Abstract
It has been long reported that abrupt onset consistently captures attention. However, the question remains that exactly what specific property of the onset causes attention capture. The topological approach suggests a novel analysis of the nature of onset: as topologically invariant transformation neither creates nor destroys an object, an object onset is essentially a topological change. A counterintuitive prediction of this topological account is that a topological change (in, for instance, holes) should, like onset, also captures attention, and even with the same potency as onset. We investigated this prediction in 11 experiments using a modified Yantis and colleagues' onset paradigm. Three main conditions were compared. In the condition of topological change of forms (TC), topological changes were specifically manipulated by introducing or removing holes in two items, one from a placeholder display and the other at the corresponding location in the test display (for instance, H-like into O-like or P-like letters, arrow into triangle, solid disk into ring or disk containing two-holes, and others; or vise versa). In the condition of onset, an item appeared in a test display at a location corresponding to a previously blank position in a placeholder display, and otherwise the placeholder and test displays were the same as those in TC. And in the condition of non-onset, changes were made between two items at the same location, from a placeholder and a test display respectively, in various non-topological form properties, luminance, and color. Subjects were required to search whether there was an odd item in test displays, and RTs were measured. The results consistently supported the topological account: both topological change of forms and abrupt onset captured attention as revealed by the same level of facilitation in visual search functions, while they both significantly speeded performance compared to the non-onset changes in non-topological features.
Meeting abstract presented at VSS 2017