[P1] Poster: Day 1

[P1-19] Characterising the spatial and temporal neural dynamics of temporal predictions in audition

*Clara Driaï-Allègre^1,2, Sophie Herbst¹ (1. Cognitive Neuroimaging Unit, INSERM, CEA, NeuroSpin (France), 2. Université Paris-Saclay (France))

Keywords:

temporal prediction,Bayesian observer,fMRI,MEG,temporal statistics

Predicting when future events will happen helps us focus and respond effectively, especially when our attentional capacity is limited. Here, I will present results from two studies (one finished, one with data acquisition in progress) that investigate the neural dynamics of temporal prediction for auditory perception using a statistical learning approach. To characterise how temporal regularities are internalised, we employ Bayesian observer models to capture the learning process over trials. In a recent EEG study (n=27), we were able to demonstrate that humans learn temporal statistics in a Bayesian manner. Specifically, target-evoked responses (P3) reflected Bayesian surprise as measured by Shannon's information. Furthermore, we will present preliminary results from an ongoing study using fMRI and MEG (acquisitions in progress). Participants perform a simple reaction time task in a foreperiod paradigm, in two separate sessions, one for fMRI and one for MEG (1-3 weeks apart), and we manipulate the mean and dispersion of the foreperiod distributions. Bayesian observer models will be fitted to reaction times to quantify participants’ temporal predictions per trial. By combining these two modalities, and informing the analyses with the information-theoretic parameters obtained from the Bayesian model (prediction error, surprise), we aim to uncover the spatial and temporal dynamics of the neural processes involved, particularly how learning to anticipate temporal probabilities enhances attentional focus over time, and how prediction error and surprise contribute to refining temporal predictions on subsequent trials. The combined fMRI-MEG approach allows us to consider cortical and subcortical brain areas, including the cerebellum, for which prior evidence suggests an implication in timing and predictive processing. By integrating neural and computational approaches, this work seeks to advance our understanding of how the brain encodes and utilises temporal statistical regularities.