DEcoding Contextual Inference of Priors in Human pERception

Objective

The DECIPHER project aims to unveil the neural mechanisms through which the brain creates, updates, and integrates sensory expectations (priors), transforming uncertain signals into a stable and adaptive perception of the multisensory world. According to Bayesian and predictive coding models, perception is not a simple passive recording, but the result of a probabilistic inference that combines sensory inputs with prior knowledge to minimize uncertainty. Understanding how priors are formed and propagated in brain networks is crucial to explaining how the brain balances stability and sensitivity to change. The project has three main objectives. (1) To map where and how priors are represented in the brain, distinguishing global (stable) components from local (dynamic) ones, through psychophysical tasks and high-resolution neuroimaging (7T fMRI). (2) Investigating autism as a clinical model of alterations in predictive coding, verifying whether the reduced use of priors observed in people on the spectrum reflects an atypical organization of brain networks, particularly in interactions between sensory systems and associative hubs such as the default mode network. (3) Test the generalization of prior-based mechanisms in multisensory contexts, using the domain of numerosity, which lends itself to being estimated and integrated through vision, hearing, and touch. DECIPHER integrates behavioral approaches, advanced brain imaging, and network analysis to provide a multiscale description of how sensory expectations are constructed and used. The expected results will not only clarify the neurocognitive foundations of perception but also offer application perspectives for education and rehabilitation, contributing to innovative strategies for improving interaction with complex and uncertain environments.