Fast saccades towards faces are robust to orientation inversion and contrast negation.
Contrast negation
Face detection
Face inversion effect
Rapid visual categorisation
Saccade
Journal
Vision research
ISSN: 1878-5646
Titre abrégé: Vision Res
Pays: England
ID NLM: 0417402
Informations de publication
Date de publication:
08 2021
08 2021
Historique:
received:
02
06
2020
revised:
22
03
2021
accepted:
30
03
2021
pubmed:
19
4
2021
medline:
27
1
2022
entrez:
18
4
2021
Statut:
ppublish
Résumé
Eye movement studies show that humans can make very fast saccades towards faces in natural scenes, but the visual mechanisms behind this process remain unclear. Here we investigate whether fast saccades towards faces rely on mechanisms that are sensitive to the orientation or contrast of the face image. We present participants pairs of images each containing a face and a car in the left and right visual field or the reverse, and we ask them to saccade to faces or cars as targets in different blocks. We assign participants to one of three image conditions: normal images, orientation-inverted images, or contrast-negated images. We report three main results that hold regardless of image conditions. First, reliable saccades towards faces are fast - they can occur at 120-130 ms. Second, fast saccades towards faces are selective - they are more accurate and faster by about 60-70 ms than saccades towards cars. Third, saccades towards faces are reflexive - early saccades in the interval of 120-160 ms tend to go to faces, even when cars are the target. These findings suggest that the speed, selectivity, and reflexivity of saccades towards faces do not depend on the orientation or contrast of the face image. Our results accord with studies suggesting that fast saccades towards faces are mainly driven by low-level image properties, such as amplitude spectrum and spatial frequency.
Identifiants
pubmed: 33866144
pii: S0042-6989(21)00064-X
doi: 10.1016/j.visres.2021.03.009
pii:
doi:
Types de publication
Journal Article
Research Support, Non-U.S. Gov't
Langues
eng
Sous-ensembles de citation
IM
Pagination
9-16Informations de copyright
Copyright © 2021 Elsevier Ltd. All rights reserved.