The auditory system relies on both local and global representations to recognize different sounds. Because local features analysis is constrained by storage capacity, average statistics of these features are computed over time to generate compact representations at the expense of temporal details availability. It is unknown whether these processes are restricted to the auditory system or interact with other modalities, such as vision. To answer this question, we examined how statistical properties of sounds are affected by atypical development of the visual system. We tested sighted and blind participants in two auditory experiments designed to target different strategies for analyzing sounds: one experiment relied on the availability of local features at specific time points; the other one benefitted from global statistical averaging over longer durations. Three groups of participants were recruited: sighted controls (SC), congenitally blind (CB), and late-onset blind (LB) individuals. SC performed as expected: when sound duration increased, local analysis gave way to global analysis. CB individuals showed almost identical patterns of behavior, revealing that auditory processing of local features, as well as statistical averaging of those features, are not altered by the complete absence of vision since birth. Surprisingly, we found that interactions between auditory computations and sight emerge later in development: LB individuals performed more poorly than those who were blind since birth. Crucially, performance was selectively degraded only for the use of local features, with no impact on statistical averaging. Taken together, these results suggested that early development is characterized by a segregation between basic auditory computations and sight, while at later developmental phases a functional interplay between the access to local auditory details and availability of visual input is involved. This interaction appears to be highly selective, as it does not affect other aspects of auditory processing, such as temporal averaging of sound statistics.
bioRxiv Subject Collection: Neuroscience