Elsevier

Cognition

Volume 208, March 2021, 104424
Cognition

Brief article
From concepts to percepts in human and machine face recognition: A reply to Blauch, Behrmann & Plaut

https://doi.org/10.1016/j.cognition.2020.104424Get rights and content

Abstract

Intact recognition of familiar faces is critical for appropriate social interactions. Thus, the human face processing system should be optimized for familiar face recognition. Blauch et al. (2020) used face recognition deep convolutional neural networks (DCNNs) that are trained to maximize recognition of the trained (familiar) identities, to model human unfamiliar and familiar face recognition. In line with this model, we discuss behavioral, neuroimaging and computational findings that indicate that human face recognition develops from the generation of identity-specific concepts of familiar faces that are learned in a supervised manner, to the generation of view-invariant identity-general perceptual representations. Face-trained DCNNs seem to share some fundamental similarities with this framework.

Section snippets

CRediT authorship contribution statement

Galit Yovel:Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing - original draft, Writing - review & editing.Naphtali Abudarham:Conceptualization, Investigation, Methodology, Software, Writing - original draft, Writing - review & editing.

Acknowledgment

This work is supported by Joint China-Israel, NSFC-ISF Research Grant Application no. 2383/18.

References (26)

  • N. Abudarham et al.

    Reverse engineering the face space: Discovering the critical features for face identification

    Journal of Vision

    (2016)
  • N. Abudarham et al.

    Same critical features are used for identification of familiarized and unfamiliar faces

    Vision Research

    (2018)
  • N. Abudarham et al.

    Face recognition depends on specialized mechanisms tuned to view-invariant facial features: Insights from deep neural networks optimized for face or object recognition

    (2020)
  • N. Abudarham et al.

    Critical features for face recognition

    Cognition

    (2019)
  • N. Blauch et al.

    Computational insights into human perceptual expertise for familiar and unfamiliar face recognition

    Cognition

    (2020)
  • J. Cavazos et al.

    Accuracy comparison across face recognition algorithms: Where are we on measuring race bias? ArXiv

    (2020)
  • F. Fang et al.

    Duration-dependent fMRI adaptation and distributed viewer-centered face representation in human visual cortex

    Cerebral Cortex

    (2007)
  • M.I. Gobbini et al.

    Neural systems for recognition of familiar faces

    Neuropsychologia

    (2007)
  • M. Gotlieb et al.

    Conceptual rather than perceptual similarity enables generalization across perceptually different appearances of familiarized faces

    (2020)
  • Jayaraman, S., Fausey, C. M., & Smith, L. B. (2015). The faces in infant-perspective scenes change over the first year...
  • R.S.S. Kramer et al.

    Natural variability is essential to learning new faces

    Visual Cognition

    (2017)
  • R.S.S. Kramer et al.

    Understanding face familiarity

    Cognition

    (2018)
  • S.M. Landi et al.

    Two areas for familiar face recognition in the primate brain

    Science

    (2017)
  • Cited by (8)

    • Perceptual similarity modulates effects of learning from variability on face recognition

      2022, Vision Research
      Citation Excerpt :

      When people do radically change across encounters, either due to a large gap between encounters (e.g., a class reunion) or radical changes in appearance (e.g., removal of facial hair), supervision is typically available as part of a normal social interaction among people, either explicitly by indicating the identity of the person or by using other identity cues including voice or the content of a conversion. Thus, familiar face recognition is enabled by generalization to images that are perceptually similar to previously learned appearances together with supervision to perceptually different appearances (Yovel & Abudarham, 2021). The role of exposure to different appearances in face identification also raises the question of the definition of familiarity.

    • Deep learning of shared perceptual representations for familiar and unfamiliar faces: Reply to commentaries

      2021, Cognition
      Citation Excerpt :

      A major point of agreement between us and both Young and Burton (2020), and Yovel and Abudarham (2020), hereafter Y&B and Y&A, respectively, is the fact that faces exhibit a large degree of idiosyncratic within-identity variability, which places fundamental constraints on face recognition performance (Young & Burton, 2018; Kramer, Young & Burton, 2018).

    View all citing articles on Scopus
    View full text