Article
Neural signatures of Bayesian perceptual adaptation during auditory motion discrimination (en)
* Presenting author
Abstract:
Auditory perception is subject to sensory noise and rapidly changing environments. To deal with ambiguous input, the auditory system needs to find the correct balance between flexibility and robustness. Bayesian inference determines the statistically optimal solution. We investigated on a behavioral and neurophysiological level whether auditory motion perception employs Bayesian inference. We had 26 young adults indicate the final direction of auditory motion sequences with random length and change points (CPs), while monitoring their neural activity via EEG. Participantsâ accuracy changed significantly based on the occurrence of CPs with a sharp decrease directly following a CP and a steady increase with additional motions in the same direction, revealing a strong bias towards momentarily established priors directly following a CP. Cluster-based permutation analysis of the EEG data revealed a centrally distributed P3b component showing later and longer sustained activity for motions directly following a CP. Momentary estimates of perceptual surprise were estimated by a Bayesian CP model and significantly predicted the cluster amplitudes on a single-sound level. These findings suggest auditory motion perception to continuously adapt to unpredictable changes as the Bayesian observer would do.