Academic journal article Attention, Perception and Psychophysics

News from the Field

Academic journal article Attention, Perception and Psychophysics

News from the Field

Article excerpt

Published online: 19 March 2013

© Psychonomic Society, Inc. 2013

SCENE PERCEPTION

Perceptual grouping in scenes

Korjoukov, L, Jeurissen, D., Kloosterman, N.A., Verhoeven, J.E., Schölte, H.S., & Roelfsema, P.R. (2012). The time course of perceptual grouping in natural scenes. Psychological Science, 23(12). doi: 10.1177/0956797612443832

The modal, hierarchical view of visual processing involves preattentive processes operating in parallel across a visual scene, followed by focal processing of a small number of objects. Under this view, perceptual organization and figure-ground determination are preattentive processes that operate rapidly to organize visual features into potential objects; the results of these organizational processes are then fed to limited capacity systems that identify and categorize objects.

As with most research on visual perception, studies of perceptual organization have relied on simple, but controlled displays. For example, studies of figure-ground organization often use bipartite displays containing two different regions or sets of regions, where one region (or set) can be perceived as a foreground figure and the other as a background. The recent paper by Koijoukov et al. (2012) departs from this approach to study perceptual organization in visual scenes, allowing them to address whether perceptual grouping occurs in parallel and without attention.

Participants viewed pictures of animals or vehicles and performed one of two tasks. In the first experiment, participants performed a grouping task. For this task, two small dots ("cues") appeared on object contours in the picture, and participants reported if the marked contours appeared on the same object or on different objects (Vecera & Farah, 1997). The distance between the dots varied across trials. If grouping was rapid and parallel, then the distance between the dots should not affect response times; the dots should be grouped rapidly when they appear on the same object, irrespective of the distance between them. In contrast with this prediction, Korjoukov et al. (2012) found an increase in response times as the distance between the dots increased, suggesting that grouping along the contours of an object involves a serial (or serial-looking) process.

In their second experiment, Korjoukov et al. (2012) asked if the apparent serial scanning result occurred after another visual process, visual categorization. Participants viewed the same displays as in the first experiment, but performed an object categorization task, reporting whether displays contained an animal or a vehicle. Participants also performed the grouping task from the first experiment. The results demonstrated that scene classification was rapid (on the order of 500 ms or less), whereas grouping was slower (600 ms or more, depending on the distance between the probed contours).

Although these findings appear to cause problems for the typical, hierarchical view of visual processing, Koijoukov et al. (2012) discuss their findings in terms of neural evidence for the time course of different tasks. Previous work has suggested that there is a fast feed-forward process that can allow some quick and dirty processing to occur, such coarse information about the content (or possibly gist) of a scene. A more finegrained analysis, such as reporting about the specific relationship between two pieces of a contour may require recurrent processes, which increases the response time associated with the fine-grained task. One important distinction that emerges from this discussion is between a processing architecture and the processing dynamics. Early visual representations that represent features and their grouping might occur at an earlier level in the visual hierarchy than recognition and categorization. But, access to this earlier information-as in Koijoukov et al.'s (2012) grouping task-might require additional time because of lateral interactions or feedback from later representations. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.