Mid-level perceptual features distinguish objects of different real-world sizes

Bria Long, Talia Konkle, Michael A. Cohen, George A. Alvarez
2016 Journal of experimental psychology. General  
Understanding how perceptual and conceptual representations are connected is a fundamental goal of cognitive science. Here, we focus on a broad conceptual distinction that constrains how we interact with objects-real-world size. Although there appear to be clear perceptual correlates for basic-level categories (apples look like other apples, oranges look like other oranges), the perceptual correlates of broader categorical distinctions are largely unexplored, i.e., do small objects look like
more » ... bjects look like other small objects? Because there are many kinds of small objects (e.g., cups, keys), there may be no reliable perceptual features that distinguish them from big objects (e.g., cars, tables). Contrary to this intuition, we demonstrated that big and small objects have reliable perceptual differences that can be extracted by early stages of visual processing. In a series of visual search studies, participants found target objects faster when the distractor objects differed in real-world size. These results held when we broadly sampled big and small objects, when we controlled for low-level features and image statistics, and when we reduced objects to texforms-unrecognizable textures that loosely preserve an object's form. However, this effect was absent when we used more basic textures. These results demonstrate that big and small objects have reliably different mid-level perceptual features, and suggest that early perceptual information about broad-category membership may influence downstream object perception, recognition, and categorization processes. Subscription orders must be prepaid. Subscriptions are on a calendar year basis only. Allow 4-6 weeks for delivery of the first issue. Call for international subscription rates. SEND THIS ORDER FORM TO
doi:10.1037/xge0000130 pmid:26709591 fatcat:sgduxq3hhbhwljq2xnhnkgwmfy