A six-month study of people recovering from an acute psychotic episode has identified a potentially modifiable cognitive process linked to the severity of delusions. Researchers report that “volatility priors”—the brain’s expectations about how unpredictable or changeable the surrounding world is—were unusually elevated in adults with schizophrenia-spectrum disorders. As delusions and paranoia became less severe during recovery, these expectations also declined, suggesting that volatility priors may be a state-sensitive marker of psychosis rather than an entirely fixed vulnerability.
The findings, published in Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, offer new insight into how delusional beliefs may emerge and persist. Delusions are among the most recognizable symptoms of psychotic disorders: they are strongly held beliefs that remain resistant to contradictory evidence and can produce profound distress, social withdrawal, disability, and danger. Yet the cognitive processes that transform ordinary uncertainty into false certainty remain poorly understood. The new research points to the way people estimate environmental instability as one possible piece of that puzzle.
In computational psychiatry, a volatility prior describes a person’s expectation that the rules governing the world are likely to change. Someone with a relatively low volatility prior may assume that a pattern will remain stable unless there is strong evidence to the contrary. Someone with a high volatility prior, by contrast, may anticipate sudden shifts and treat the environment as unreliable, chaotic, or difficult to predict. Such expectations can influence how rapidly the brain updates its beliefs when new information arrives. When a person expects constant change, ambiguous events may be given disproportionate weight, potentially encouraging interpretations that become increasingly detached from external evidence.
Julia M. Sheffield, PhD, of Vanderbilt University Medical Center’s Department of Psychiatry and Behavioral Sciences, led the investigation to determine whether elevated volatility priors are a temporary feature of acute psychosis or a stable characteristic of people vulnerable to delusions. The distinction is clinically important. If volatility priors remain consistently high regardless of symptoms, they could represent a trait-like risk factor and a possible target for prevention. If they change alongside psychotic symptoms, they may instead function as a state marker—an indicator of the patient’s current clinical condition that could be influenced through treatment and recovery.
The research team recruited 75 adults with schizophrenia-spectrum disorders who had recently experienced an acute psychotic episode involving delusional thought content. Participants were assessed at six timepoints across six months, allowing investigators to track changes in cognition and symptoms rather than relying on a single snapshot. Their results were compared with data from 71 people without a diagnosed psychiatric disorder. To estimate volatility priors, participants completed a probabilistic reversal learning task, a behavioral experiment designed to measure how people respond when the relationship between choices and outcomes changes.
In this task, participants learn that one choice is more likely to produce a reward than another, but the probabilities periodically reverse. Successful performance requires tracking the environment, detecting when a pattern has changed, and adjusting behavior accordingly. The researchers used computational modeling to separate different components of decision-making, including how strongly participants expected the environment to shift. This approach does not directly measure a belief such as “the world is dangerous” or “people are plotting against me.” Instead, it estimates a more basic parameter governing belief updating: how much unpredictability a person expects before new evidence is presented.
At the beginning of the study, participants with schizophrenia-spectrum disorders showed significantly higher volatility priors than the non-clinical comparison group. Their delusional thinking, paranoia, and volatility expectations remained elevated at the end of the six-month follow-up and did not fully return to control levels. Nevertheless, all three measures generally declined over time, and the longitudinal analyses revealed that reductions in volatility priors were associated with improvements in delusion severity and paranoia. The relationship was specific: volatility priors were not significantly linked to changes in depression or worry, suggesting that the computational signal may be more closely related to psychotic belief formation than to general emotional distress.
The association remained after researchers accounted for antipsychotic medication and baseline cognitive ability. That finding does not show that changes in volatility priors caused delusions to improve, nor does it establish that modifying these expectations would necessarily eliminate psychotic symptoms. However, it strengthens the case that volatility priors track an important dimension of psychosis beyond medication exposure or overall intellectual performance. The researchers also found some evidence that the relationship was stronger for paranoid or persecutory delusions, in which people believe they are being watched, targeted, threatened, or harmed by others. Such beliefs may be particularly sensitive to expectations that social environments are unstable and unpredictable.
Repeated testing introduced an important complication. Participants completed the same type of task six times, and practice can alter performance independently of any genuine change in cognition. People may become faster, more accurate, or more familiar with the structure of an experiment simply because they have seen it before. The investigators attempted to reduce and account for these practice effects, but they acknowledge that the study remains partly confounded by repeated exposure. At the same time, the continued elevation of volatility priors after six months and six task sessions provided some reassurance that participants were not merely over-learning the experiment. The modeling results appeared to retain sensitivity to individual differences in volatility expectations beyond simple task familiarity.
The study’s implications extend beyond measurement. Antipsychotic medications, which broadly influence dopamine signaling, remain a central treatment for psychotic disorders, but responses vary substantially and many patients continue to experience delusions. Cognitive behavioral therapy for psychosis and other psychological interventions might eventually use computational information about belief updating to help patients examine how they interpret uncertainty and how rapidly they revise conclusions. A future treatment could, in principle, focus on recalibrating expectations about instability, helping individuals distinguish genuine changes in their environment from ordinary fluctuations or ambiguous events. Such an approach would need to be tested rigorously in controlled clinical trials before it could be considered an established therapy.
Sheffield and colleagues describe their results as foundational evidence that volatility priors are elevated during psychosis, decline during recovery, and move in parallel with the severity of specific delusional experiences. The work does not suggest that people with schizophrenia-spectrum disorders simply misread reality because they expect everything to change, nor does it reduce delusions to a single computational mechanism. Psychosis is biologically and psychologically diverse, and several interacting processes—including dopamine function, attention, learning, memory, social reasoning, and prior experiences—are likely to shape symptoms. Even so, identifying a measurable process that changes with recovery could help researchers connect laboratory models of belief updating with clinical treatment. The ultimate goal is to translate that connection into more precise and effective ways to reduce the disabling burden of delusions for patients and their families.
Subject of Research: People
Article Title: Longitudinal Associations Between Volatility Priors and Delusions in Individuals Recovering from an Acute Psychotic Episode
News Publication Date: August 18, 2026
Web References: https://doi.org/10.1016/j.bpsc.2026.06.013; Biological Psychiatry: Cognitive Neuroscience and Neuroimaging: https://www.sobp.org/bpcnni
References: Sheffield JM et al., “Longitudinal Associations Between Volatility Priors and Delusions in Individuals Recovering from an Acute Psychotic Episode,” Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, published July 3, 2026. DOI: 10.1016/j.bpsc.2026.06.013
Keywords: schizophrenia-spectrum disorders, delusions, psychosis, paranoia, volatility priors, computational psychiatry, belief updating, probabilistic reversal learning, cognitive markers, psychiatric treatment
