Purchase this article with an account.
W. Drew Bromfield, Thomas James; Task-Dependent Reliance on Image Fragments in Humans . Journal of Vision 2014;14(10):1295. doi: 10.1167/14.10.1295.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Previous work has shown that image fragments informative for one task are not always the most informative for a different task with same set of stimuli and that neural responses in humans are sensitive to differences in task-dependent information (Nestor, Vettel, & Tarr, 2008). Thus, humans may actively rely on different aspects -- possibly visual fragments or features -- of a stimulus depending on the current task. We tested this possibility using an aperture-viewing paradigm (James, Bushmakin, et al. 2011) and a 2 interval same-different task. We used two categories of stimuli: female faces and cars. There were 4 individuals in each category, with two distinct instances per individual. Different instances for faces were expressions (happy, neutral), and for cars were viewpoints. Subjects were first presented with an image behind an occluder and explored the image through an aperture that was moved continuously by the subject via the mouse. A second image was then presented and a response was made to indicate whether or not it was 'the same' as the first image. Task condition dictated the criteria for a successful same response. For categorization, two images were the 'same' if they were from the same category (cars, faces) regardless of which individual or which instance. For individuation, two images were the 'same' if they depicted the same individual, regardless of which instance. Heat maps of viewing time were thresholded to select fragments that subjects relied on for success. Both face and car fragments revealed task-dependent differences regarding which regions subjects relied on. However, face fragments exhibited significantly more overlap across individuation and categorization tasks than did car fragments. These findings suggest that while humans do rely on different regions of a stimulus depending on task, they may employ a more consistent viewing strategy across situations for faces than cars.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only