Visual search does not always predict performance in tasks that require finding targets among distractors: The case of line-ending illusory contours

Does the standard search task predict performance in related tasks for Kanizsa-style illusory contours?

It is often assumed that results from standard visual search tasks will be replicated in related tasks but his idea is rarely tested. In a conceptual replication of Li, Cave, and Wolfe (2008), we investigated the attentional demands of Kanizsa-style illusory contours using orientation-based search, comparing performance for items defined by real- as compared to illusory contours. After confirming the initial findings in standard search, we tested the same manipulation in multiple-target search, Thornton and Gilden's (2007) hybrid standard/multiple-target search, and simple- and selective enumeration. The RT slope differences between real- and illusory contours did not replicate in Thornton and Gilden's task, though they did in multiple-target search and selective enumeration. In fact, absolute differences between real- and illusory contours in RT costs per distractor were 2 - 6 times larger than in standard search. To determine whether performance differences between real and illusory contours originated from shape-definition (necessary for distinguishing target shapes from distractors) or unit formation (grouping disconnected parts to define an item/unit), simple and selective enumeration were compared. The differences between real- and illusory-contours only emerged in selective enumeration (enumerating targets among distractors), which suggests the discrepancies between conditions originate from shape definition rather than unit formation processes. There was no evidence of subitizing in selective enumeration for illusory contour figures, but contrary to attention-based theories of enumeration, there was no subitizing for the real-contour controls either. This study contributes to research on illusory contours but it is especially important to the study of search and enumeration.

Why doesn't emotional valence affect subitising and counting in simple enumeration?

Accurate visual-spatial enumeration involves either the subitising process (for 1-4 items) or the counting process (for larger numbers of items). Although these processes differ, both are thought to involve attentional selection. Many studies show that emotional valence, the negativity or positivity of a stimulus, influences attention and yet Watson and Blagrove found valence had no effect on simple enumeration (enumeration without distractors). To shed light on this surprising finding, we had participants enumerate 1 to 9 dots after viewing emotional scenes, using images from the International Affective Picture System (IAPS) to manipulate valence and arousal. To ensure valence and arousal categorisations were valid for each participant, we individualised them based on their own ratings. Results indicated that both valence and arousal affected enumeration latencies, with enumeration fastest after positive high arousal images and slowest after negative low arousal images. Disengagement deficits were apparent from slowed enumeration after negative images, but there was no evidence that valence affected the breadth of the attentional focus (no interactions with display area). Despite hints that valence may affect subitising and counting differently (weak trends to a cross-over interaction in RT slopes), no firm conclusions can be made because differences were small (<20 ms/item).