TALK DETAILS
Neurocomputational underpinnings of predictive perception
Luca Tarasi β Giuseppe di Pellegrino, Vincenzo Romei
28 September 2022
Background and Objectives. Predictive coding theory proposes that perceptual experience arises from the integrations between subjective (e.g., prior) and objective reality (e.g., visual input). The purpose of the present work was to map the behavioural and electrophysiological mechanisms behind the integration of prior knowledge into human perceptual decision-making.
Methods. We employed computational (signal detection theory, drift diffusion model) and electrophysiological (time-frequency decomposition) analyses to explore the impact that prior information has on perceptual decision. To this end, we manipulated prior knowledge by introducing uninformative or informative (low and high) target probability expectations in a detection task while simultaneously recording brain activity in 68 human participants using EEG.
Results. The behavioural results reveal that participants were able to incorporate prior information into their decision-making. Specifically, manipulation of perceptual expectations was able to elicit response bias (i.e., shaping criterion and starting point parameters) while leaving objective performance (i.e., d-prime and drift rate parameters) unchanged. At the neural level, the induction of response bias was reflected in the change in pre-stimulus alpha amplitude, with increased anticipation of target presence associated with increased alpha desynchronization in perceptual regions. Notably, the differentiation of alpha wave amplitude in the liberal vs. conservative condition was associated with the magnitude of change in criterion and starting point (but not d' and drift rate) at the individual level.
Discussion. We revealed that prior information shapes the content of perceptual representations rather than their fidelity and that this process is enacted through a preparatory and voluntary process mediated by alpha-band cortical oscillations in perceptual regions. These findings offer an advance in the understanding of the neural mechanism that drives perceptual decision-making and the integration of expectation-like information in humans, providing the oscillatory basis on which predictive brain theory is rooted.
doi.org/10.57736/nmc-ab93-1820π