We live in a world where some things are constantly changing while others stay the same. In order to navigate and act optimally in our environment, we therefore need to both recognize changes when they occur and learn about the regularities that are stable. Most of us are quite successful in doing so, but persons that are affected by autism or schizophrenia spectrum disorders, seem to perceive their environment as less stable than is warranted. This may explain why they experience the world differently, behave differently and (sometimes) make suboptimal decisions. We study this belief in volatility or changes in the environment with computerized tasks and eye-tracking. The latter is used because we can see surprise in the eyes, that is unexpected changes let the pupil dilate.
Based on previous studies our hypothesis was that patients with schizophrenia and persons with autism see the world as more volatile.
Our results show a more complex picture, where only some of these individuals see the world as more volatile, and a decreased adaptation of pupil size to uncertainty indicates problems with optimally adjusting to a changing world.
We identified factors causing the jumping to conclusion bias. When controlling for comprehension, motivation and by using computational modelling we find a role for memory for this bias, as well as perceiving the world as less stable contributes to the bias, particularly in psychosis.
We also found an overlap between autism and psychosis, in task performance as well as traits, questioning the opposite bias view. This should inform treatment; e.g., in high-functioning autism social deficits may arise from suspicion and paranoid ideation. In psychotic disorders, social withdrawal may prevent the calibration of beliefs and strengthen delusions. Viewing autism and psychosis as polar opposites is not supported by our research. We therefore recommend a precision medicine approach to psychiatry.
We also recommend controlling for comprehension, motivation and cognitive abilities as a mental disorder places a cognitive burden on the patient's thinking and reasoning.
More than 1% of the population are affected by schizophrenia and autism, posing tremendous challenges for the affected, their families and society. Schizophrenia and autism are detected only after onset of the pathology and preventive treatments are still in its infancy. Despite a plethora of behavioural and neuroscientific data, predictions linking behaviour and biology can currently not be made. A promising approach is the predictive coding framework, seeing the brain as an inference machine (Friston, 2005). Models based on this account, however, make opposite predictions for the same disorder or are similar for autism and psychosis.
My project aims at testing experimentally the predictions from the models. It will try to circumvent the limitations of the currently available models by using newly developed tasks that directly measure the parameters of the models. Three experimental tasks will be administered: measuring expectations in a probabilistic inference task, assessing the precision of one's memory, as well as a social inference task. This project will target both healthy people and patients. By measuring pupil dilation, a marker for neuronal gain, we can test whether aberrant processing is due to differences in gain (Friston 2005). By building computational models we test which of the predictive coding accounts of autism and psychosis explains the data across the three tasks best. Differences in performance and bias will identify subgroups, an important step for classifying cognitive endophenotypes. Second, we will scrutinize the obtained predictions by testing patients with psychosis before and after metacognitive therapy, translating the basic research into the clinical setting. Our interdisciplinary approach will test the contradictory models and develop a new model grounded in the experimental data. Identification of the computational parameters will contribute to a screening tool, detecting aberrant decision-making before it becomes pathological.