral Control

Knowing that a behavior is harmful does not always make it possible to stop. Someone with obsessive-compulsive disorder may check a locked door dozens of times and still feel compelled to check again. Someone with severe binge eating disorder may recognize the circumstances that trigger a binge and still be unable to interrupt it.

That gap between awareness and control is central to understanding compulsive behavior. It also raises a question: Can we identify the brain activity that precedes a harmful urge and intervene before that urge becomes action?

Deep brain stimulation and other forms of neuromodulation offer ways to investigate that question. In my discussion with Dr. Casey Halpern, a functional neurosurgeon at Penn Medicine, we explored what established treatments for movement disorders can teach us about OCD, cravings, binge eating, and impulsivity. The promise is substantial, but it is important to distinguish treatments already used clinically from approaches still being tested.

Table of Contents

• How Deep Brain Stimulation Changes Brain Function
• OCD: When Repetition Becomes a Disorder
• The Brain Circuits Linking OCD, Reward, and Compulsion
• Can Researchers Detect a Craving Before a Binge?
• Non-Invasive Brain Stimulation: TMS and Focused Ultrasound
• What Epilepsy Research Can Teach Us About Psychiatric Treatment
• Why Awareness Helps, but May Not Be Enough
• Could AI Predict Harmful Impulses Before We Notice Them?
• The Goal Is Better Control, Not the Elimination of Desire

How Deep Brain Stimulation Changes Brain Function

Neurosurgery encompasses far more than operating on the brain. Neurosurgeons treat brain tumors, aneurysms, traumatic injuries, spinal conditions, and peripheral nerve disorders. Functional neurosurgery is a specialized part of that field, focused on changing the activity of neural circuits to improve symptoms.

Deep brain stimulation, or DBS, involves placing a thin, insulated electrode into a carefully selected brain region. Small contacts near the end of the electrode deliver electrical stimulation.

The electrode is the delivery system. The electrical stimulation is the treatment. One useful analogy is a device that delivers medication, except that the therapeutic input is electricity rather than a chemical compound.

For certain movement symptoms, the effects can be remarkably fast. A person who has lived with tremor for years may experience immediate relief when stimulation is activated in the appropriate circuit. That striking relationship between circuit activity and symptoms helped inspire Halpern to pursue neurosurgery.

Why a movement treatment can also affect mood

Brain circuits are closely interconnected. An electrode positioned to influence movement may sit only millimeters from pathways involved in emotion. Stimulating nearby tissue can sometimes produce a brief experience of laughter or panic. In such cases, clinicians can turn off the stimulation at that contact.

These observations are not simply side effects to avoid. They also provide clues about how the brain works. Some patients receiving DBS for Parkinson’s disease report changes in mood, compulsive behavior, or problematic gambling.

There can be more than one explanation. Relieving a disabling tremor could improve mood indirectly. But stimulation may also influence emotional, or limbic, circuits more directly. Understanding those distinctions is part of developing treatments that target psychiatric symptoms deliberately rather than incidentally.

OCD: When Repetition Becomes a Disorder

Being meticulous, safety-conscious, or committed to a routine is not the same as having obsessive-compulsive disorder. Some degree of repetitive checking can serve a useful purpose. A surgeon should be exacting about safety procedures.

The critical distinction is whether thoughts and behaviors become difficult to control and interfere with life. In OCD, obsessions can generate distress and a powerful drive to perform compulsions. Those compulsions may provide temporary relief while consuming increasing amounts of time and attention.

Examples include checking a home for safety until the early morning despite needing sleep, or repeatedly washing because a sense of contamination will not resolve. These are not simply preferences for cleanliness or order. The behavior continues despite meaningful costs.

For a broader clinical overview, the National Institute of Mental Health provides information on OCD symptoms and treatment.

Medication and exposure and response prevention

Several nonsurgical treatments can be very helpful for OCD:

• Selective serotonin reuptake inhibitors, or SSRIs: Common first-line medications that act on serotonin signaling.
• Certain tricyclic antidepressants: Another medication option discussed by Halpern.
• Exposure and response prevention, or ERP: A specialized behavioral treatment that involves encountering triggers while learning not to carry out the associated compulsion.

ERP is a form of cognitive behavioral therapy with a specific focus on the relationship between triggers, distress, and compulsive responses. The goal is not merely to tell someone that a fear is unreasonable. It is to help them respond differently when that fear or urge arises.

