Reframing Binge Eating Disorder Through Circuit-Level Neuroscience

Binge eating disorder (BED) has largely been discussed in terms of heightened food reward, impaired self-control, or disrupted homeostatic regulation, leaving the cortical circuitry that links higher-order regulatory regions to hypothalamic feeding systems poorly defined. Fu Yu, PhD, and colleagues identified a projection from the medial prefrontal cortex (mPFC) to GABAergic neurons in the rostral zona incerta (rZI) that bidirectionally regulated punishment-resistant consumption of palatable food in mice, with repeated binge experience reorganizing this population into persistent food seeking.¹ Human imaging across 3 cohorts linked mPFC/anterior cingulate cortex-rZI connectivity to body weight and maladaptive eating measures spanning childhood through adulthood, providing correlational support for a translational role of this circuit. Yu noted these findings offer a mechanistic rationale for individualized, circuit-informed neuromodulation strategies that could refine existing dorsolateral prefrontal cortex stimulation approaches, which have reduced binge eating episodes in a randomized trial but rely on fixed anatomical targeting.²

Psychiatric Times: How does your paper fit into the larger context of binge eating disorder research?

Fu Yu, PhD: Binge eating disorder is often discussed in terms of heightened food reward, impaired self-control, or disrupted homeostatic regulation. An important unanswered question has been how higher-order cortical regions actually engage the brain's feeding and motivational systems to produce eating that persists despite negative consequences.

Our study identifies one such pathway: neurons in the medial prefrontal cortex, or mPFC, that project directly to GABAergic neurons in the rostral zona incerta, or rZI. In mice, manipulating this pathway in either direction changed punishment-resistant consumption of palatable food. Longitudinal imaging further suggested that repeated binge experience reorganized the activity of these projection-defined mPFC neurons, creating a persistent neural state biased toward palatable food seeking.

The broader contribution is therefore to frame compulsive eating as more than simply excessive hunger. It may involve experience-dependent changes in a cortical-hypothalamic circuit that gives palatable food unusually high priority, even when pursuing it carries a cost. We also found convergent associations between mPFC-rZI functional connectivity and body weight or maladaptive eating measures in three human cohorts. Those human findings are correlational, but they provide a translational bridge that warrants further study.

PT: What led you to research specific mPFC neurons?

Yu: The mPFC has a well-established role in behavioral control, value-based decision-making, and compulsive behaviors such as addiction, making it a strong candidate for providing top-down signals in compulsive eating. At the same time, human fMRI studies of obesity and binge eating disorder have produced inconsistent findings: depending on the cohort, task, and area examined, mPFC responses to food cues have been reported as either increased, decreased or no change. We felt that one reason may be that the mPFC is too large and functionally diverse to treat as a single brain region. Its different subregions and neuronal populations project to different targets and can support very different, sometimes opposing, functions.

We therefore took a circuit-specific approach. Rather than asking whether the mPFC as a whole was activated, we defined the relevant neurons by where they project. Our brain-wide tracing showed that the mPFC is a major source of direct input to rostral ZI GABAergic neurons, which were already of interest because the ZI integrates sensory, motivational, and affective information and can promote nonhomeostatic feeding. The major inputs arose from the prelimbic and medial orbital regions, and most of the rZI-projecting neurons expressed Npr3. They were also distinguishable from the better-known mPFC population projecting to the nucleus accumbens.

This allowed us to ask a more precise question: does the particular mPFC output channel communicating with the rZI regulate compulsive eating? Our causal manipulations and in vivo imaging supported that possibility. More broadly, resolving specific mPFC subregions and projection-defined populations may help explain why human imaging studies examining the mPFC more globally have produced variable results.

PT: What do you see as the most clinically relevant findings from this paper?

Yu: The most clinically relevant finding is the identification of a clear functional connection between the medial prefrontal cortex and the rostral zona incerta, and the demonstration in mice that specific mPFC inputs to the rZI causally regulate compulsive eating. Importantly, we found corresponding changes in mPFC/ACC–rZI connectivity across several human cohorts, from eating in the absence of hunger in young children to binge eating disorder in adolescents and maladaptive eating symptoms in adults with obesity. Although the human findings are correlational, this cross-species convergence suggests that the circuit may be relevant to human disease.

This discovery provides a potential mechanistic foundation for developing neuromodulation treatments for binge eating disorder. A recent double-blind randomized trial also found that repeated stimulation of the left dorsolateral prefrontal cortex reduced binge-eating episodes in women with binge eating disorder and obesity. However, the mechanism underlying this effect remains unclear, and stimulating the same anatomical DLPFC location in every patient may not optimally engage the disease-relevant circuit. Our findings suggest that the therapeutic effect could involve modulation of specific prefrontal–subcortical networks that sustain compulsive food seeking.

