Elevated expectations of environmental volatility tracked greater delusion severity over 6 months among patients with schizophrenia-spectrum disorders recovering from acute psychotic episodes. These volatility priors may represent a state-sensitive marker of delusional thinking.
Investigators followed 75 patients with schizophrenia-spectrum disorders and 71 nonclinical comparison participants for 6 months. The participants were identified while receiving treatment for an acute psychotic episode and were eligible based on documented delusional thought content. They completed up to six assessments, including a three-option probabilistic reversal-learning task used to estimate their initial expectations about the stability of the task environment. Higher values reflected a greater expectation that environmental contingencies would shift erratically.
Delusion outcomes included clinician-rated severity among patients with schizophrenia-spectrum disorders and self-reported paranoia and delusional ideation in both groups. The investigators also assessed depression, worry, well-being, and amotivation to examine the specificity of associations with delusional thinking.
At baseline, patients with schizophrenia-spectrum disorders had higher volatility priors compared with nonclinical participants. Patients also had greater win-switch behavior, indicating more frequent switching to another card deck following a rewarded choice. Differences in sensitivity to volatility, the meta-volatility learning rate, and lose-stay behavior did not remain statistically significant after adjustment.
Paranoia and delusional ideation remained elevated compared with the nonclinical group at 6 months, but clinician-rated delusion severity, paranoia, and delusional ideation decreased among patients with schizophrenia-spectrum disorders. A reliable change analysis showed that the reduction in paranoia met the investigators’ threshold for reliable change, whereas reductions in delusional ideation and clinician-rated delusions may have been partly influenced by measurement invariance despite being statistically significant.
Volatility priors also decreased during the 6-month period among patients with schizophrenia-spectrum disorders but remained elevated relative to nonclinical participants at every assessment. The comparison group showed a sharp decline between the first and second visits but no statistically significant overall change across follow-up.
Longitudinally, higher volatility priors were associated with greater clinician-rated delusion severity among patients with schizophrenia-spectrum disorders. The association persisted among those who completed all study visits and after exclusion of the first assessment and examination of only data collected following initial exposure to the task. Volatility priors tracked paranoia and delusional ideation over time, with stronger associations among patients with schizophrenia-spectrum disorders.
Further evaluation showed that volatility priors were associated with persecutory but not grandiose delusions and with delusional conviction and preoccupation but not distress. There were no statistically significant longitudinal associations between volatility priors and depression, worry, well-being, or amotivation, including following adjustment for antipsychotic medication, cognitive ability, and premorbid IQ and when an alternative computational model was used.
The findings did not establish that volatility priors were exclusively state-dependent. Delusions and paranoia remained elevated at the end of follow-up, as did volatility priors. The investigators noted that it remained unclear whether volatility priors would eventually normalize or instead represent a hybrid state-trait marker reflecting both current delusion severity and proneness.
Study limitations included attrition and possible practice effects. Just 55% (n = 41) of the 75 patients with schizophrenia-spectrum disorders enrolled at baseline completed all assessments compared with 87% (n = 62) of nonclinical participants. Although analyses restricted to completers yielded a similar pattern, the full effect of attrition was unknown. Repeated exposure to the learning task remained a potential confound despite the use of different stimuli and analyses excluding the first assessment. The investigators noted that other computational models could have been used to characterize the learning data.
The findings were “an important step toward leveraging computational psychiatry to identify a modifiable latent cognitive mechanism” that could potentially be targeted in patients experiencing delusions or paranoia, wrote lead study author Julia M. Sheffield, of the Department of Psychiatry and Behavioral Sciences at the Vanderbilt University Medical Center, and colleagues.
Full disclosures of the study authors can be found in the study.
Source: Biological Psychiatry: Cognitive Neuroscience and Neuroimaging
