Abstract
Using a variant of the visual search task, we investigated how figure-ground segmentation processes might lead to the camouflage of a target item when it is embedded in an array of distractor items. Each search item consisted of a black region and a white region, roughly equal in area, such that either the white or black region denoted a meaningful and familiar object (i.e., it had high denotivity), whereas the other region did not depict a known object (i.e. it had low denotivity). Participants were first shown each image in isolation and asked to identify which region denoted a meaningful object. During subsequent search trials, participants were presented with a 4x4 array of items and instructed to locate the specified target item as quickly as possible on each trial. Both eye movements and response data were collected during search. Participants made a speeded detection response with the space bar, followed by a localization response to a masked display using the mouse (to evaluate accuracy). Critically, congruency of the target and distractors was manipulated. On congruent trials, the highly denotive region in each of the images was the same color (e.g., all black). On incongruent trials, the highly denotive region in each of the distractor images was the same color (e.g., black), while the highly denotive region of the target was the opposite color (e.g., white). We found that incongruent trials produced a ‘camouflage’ effect, whereby participants were significantly slower to locate incongruent target items compared to their congruent counterparts. This result indicates that segmentation of individual items is modified by the dominant global segmentation, inducing participants on incongruent trials to misclassify the ground region of the target as the figure, thus leading to a delay in target recognition.
Meeting abstract presented at VSS 2012