In other disorders like depression, personalized TMS targeting can begin by identifying the disease-relevant sgACC region and then selecting a DLPFC site that is strongly negatively connected with it; a conceptually similar strategy could be explored for binge eating disorder. Food-cue fMRI could first identify the ACC or sgACC region showing the strongest craving-related response in an individual patient. Researchers could also localize the patient’s rZI and identify the ACC region showing the strongest abnormal coupling with it. Finally, the accessible DLPFC site most negatively connected with this patient-specific ACC–rZI network could be selected for stimulation.

This approach could move the field from applying rTMS to the same broadly defined DLPFC location in every patient toward individualized, circuit-informed neuromodulation. It may be especially valuable for patients who do not respond adequately to existing behavioral or pharmacological treatments, potentially including GLP-1 receptor agonists. However, this remains a testable clinical hypothesis: our study does not yet establish that stimulating an anticorrelated DLPFC site will normalize ACC–rZI function or reduce binge eating in patients.

PT: Do you see any relevant connection of these findings to other compulsive issues or disorders?

Yu: Potentially, yes, but mainly at the level of a shared neural computation rather than a claim that the same circuit causes every compulsive disorder. A core feature of compulsivity is persistence despite adverse consequences. In our mice, repeated binge experience was associated with a persistent prefrontal population state that continued to favor palatable food even when its value should have been updated by bitterness or punishment.

Similar themes, such as altered valuation, inflexible updating, persistent motivational states, and prefrontal plasticity, are relevant to substance-use disorders and obsessive-compulsive and related disorders. The mPFC is involved broadly in action selection and behavioral control, while the ZI has roles in motivation, positive valence, and action initiation. It is therefore plausible that related cortical-subcortical architectures help stabilize other maladaptive behaviors.

That remains a hypothesis. Different forms of compulsivity have distinct symptoms and likely recruit partly distinct circuits. We did not test drug seeking, repetitive rituals, or other compulsive behaviors in this study. A useful next step will be to determine whether the mPFC-rZI pathway is specific to palatable-food seeking or participates in a more general mechanism of persistent, consequence-resistant motivation.

PT: How can practicing psychiatric clinicians employ your findings to educate their practice in treating eating disorders?

Yu: These findings will not change current clinical practice on their own, but they can support a useful and destigmatizing way of explaining binge eating disorder. Persistent binge eating is not simply a failure of willpower. Repeated binge-eating experiences may reshape brain systems involved in assigning value, guiding actions, and updating behavior when consequences become negative.

Clinically, the study also reinforces the importance of distinguishing obesity or generalized overeating from the defining features of binge eating disorder, including loss of control and the persistence of eating behavior despite distress or harmful consequences. Our mouse findings suggest that compulsive eating can be partly dissociated from ordinary hunger. This supports the need to assess the behavioral pattern and the patient’s experience of control, rather than infer the presence or absence of an eating disorder from body weight alone.

The work may also help clinicians explain why food-related cues, learned associations, and repeated exposure to highly palatable foods can acquire such powerful motivational influence, and why relapse does not mean that a patient lacks motivation or commitment to recovery. Existing evidence-based psychological and pharmacological treatments should remain the foundation of clinical care. Although neuromodulation is promising, circuit-informed stimulation based on our findings remains experimental and should not yet be used clinically outside properly designed research studies.

PT: How do you see brain imaging influencing eating disorder research in the future?

Yu: Brain imaging is increasingly moving beyond asking which individual brain regions are more or less active to examining how networks interact, and how those interactions change with experience, symptoms, and treatment. In eating disorders, this approach could help identify biologically meaningful subtypes that are not captured by body weight or diagnostic categories alone.

Longitudinal imaging will be particularly important. Rather than scanning patients only after an eating disorder is established, researchers could examine whether particular circuit features precede binge eating, emerge as the behavior becomes persistent, or normalize with successful treatment. Combining resting-state imaging with food-cue tasks, behavioral measures, and computational analyses may eventually help predict vulnerability, relapse, and treatment response. Imaging could also guide and monitor circuit-informed interventions. For example, it may allow researchers to identify an individual patient’s food-responsive ACC–rZI network and select a functionally connected cortical target for personalized noninvasive neuromodulation.

There are also important technical and interpretive limitations. The zona incerta is a small and anatomically complex structure, making high-resolution image acquisition, careful anatomical localization, standardized analyses, and independent replication essential. Higher-field MRI may be particularly valuable for studying small hypothalamic and subthalamic nuclei in humans. Functional connectivity, however, does not establish causal direction or reveal the underlying cell types. The greatest progress will therefore come from integrating human imaging with mechanistic animal studies, while avoiding the premature use of group-level associations as diagnostic or treatment-selection tools for individual patients.

Dr Yu is a senior principal investigator and director of the neurometabolism division at the Institute of Molecular and Cell Biology in Singapore.

References

1. Leow YN, Senol E, Qian X, et al. A cortical-hypothalamic neural circuit for compulsive eating in mice. Neuron. 2026.

2. Maranhão MF, Estella N, Cury MEG, et al. Repetitive transcranial magnetic stimulation of the left dorsolateral prefrontal cortex in binge eating disorder: a double-blind randomized controlled trial. Psychol Med. 2025;55:e149.