Purchase this article with an account.
Bruce C Hansen, Benjamin Thompson, Robert F Hess, Dave Ellemberg; Reverse correlation between the N170 and fractal noise yields human faces: A time-frequency spectrum analysis. Journal of Vision 2009;9(8):465. doi: https://doi.org/10.1167/9.8.465.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Reverse correlation based on behavioral responses to white noise stimuli can render in external space the internal representation of faces held by human observers. Interestingly, such external renditions of the internal face representation can be obtained in tasks where no face signal is ever presented (Gosselin & Schyns, Psych Sci, 2003). Here, we sought to bypass behavioral measures and derive whole face representations directly from neural activity in the human brain. We presented fractal noise stimuli (i.e., noise patterns with amplitude spectra slopes similar to human faces) in combination with a behavioral response continuum while simultaneously recording EEGs. Fractal noise stimuli were presented for 750ms (preceded by a 250ms base-line mean luminance blank) followed by a 1500ms response interval. EEGs over the P7, P8, P9, P10, P07, and P08 electrodes were subjected to a time-frequency analysis near 170ms (i.e., negative amplitudes at 170ms at these electrodes have been linked to early face processing) on a trial-by-trial basis regardless of the behavioral response. Fractal noise patterns eliciting significant negative EEG amplitudes as well as those that elicited weak or non-negative amplitudes were differenced to generate separate sets of face classification images for delta, theta, alpha, beta, and gamma frequency (Hz) ranges as well as across all frequencies. EEG derived face classification images based on the entire time-frequency spectrum at and near 170ms produced the strongest external renditions of faces, suggesting that it is the entire range of EEG frequencies which contribute to the N170 representation of human faces. Convincing EEG derived face classification images were also generated from the theta-alpha frequency range near 220ms (negative amplitude), theta frequency range near 380ms (negative amplitude), and theta-beta frequency range near 500ms (positive amplitude). In none of the observed cases did the EEG derived face classification images match those derived from the participants' behavioral responses.
This PDF is available to Subscribers Only