Multi-level reorganization in the temporal dynamics of sound processing in early blind people

Early blindness triggers reorganization in the brain networks that code for sound processing. However, how visual deprivation impacts the temporal dynamics of different stages of auditory discrimination (acoustic to categorical coding) remains mostly unexplored. We characterized the time course of brain activity using electroencephalography (EEG) while congenitally blind (CB) and sighted individuals (SC) listened to 1-second-long sounds belonging to eight categories. Multivariate decoding analyses revealed enhanced sound decoding in the brain of the CB compared to sighted from 167 to 1007 ms after sound onset. Furthermore, the classifier weights transformed and projected on the sensors were enhanced in the CB with the topography evolving along a frontal-posterior axis as the sound unfolded in time. To investigate which format of sound processing were enhanced in CB, we used representational similarity analysis (RSA) with three classes of models for sound representation: (i) the Modulation Transfer Function (MTF) to simulate early stage of acoustic processing, (ii) layers of a deep neural network (YAMNET), (iii) a high-level model based on the categorical membership of sounds and participant specific similarity ratings of each pair of sounds. MTF emerged early in the EEG of each group, peaking at around 200 ms however, no differences were found between the two populations. In contrast, the correlations with the DNN layers displayed distinctions at around 200 ms and post stimulus offset specifically in few a layers representing intermediate acoustic processing. Categorical representation of sounds emerged at around 250 ms in both populations, CB showed an enhanced categorical representation peaking at 550 ms. Furthermore, regression analysis indicated that the DNN layers could uniquely explain the maximum variance in the EEG of both populations early in time while unique contribution of categorical models peaked at 550 ms. The DNN layers representations in the brain were found to be mediated by enhanced theta oscillations in CB. All these results suggest that early blindness triggers a multi-level reorganization in brain networks coding for sounds, with enhanced intermediate-level acoustic discrimination earlier in time, followed by an increased categorical coding later.

Affiliations

UCLouvainSSH/IPSY - Psychological Sciences Research Institute

Citations

APA
Chicago
FWB

Talwar, S., Mattioni, S., Giraudet, E., Calce, R. P., Barbero, F., Barilari, M., & Collignon, O. (2024). Multi-level reorganization in the temporal dynamics of sound processing in early blind people. Cognitive Neuroscience Society, Toronto. https://hdl.handle.net/2078.5/213860