Although medications are often described in terms of a particular neurotransmitter, the brain does not operate as a set of isolated chemical systems. Serotonin interacts with dopamine and noradrenergic signaling, among other systems. That complexity helps explain why predicting an individual’s response to medication can be difficult.

When standard treatments are not enough

Halpern estimated that roughly 30% of patients continue to struggle with OCD despite available treatments, with some experiencing severe symptoms. As a surgeon, he tends to encounter the most treatment-resistant cases: people who have already tried medications and specialized behavioral care without adequate relief.

For selected patients, surgical options include DBS and capsulotomy. These work differently:

• DBS modulates activity: Implanted electrodes deliver stimulation that can be adjusted or switched off.
• Capsulotomy creates a small lesion: A targeted area of tissue is destroyed to interrupt pathways implicated in symptoms.

Very small lesions, on the order of a few millimeters, can sometimes improve symptoms without obvious functional deficits on clinical testing. That does not make them trivial procedures or mean that the tissue has no function. Both stimulation and ablation require careful consideration of benefits, risks, and alternatives.

For the severe OCD cases under discussion, Halpern described a surgical responder rate of approximately 50%. Importantly, a response is not the same as a cure. Patients who improve may still have significant OCD symptoms. That limitation is a major reason to pursue more precise, symptom-specific targeting.

The Brain Circuits Linking OCD, Reward, and Compulsion

OCD is not a problem confined to one small brain structure. It involves networks connecting the cortex with deeper, subcortical regions.

Prefrontal and orbitofrontal areas can show abnormal, often heightened, activity in OCD. These regions communicate with the basal ganglia, including the caudate and putamen, which are part of the dorsal striatum. Those networks also interact with the ventral striatum, which includes the nucleus accumbens.

The nucleus accumbens is part of the brain’s reward circuitry. It helps regulate reward-seeking behavior and connects with many other regions. Under healthy conditions, having an urge to pursue a reward is entirely normal. Treatment is not intended to eliminate motivation or pleasure.

The problem arises when a drive repeatedly overrides the consequences of acting on it.

Continuing despite the cost

A useful thread connecting several conditions is pursuing an urge despite substantial risk or harm.

In an animal experiment, that might look like continuing to pursue a reward despite receiving a foot shock. In a person with OCD, it might mean checking locks repeatedly despite losing a night of sleep. In an eating disorder, it might involve bingeing or purging despite health risks. In addiction, it may involve seeking a drug despite dangerous consequences.

These disorders are not interchangeable. OCD, binge eating disorder, and substance addiction each have distinct features. But overlapping circuits may contribute to the difficulty of stopping a behavior once an urge has become powerful.

Repeated exposure to a particularly strong reward, such as a drug of abuse, can alter the normal operation of reward circuitry. Halpern’s research asks whether targeted interventions can interrupt a recurring, harmful pattern while preserving healthy reward-seeking.

Can Researchers Detect a Craving Before a Binge?

One of the most useful models for this work comes from Parkinson’s disease surgery.

During some DBS procedures, researchers record electrical activity through very fine electrodes. That activity can be converted into sound. In certain motor regions, the pattern of firing can correspond to the rhythm of a patient’s tremor.

This gives the surgical team a physiological signal related to the symptom they are trying to treat. Stimulation in an appropriate region, such as the subthalamic nucleus in relevant Parkinson’s cases, can disrupt the tremor circuit.

The corresponding question for binge eating is straightforward to ask but difficult to answer: What is the brain signal that precedes the urge to binge?

Why focus on craving?

Halpern’s group chose craving partly because it is a term patients recognize. People may have different interpretations of “bingeing” or “losing control,” and some may not describe an episode as out of control even when their behavior suggests otherwise. Craving provides a relatable starting point for identifying an internal state.

The aim is to find activity that reliably accompanies that state and could eventually guide stimulation. This is research, not a claim that DBS is a routine treatment for binge eating disorder or obesity.

The group also explored symptom-linked activity in an individual with OCD, identifying candidate signals related to obsessions while refining electrode placement. That was a single-case proof of concept, not evidence that a universal “obsession cell” or “craving cell” had been established.

Some early studies require patients to remain awake so researchers can relate their experiences to recorded activity. That does not mean every DBS procedure requires an awake patient. The need depends on the clinical procedure and the research question.

