Anhedonia—the diminished ability to experience pleasure—has long challenged clinicians because it cuts across diagnostic boundaries. It can appear in depression, post-traumatic stress disorder, schizophrenia-spectrum conditions, and other forms of mental illness, yet standard treatments do not always restore a person’s capacity to anticipate, pursue, or enjoy rewarding experiences. A new proof-of-concept trial described in Translational Psychiatry is testing a novel way to address this problem: using neurocomputational training to strengthen the brain processes involved in learning from rewards and detecting their value.
The study, led by K.M. Harlé, M.B. Stein, A.N. Simmons, and colleagues, focuses on a question that is increasingly central to modern psychiatry: can measurable features of learning and motivation be directly trained, rather than treated only through broad diagnostic categories? Instead of asking whether a participant meets criteria for depression or another disorder, the researchers target anhedonia itself as a shared psychological and neural process. This transdiagnostic approach could help reshape treatment development around the mechanisms that symptoms have in common.
Reward is not a single experience in the brain. It includes anticipation, motivation, learning, decision-making, and the pleasurable response that follows a rewarding event. A person with anhedonia may struggle to predict that an activity will feel good, may fail to update expectations after a positive experience, or may find the experience itself less satisfying. These components can overlap, but they are not identical. Neurocomputational methods provide a framework for separating them and describing how the brain updates its expectations when outcomes are better or worse than predicted.
At the center of this framework is reinforcement learning, a computational theory of how organisms adapt through experience. When an outcome differs from what was expected, the discrepancy can be represented as a prediction error. A reward that is larger than anticipated generates a positive prediction error, encouraging the brain to repeat the associated behavior. A disappointing outcome generates a negative prediction error, prompting adjustment. In anhedonia, some of these learning signals may be weakened, poorly integrated, or less influential in guiding future choices. The trial is designed to explore whether targeted training can improve this system.
The phrase “neurocomputational training” refers to an intervention informed by models of brain function rather than relying solely on conventional psychological exercises. Such training may use behavioral tasks, feedback, or repeated decision-making experiences designed to engage specific computational processes. The goal is not simply to make participants encounter pleasant events, but to strengthen the mechanisms that allow the brain to recognize, learn from, and respond to positive outcomes. By focusing on reward sensitivity and reward learning, the researchers are attempting to move from symptom description to process-level intervention.
The trial’s proof-of-concept status is important. A proof-of-concept study is generally intended to determine whether an approach is feasible, theoretically justified, and capable of producing signals that warrant larger investigations. It is not, by itself, a definitive demonstration that a treatment works for everyone or that it should replace established clinical care. The publication presents the work as an early test of whether a computationally guided strategy can be applied to anhedonic individuals across diagnostic categories. That framing makes the study scientifically significant while also placing appropriate limits on what can be concluded.
A transdiagnostic design may be especially valuable because anhedonia often behaves differently from other symptoms within the same diagnosis. Two people with depression, for example, may share a diagnostic label while having very different difficulties: one may be dominated by anxiety and rumination, while another may primarily lose motivation and pleasure. Conversely, people with different diagnoses may exhibit similar impairments in reward processing. If these mechanisms can be measured reliably, future treatments could be assigned according to a patient’s computational profile rather than diagnosis alone.
The research also reflects a broader shift toward precision psychiatry. Traditional psychiatric diagnoses are primarily based on clusters of reported symptoms, but those categories do not always reveal the biological or cognitive processes driving an individual’s difficulties. Computational psychiatry seeks to bridge this gap by translating behavior into mathematical parameters, such as how strongly a person learns from positive feedback, how much value they assign to anticipated rewards, or how consistently they update beliefs. These parameters may eventually help researchers identify who is most likely to benefit from a particular intervention.
Interest in reward circuitry has grown because anhedonia is associated with disruptions across interconnected brain systems involved in valuation, motivation, and learning. Regions such as the ventral striatum and prefrontal cortex are frequently studied in relation to reward processing, while neuromodulators including dopamine help regulate the way motivational information is represented. However, anhedonia is not reducible to a single brain region or chemical. It emerges from interactions among neural circuits, cognition, behavior, and environment. A computational training program therefore offers a way to examine how these systems operate together, even when the underlying biology is complex.
If future research confirms that reward learning and sensitivity can be improved through targeted training, the implications could extend beyond one intervention. Clinicians might eventually use brief behavioral assessments to identify specific reward-processing deficits, deliver individualized exercises, and track change through computational measures alongside symptom questionnaires. Such a model could complement psychotherapy and medication rather than compete with them. For now, the Harlé and colleagues study represents an early but potentially influential step toward treating anhedonia as a modifiable mechanism—one that may link multiple psychiatric conditions and provide a new route to recovery.
Subject of Research: Neurocomputational training to improve reward learning and reward sensitivity in individuals experiencing anhedonia across psychiatric diagnoses.
Article Title: Neurocomputational training to boost reward learning and sensitivity in anhedonic individuals: a transdiagnostic proof-of-concept trial.
Article References: Harlé, K.M., Stein, M.B., Simmons, A.N. et al. “Neurocomputational training to boost reward learning and sensitivity in anhedonic individuals: a transdiagnostic proof-of-concept trial.” Translational Psychiatry (2026). https://doi.org/10.1038/s41398-026-04355-1
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s41398-026-04355-1
Keywords: anhedonia, reward learning, reward sensitivity, neurocomputational training, computational psychiatry, transdiagnostic research, precision psychiatry, mental health
