Assessing the predictive coding accounts of Autism Spectrum Disorders

The MEMO Team cordially invites you to a talk

by

Laurie-Anne Sapey-Triomphe (PhD)

1. Department of Brain and Cognition, Leuven Brain Institute, KU Leuven, 3000 Leuven, Belgium.

2. Leuven Autism Research (LAuRes), KU Leuven, 3000 Leuven, Belgium.

Title: Assessing the predictive coding accounts of Autism Spectrum Disorders

Abstract

Autism Spectrum Disorder (ASD) is characterized by heterogeneous social and non-social symptoms. Recent predictive coding theories have attempted to account for this heterogeneity in an integrated theory by suggesting that atypical perceptual learning could play a central role in ASD. Specifically, priors, which capture the underlying statistical regularities of the environment, may be atypically learnt in ASD. Other hypotheses suggest that sensory sensitivity may be too high in ASD, leading to an atypical sensory/prior ratio. We conducted a series of behavioral and neuroimaging experiments aiming at characterizing how autistic individuals learn and adjust their priors. Across these studies, we showed that autistic adults can implicitly learn a prior and have their percepts biased by their expectations. Yet, they tend to have more difficulties in adjusting their priors to the context than neurotypicals. Furthermore, we identified some neural and molecular correlates of prediction learning in ASD using model-based fMRI and Magnetic Resonance Spectroscopy. The neural network encoding prior knowledge and prediction errors was generally similar in adults with and without ASD, but the ASD group activated more strongly some regions involved in encoding high-level priors and high-level prediction errors, while no differences were found at the lower level of the hierarchy. Furthermore, we found that the glutamate/GABA ratio in a region involved in prediction learning (i.e., inferior frontal gyrus) was correlated with the ability to learn predictions. Finally, using EEG associated with fast periodic visual stimulation, behavioral tasks and questionnaires, we characterized sensory sensitivity at multiple levels to test the hypothesis of an increased sensory precision in ASD. Altogether, these results help refining the predictive coding theories of ASD and shed light on the potential underlying neural mechanisms.