From individual neurons to a usable device signal

There is an important technical gap between detecting activity during surgery and detecting it through a chronically implanted device.

The fine recording electrodes used in the operating room can capture activity at the level of individual cells. Larger implanted contacts detect the combined activity of many thousands of cells. A signal useful for long-term treatment therefore has to be recognizable at that broader scale.

The research goal is to identify an electrical pattern an implanted device can detect and use to trigger stimulation at a relevant moment, rather than relying only on stimulation delivered without regard to the person’s current state.

Non-Invasive Brain Stimulation: TMS and Focused Ultrasound

Implanted electrodes provide valuable access to brain circuits, but surgery cannot be the answer for everyone. Non-invasive approaches are essential to making circuit-based treatments more broadly useful.

Two important approaches are transcranial magnetic stimulation and MRI-guided focused ultrasound. They differ substantially in how they affect the brain.

Transcranial magnetic stimulation

Transcranial magnetic stimulation, or TMS, uses magnetic stimulation delivered from outside the head to influence brain activity. Halpern noted FDA-authorized uses for depression, OCD, and smoking cessation.

One challenge is precision. Researchers need to establish which stimulation sites influence the relevant network and which patients are likely to benefit. Knowing that an intervention helps does not necessarily mean its mechanism is fully understood, a limitation that also applies to DBS.

TMS may also serve as a way to test circuit hypotheses. If stimulating a particular network temporarily reduces OCD symptoms, that response could help guide further research into more durable interventions.

For eating disorders and obesity, Halpern emphasized how much remains to be studied. The existence of TMS treatments for some conditions does not establish that the same approach will work for every disorder involving urges or reward.

MRI-guided focused ultrasound

Focused ultrasound can deliver energy to a selected brain region without an incision. MRI guidance helps clinicians locate and monitor the target.

In the tremor treatments Halpern described, focused ultrasound creates an ablation: a small, deliberately placed lesion. He uses it for selected patients with essential tremor or Parkinson’s-related tremor, often treating symptoms affecting one hand, such as the dominant or more severely affected hand.

The effects can be dramatic, but non-invasive does not mean the treatment leaves the brain unchanged. An ultrasound ablation destroys targeted tissue even though it does not require an implanted electrode or an open incision.

Researchers are also exploring other uses of ultrasound:

• Modulating neural activity without creating a permanent lesion.
• Increasing or inhibiting activity in specific regions.
• Temporarily opening the blood-brain barrier to help deliver a medication to a selected location.

These investigational uses should not be confused with established ultrasound ablation procedures for tremor.

A more precise tool still needs the right target

For psychiatric disorders, one of the largest obstacles is not simply delivering energy accurately. It is knowing where to deliver it.

Using ultrasound to reproduce an existing OCD capsulotomy target might avoid an incision, but it would not necessarily improve the clinical outcome. A new delivery method does not automatically solve the limitations of the target itself.

This is where stimulation and recording studies become especially valuable. Researchers can first learn which circuits contribute to symptoms, then determine whether those circuits can be treated safely through a non-invasive approach.

What Epilepsy Research Can Teach Us About Psychiatric Treatment

Epilepsy care offers a powerful model for mapping brain circuits in humans.

In stereoelectroencephalography, or stereo-EEG, clinicians place thin electrodes into selected brain regions to determine where seizures begin and how they spread. Patients are monitored in the hospital, and the resulting recordings can guide surgical removal, laser ablation, or implanted stimulation.

Although this remains an invasive brain procedure, Halpern emphasized that it can be well tolerated and has become an important part of specialized epilepsy care.

Researchers are adapting aspects of this strategy to study psychiatric symptoms. Teams working on depression have combined invasive recording methods with psychiatric expertise to investigate more individualized stimulation targets. Halpern described efforts to pursue related approaches for OCD.

Stimulation has a useful advantage during discovery: it can be adjusted or turned off. Electrodes can also be removed, although implantation and removal remain surgical procedures. A permanent lesion offers less flexibility if the target proves unsuitable.

Over time, sufficiently consistent findings could help define targets for non-invasive treatments. That is a potential path forward, not an established endpoint.

Why human studies are necessary

Animal models are valuable, but they do not capture every aspect of human psychiatric illness.

Halpern’s group developed a candidate target for binge eating and obesity from mouse research because certain aspects of feeding and reward-seeking can be modeled experimentally. Human obsessions and depressive experiences are more difficult to reproduce faithfully in a mouse.

