Abstract
The discovery that deep convolutional neural networks (DCNNs) achieve human performance in realistic tasks offers fresh opportunities for linking neuronal tuning properties to such tasks. Here we show that the face-space geometry, revealed through pair-wise activation similarities of face-selective neuronal groups recorded intracranially in 33 patients, significantly matches that of a DCNN having human-level face recognition capabilities. This convergent evolution of pattern similarities across biological and artificial networks highlights the significance of face-space geometry in face perception. Furthermore, the nature of the neuronal to DCNN match suggests a role of human face areas in pictorial aspects of face perception. First, the match was confined to intermediate DCNN layers. Second, presenting identity-preserving image manipulations to the DCNN abolished its correlation to neuronal responses. Finally, DCNN units matching human neuronal group tuning displayed view-point selective receptive fields. Our results demonstrate the importance of face-space geometry in the pictorial aspects of human face perception.
Highlights
The discovery that deep convolutional neural networks (DCNNs) achieve human performance in realistic tasks offers fresh opportunities for linking neuronal tuning properties to such tasks
What could be the function of the observed face-exemplar selectivity? To find out whether a DCNN with human-level face recognition performance (VGG-Face) could serve as a realistic functional model of these selectivities, we examined whether the face-space geometry of face exemplars, as determined by pair-wise distances between their activation patterns[25,26], was similar between the human cortex and individual DCNN layers
Given the ongoing debate concerning the functional distinction between the two face patches, we separately examined their match to the DCNN layers
Summary
The discovery that deep convolutional neural networks (DCNNs) achieve human performance in realistic tasks offers fresh opportunities for linking neuronal tuning properties to such tasks. We show that the face-space geometry, revealed through pair-wise activation similarities of face-selective neuronal groups recorded intracranially in 33 patients, significantly matches that of a DCNN having human-level face recognition capabilities. This convergent evolution of pattern similarities across biological and artificial networks highlights the significance of face-space geometry in face perception. Systems neuroscience research has been one of the fastest growing fields of science in recent years, accumulating detailed depictions of neuronal functional properties. Despite this progress, two fundamental questions remain unsolved. The problem is largely due to the lack of models whose functional performance can achieve realistic human or animal levels[15]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
