Abstract
Effects of display heterogeneity on visual search efficiency are well documented (Duncan & Humphreys, 1989). Even when searching a clearly distinguishable feature singleton, attentional salience falls down with heterogeneity of distractors (e.g., Santhi & Reeves, 2004). It is presumed that the visual system is able to preattentively separate heterogeneous features to homogenous subsets and attend each subset serially to find a singleton. The issue we addressed in our study was as follows: How does the visual system process heterogeneous sets that can’t be clearly distinguished? Theoretically, it should conjoin all heterogeneous items under the same subset representation and a singleton, therefore, would become more salient despite large heterogeneity. In our visual search task observers searched for an odd-sized target (either small, or large) among 13, 25, or 37 differently sized items. There were two homogenous conditions: (1) all distractors were of medium or (2) opposite size (e.g., large distractors with small targets and vice versa). Above, two heterogenous conditions were tested. In one such conditions all distractors were of (3) medium and opposite sizes (the difference between medium and each opposite size were clearly distinguishable ). Finally, in condition (4) four transition sizes filled the gap between medium and opposite distractors providing six heterogeneous sizes. We found in the result near parallel pattern of search performance in all positive conditions. The fastest detection was predictably found for homogenous displays with opposite sizes. The slowest detection was found for two distinct sizes of distractors. The intermediate efficiency was found for both medium homogenous and heterogeneous sets with transition sizes. RTs were substantially the same in these two conditions. This suggests that the visual system does fail to separate such transitional sets to subsets and treat them as a unitary perceptual entity opposing to a singleton (despite large heterogeneity and wide range of differences).
Meeting abstract presented at VSS 2013