Purchase this article with an account.
Dan Nemrodov, Ilya Nudnou, Adrian Nestor; Imagery and perception-based decoding of facial identity from EEG signals. Journal of Vision 2020;20(11):311. doi: https://doi.org/10.1167/jov.20.11.311.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Despite recent progress in unravelling the neurocomputational mechanisms for face perception,
their counterparts, that subserve visual memory and imagery for faces, are yet to be elucidated.
To address this challenge, here, we used electroencephalography (EEG) in healthy adults to
assess the neural response elicited by familiar faces, as participants viewed or recalled their
appearance in response to a cue. Specifically, we appealed to pattern analysis to decode the
identity of famous female faces and to characterize the structure of face space from imagery
and perception. The outcome of this investigation showed, first, that facial identity could be
decoded from both perception and imagery. Second, the temporal profile of face decoding
showed, for imagery, particular reliance on the interval between 1-2 seconds after cue onset
while, as expected, this profile exhibited an earlier peak for perception, around 400 ms after
stimulus onset. Third, despite their divergence in the temporal domain, imagery and perception
were spatially similar in their reliance on parietal and centro-parietal channels for decoding
purposes. Last, an assessment of representational spaces showed that imagery-based space
structure could be explained to some extent, but not entirely, by its perceptual counterpart.
Thus, the present findings shed light on the visual representations underlying face imagery, on
their spatiotemporal dynamics and on their relation with perception. Moreover, methodologically,
they demonstrate the ability of EEG signals to carry fine-grained information regarding visual
memory and imagery for faces.
This PDF is available to Subscribers Only