Purchase this article with an account.
Rebecca M. Jones, Jonathan D. Victor, Mary M. Conte; Classification image analysis reveals different cognitive strategies for symmetry and face processing. Journal of Vision 2012;12(9):632. doi: 10.1167/12.9.632.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Introduction: Symmetry and face processing have some common characteristics, but they may rely on different cognitive strategies. Here we used classification image analysis (Eckstein & Ahumada, 2002) to determine subject’s strategies, and to distinguish parts-based from holistic mechanisms. Methods: By manipulating face photographs, we created a library of 18 x 24-check binary (black and white) images in which symmetry and face-likeness varied independently. Subjects (N = 6) identified the target that was most symmetric among three distractors of equal face-likeness (experiment 1), and identified the target that was most face-like among three distractors of equal symmetry (experiment 2). Classification images were constructed by correlating the symmetry of each check-pair (i.e., whether the luminance of a check matched the luminance of its mirror reflection in the opposite half of the image) with the probability that the image was identified as the target. Results: Classification images based on check pairs showed that subjects employed different strategies for the two tasks, and for upright versus inverted faces. In experiment 1, the largest number of significant contiguous checks was in the eye region; the observed clusters would have occurred less than ~ 2% of the time by chance. In experiment 2, the largest cluster of contiguous checks was on the midline; a cluster this size along the midline would only occur ~ 4% of the time by chance. Conclusions: Task demands and image orientation both influenced the classification images, suggesting subjects relied on different holistic processing strategies for symmetry processing and face identification.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only