A new framework proposed by researchers E.M. van Fenema and G.E. Jacobs suggests that a wide range of psychiatric symptoms may be understood through a common failure in the brain’s prediction machinery. Rather than treating disorders as entirely separate categories defined only by symptoms, the authors argue that psychiatry could benefit from examining how the brain generates, updates and uses predictions about the body, the environment and other people. Their perspective, published in Translational Psychiatry, places “predictive dysfunction” at the center of a possible mechanism-based model for mental illness.
The proposal builds on the predictive-processing theory of brain function. In this view, the brain is not a passive receiver of sensory information. It continuously generates hypotheses about what is happening and compares those expectations with incoming signals. The difference between a prediction and the information actually received is known as a prediction error. The brain then adjusts its internal model, changes its expectations or alters its behavior to reduce uncertainty. This process operates across multiple levels, from basic sensory perception to complex beliefs about identity, social relationships and the future.
According to the framework, psychiatric symptoms may emerge when one or more elements of this predictive system become improperly calibrated. The problem could involve the content of predictions, the strength assigned to prior beliefs, the reliability given to sensory evidence or the way prediction errors are transmitted through neural networks. In technical terms, the brain may assign abnormal “precision” to either expectations or incoming data. A prediction that is normally treated as flexible could become excessively dominant, while a genuinely important signal might be discounted. These imbalances could help explain why experiences that are internally generated sometimes feel more compelling than external evidence.
This approach may offer a way to connect symptoms that are traditionally divided among diagnostic categories. Hallucinations, delusions, anxiety, depression, compulsive behavior and disturbances in the sense of self appear clinically different, but each can involve altered expectations about what is happening or what is likely to happen next. A person with severe anxiety may predict danger too readily and interpret ambiguous signals as threatening. Someone experiencing depression may develop highly persistent predictions of failure, loss or low personal value. In psychosis, internally generated interpretations may acquire excessive confidence, making contradictory evidence difficult to incorporate.
The researchers’ central argument is not that every psychiatric condition has one single cause. Instead, predictive dysfunction is presented as a possible organizing mechanism that can operate differently across individuals and disorders. Biological vulnerability, developmental experience, stress, inflammation, trauma, sleep disruption and social context could all influence how predictive models are formed and maintained. The same underlying computational disturbance might therefore produce different symptoms depending on the brain systems involved and the person’s history. This could help explain why people with the same diagnosis often show markedly different symptoms and why similar symptoms can appear in different disorders.
A mechanism-based framework could also change how researchers design studies and treatments. Current psychiatric diagnoses are largely descriptive: they classify patterns of behavior and reported experience rather than identifying a single biological process. Predictive-processing models could encourage investigators to measure specific computational functions, such as how strongly a person relies on prior expectations, how quickly beliefs are updated and how uncertainty is represented. These functions could be examined using behavioral experiments, computational modeling, neuroimaging, physiological measurements and longitudinal clinical assessments.
The framework may have implications for treatment, although it does not itself establish a new therapy. Medication, psychotherapy and environmental interventions could be interpreted as influencing different parts of the predictive system. Pharmacological treatments might alter the neural signaling involved in precision and learning, while psychotherapy could help patients test rigid predictions and develop more adaptive models of themselves and the world. Changes in sleep, social safety and daily routines might also modify the signals used by the brain to update its expectations. The long-term goal would be to match interventions to the mechanism maintaining a person’s symptoms rather than relying only on diagnostic labels.
However, predictive-processing theories also face important challenges. Predictions are difficult to measure directly in living humans, and computational models can produce different interpretations from the same behavioral data. A mathematical description of abnormal belief updating does not automatically identify its neural cause, nor does it prove that predictive dysfunction is the primary driver of a disorder. Psychiatric experiences are shaped by meaning, culture, relationships and personal history, factors that cannot be reduced entirely to a single computational parameter. The framework will therefore need to be tested against clinical data and compared with other biological, psychological and social explanations.
Van Fenema and Jacobs’ proposal arrives as psychiatry increasingly seeks explanations that move beyond symptom checklists without abandoning the complexity of human experience. By focusing on how the brain anticipates the world and responds when expectations fail, the predictive-dysfunction model offers a common language for studying perception, emotion, cognition and behavior. Its influence will depend on whether future research can translate the theory into reliable measurements and clinically useful predictions. If that happens, psychiatric diagnosis could gradually shift from asking which label best describes a patient to asking which predictive processes have become disrupted—and how those processes can be restored.
Subject of Research: Predictive dysfunction as a unifying, mechanism-based framework for understanding psychiatric disorders.
Article Title: Predictive dysfunction: toward a unifying mechanism-based framework for psychiatry
Article References: van Fenema, E.M., Jacobs, G.E. “Predictive dysfunction: toward a unifying mechanism-based framework for psychiatry.” Translational Psychiatry (2026). https://doi.org/10.1038/s41398-026-04356-0
Image Credits: AI Generated
DOI: https://doi.org/10.1038/s41398-026-04356-0
Keywords: predictive processing, predictive dysfunction, psychiatry, mental disorders, prediction error, computational psychiatry, brain mechanisms, psychiatric diagnosis, neurobiology, mental health