Recordings from patients already undergoing epilepsy monitoring may offer opportunities to study co-occurring psychiatric symptoms. Ultimately, however, researchers need evidence from people with the condition they intend to treat.

Why Awareness Helps, but May Not Be Enough

Recognizing an approaching craving or a change in emotional state could create an opportunity to respond differently. That is one reason therapies that improve awareness and behavioral skills can be so valuable.

But there is a crucial distinction between recognizing an urge and being able to control what follows.

Halpern described patients with severe binge eating disorder who had already tried behavioral management, cognitive behavioral therapy, and medication. They understood their problem and could identify relevant triggers. Nevertheless, they continued to experience episodes they could not reliably stop.

Studying binges under controlled conditions

To investigate the brain signals associated with these episodes, the research team brings patients with implanted recording devices into a laboratory setting.

A psychiatrist or eating disorder specialist uses a structured mood-provocation procedure related to the patient’s own account of what precedes a binge. Researchers synchronize brain recordings with video and eye tracking to identify what the person is looking at, what they eat, and the timing of individual bites.

That timing matters because the desired signal is not simply activity that occurs during eating. It is activity that may help explain what happens immediately before the behavior begins.

Even in this highly structured setting, with participants fully aware that their eating behavior is being studied, binge episodes can still occur. For these severely affected patients, knowledge of the situation does not necessarily restore control.

Treatment resistance is not evidence that someone lacks insight or is unwilling to change.

Behavioral therapy remains valuable. Some people retain benefits long term, while others need ongoing support to sustain them. The point is not to set therapy against neuromodulation. It is to recognize that different levels of illness may require different combinations of care.

Could AI Predict Harmful Impulses Before We Notice Them?

Brain recordings are one possible source of early warning signals. Voice, breathing, sleep, and other physiological or behavioral measures may provide additional clues.

During our discussion, I raised research exploring whether patterns in speech and related signals could help identify an approaching depressive or suicidal crisis before a person recognizes a meaningful change. The underlying idea is that a machine-learning system might detect a recurring pattern across many measurements that would be difficult for a person to notice unaided.

Halpern saw promise in this direction, while emphasizing that the field is still early. His laboratory uses machine learning to study brain signals, but identifying a promising pattern is not the same as establishing a reliable clinical tool.

Compulsivity and impulsivity are related, but different

Much of the discussion centered on compulsivity: continuing to pursue an urge despite harmful consequences.

Impulsivity involves a related but distinct difficulty, such as acting before an appropriate cue or failing to wait. In a simplified animal task, a mouse may be trained to wait for a tone before seeking food. Seeking the food before the tone can provide a measure of impulsive responding.

Neither concept should be reduced to a character flaw. Both concern how behavior is regulated, and both can vary across people and circumstances.

The research question is whether changes in brain activity or bodily physiology can reliably indicate when control is becoming more difficult, and whether that information can guide a useful intervention.

Prediction must lead to meaningful help

The need for scalable treatments is enormous. Neurosurgery can help selected patients with severe illness, but surgical teams cannot treat everyone affected by addiction, depression, OCD, or disordered eating.

One of the broader contributions of invasive research may therefore be the knowledge it generates. Understanding symptom-linked brain activity could help researchers evaluate signals measured outside the brain and develop more accessible interventions.

However, an impressive-looking wearable or a sophisticated algorithm is not automatically a treatment. A useful tool needs evidence that it measures something clinically meaningful and helps improve outcomes. Otherwise, it risks consuming patients’ time and money without providing real benefit.

The Goal Is Better Control, Not the Elimination of Desire

Reward-seeking, motivation, and desire are normal parts of being human. The objective of this research is not to remove them. It is to help people regain the ability to choose how they respond when an urge becomes dangerous, disabling, or impossible to interrupt.

Deep brain stimulation demonstrates that changing activity in a small part of a circuit can sometimes produce a profound improvement in symptoms. It also reveals how much remains unknown, especially when moving from a visible motor symptom such as tremor to experiences such as obsession, craving, or depression.

The most promising path connects careful clinical care with precise recording, adjustable stimulation, and rigorous testing of non-invasive tools. Awareness can be powerful. For those whose symptoms persist despite awareness and treatment, restoring control may require intervening more directly in the circuits that make that control possible.