The role of conditioning in perceptual inference

• Applicant: Professor Philipp Sterzer, Ph.D.
• Subject Area: Biological Psychiatry ; Human Cognitive and Systems Neuroscience
• Term: from 2014 to 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 250227358

The sensory data available to our brains are a fundamentally impoverished source of information. However, we nonetheless usually experience our perception of the world as rich, unitary and coherent. To generate such coherent perception that can be used as a reliable guide to behaviour, our brains need to infer the causes underlying the sensory data. Perception is therefore commonly conceptualized as an active inferential process that uses endogenous predictions to 'make sense' of the available sensory data. Perceptual inference is adaptive in that these predictions are constantly updated by new sensory evidence. It seems obvious that such an adaptive process should be shaped by its consequences, that is, whether a percept is associated with favourable or aversive events. Surprisingly, however, little is known about the role of such conditioning in perceptual inference. The key assumption of the current proposal is that perceptual inference is an active process that is modulated by conditioning in a similar way as behaviour. In addition to such operant-like conditioning, which is here referred to as percept conditioning, it is also assumed that a form of classical conditioning (here: stimulus conditioning) may also play a role in perceptual inference. This project aims to elucidate the roles of percept and stimulus conditioning in perceptual inference using the paradigm of perceptual ambiguity in a series of behavioural and functional magnetic resonance imaging (fMRI) experiments. The mechanisms of percept vs. stimulus conditioning will be disentangled by either associating perceptual choices with monetary wins or losses, just as in operant conditioning; or by associating visual stimuli that compete for awareness with monetary wins or losses, as in classical conditioning. Crucially, subjective perceptual alternations that occur during the viewing of ambiguous stimuli will be assessed with objective measures in order to preclude potential confounding effects of subjective reports. Behaviourally, perception will be tracked indirectly with a probe detection task. To provide an objective measure of perception at the neural level, multivariate pattern analysis of fMRI data will be used to decode perception from percept-specific brain signals. Moreover, the neural basis of conditioning in perceptual inference will be investigated with a univariate statistical approach in conjunction with models of effective connectivity. It is expected that this research will provide evidence for dissociable effects of percept and stimulus conditioning both at the behavioural and the neural level. The proposed project will thus further our understanding of the mechanisms underlying conscious perception by shedding light on a hitherto largely unexplored aspect of perceptual inference. It will also pave the way for future research into the mechanisms underlying altered perceptual experience in psychiatric diseases such as affective disorders and schizophrenia.

• DFG Programme: Research Grants