Perceptual decision-making under changing prior beliefs

• Applicant: Professor Dr. Andreas Heinz, since 5/2022
• Subject Area: Human Cognitive and Systems Neuroscience ; Clinical Psychiatry, Psychotherapy, Child and Adolescent Psychiatry
• Term: from 2018 to 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 407062764

The way we select information for perceptual decisions is shaped by expectations and beliefs regarding the world. Such selective sampling of sensory information has been conceptualized within the framework of hierarchical predictive coding as the result of an inference process that uses prior beliefs to infer states of the world. This process can be modelled as Bayesian inference, in which perceptual choices result from the combination of probabilistic prior beliefs and sensory data and whereby prior beliefs stored at higher hierarchical levels in the brain are fed back to lower levels. If there is a mismatch between the incoming sensory data and prior beliefs, a prediction error signal is fed forward to update the beliefs at higher levels. Prior beliefs have been proposed to modulate neural responses in sensory areas by sharpening representations of expected stimulus features, thereby selectively enhancing neural sampling of these features. A key question that has barely been investigated is how prior beliefs are updated in dynamically changing sensory environments and how this affects the selective sampling of sensory information. To deal with changes in the world, beliefs regarding the state of the world must be flexibly modulated by higher-level beliefs regarding environmental volatility. This is particularly important in light of the current models of schizophrenia: Symptoms of schizophrenia are thought to result from altered inference mechanisms that give rise to exaggerated prediction error signalling, rendering irrelevant sensory events overly salient and thus inducing delusional beliefs. These false beliefs are highly fixed, which may be the result of a bias towards believing that the world is less changeable. In this project, we aim to elucidate the neural mechanisms underlying the influence of prior beliefs on selective sampling of sensory information and the modulation of this influence by higher-level beliefs regarding the volatility of the environment. Moreover, we aim to extend the understanding of these processes by investigating selective sampling of sensory information in patients with schizophrenia as a model for dysfunctional decision-making. Using behavioural experiments, functional magnetic resonance imaging, and computational modeling in a hierarchical Bayesian predictive coding framework, we will investigate (1) how beliefs induced by audio-visual associative learning influence selective neural sampling under perceptual uncertainty in a random-dot motion-discrimination task; (2) how the influence of such beliefs on selective sampling is modulated by experimentally induced variations in environmental volatility; and (3) how the effect of prior beliefs and their modulation by environmental volatility on selective sampling relates to delusions and other psychotic symptoms in patients with schizophrenia.

• DFG Programme: Research Grants
• Ehemaliger Antragsteller: Professor Philipp Sterzer, Ph.D., until 4/2022