Sensory substitution devices (SSDs) involve translation of image to sound, with the aim to equip blind individuals with functions traditionally considered “visual”, like face detection. Previous work showed that image-to-sound SSDs recruit cortices specialised for visual functions. However, the brain dynamics of these activations, and thus, of SSD-supported perception, remain unknown. Visual cortices specialized for specific categories might be the first locus (or one of the loci) of perception of SSDmediated objects. Alternatively, the activation of those cortices may be a byproduct of perceptual processes unfurling first in auditory cortices. Resolving this uncertainty is critical for understanding to what extent the blind truly “see” via SSDs. Analyzing with electrical neuroimaging the EEG data from congenitally blind adults, we show for the first time that inferotemporal visual cortices are some of the earliest sites of face mediated by SSDs. These activations onset at ˜450ms, i.e., within the first 1/3 of the sweep through the 1,500ms face soundscapes. No earlier effects were detected, including with univariate methods. Interpreted together with subjective reports of blind SSD users, our results suggest to be the evidence that the blind can “see” with SSDs. Control analyses showing the specificity of these effects are presented. By identifying the temporal dynamics of SSD perception, our findings offer new support for the hypothesis that brain cortices are characterised by task-contingent functional organisation
