Purchase this article with an account.
Anna Cragin, Amanda Hahn, James Pomerantz; Emergent Features Predict Grouping in Search and Classification Tasks. Journal of Vision 2012;12(9):431. doi: https://doi.org/10.1167/12.9.431.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Emergent Features (EFs) are properties that are not possessed by any individual part, but arise only from the configuration of parts. They are furthermore processed more quickly than are the properties of the parts, and the presence of EFs makes configurations more salient than their parts. Converging evidence from two tasks showed strong support for EFs as the basis for grouping. In the Search Task, discrimination between the orientations of line segments (bases) was facilitated by the addition of "context" line segments, but only when those contexts created EF differences between the target shape and the shape of the distractors. Faster search times in the "composite condition" relative to the base (i.e., no context added) indicate the presence of a Configural Superiority Effect (CSE; Pomerantz & Portillo, 2011). In the Classification Task, either the base or the context was relevant for classifying composite images into predetermined response categories. Variation in the irrelevant portion of the image slowed performance relative to a control condition where the irrelevant portion does not vary from trial to trial. Such a detriment in performance is termed Garner Interference (GI; Pomerantz & Garner, 1973), and is indicative of subjects’ reduced ability or preference to pay selective attention, electing instead to pay more attention to the whole configuration. We reasoned that when elements are perceptually organized, forming a Gestalt, (1) discrimination is enhanced when contexts produce more salient properties (yielded a CSE), and (2) the ability to selectively attend to any one element is impaired (GI). Thus, CSE and GI effects can be diagnostic of grouping. Results indicated that the type and number of EF differences and similarities, coded by using the Tversky Contrast Model (Tversky, 1977), can successfully predict both of these effects and, by extension, grouping.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only