Abstract
Purpose: When subjects search for a target defined by a conjunction of two features, search can be quite efficient (shallow RT x set size functions). Various models, including our Guided Search model, propose that information about the target's defining features can be used to guide the deployment of attention. For example, in a search for a red vertical item, attention can be guided to red items and to vertical items. Is guidance available immediately? If not, we hypothesized that a brief search through a small array should be qualitatively different from a longer search through a larger array. Method: In three exps, we compared performance at small (1,2,3,4) and large (8,16,24,32) set sizes for “guided” conjunction search (red vertical among green vertical and red horizontal) and inefficient search (T among Ls in Exps 1 & 2, 2 among 5s in Exp 3). Small set sizes were run as a separate block in Exp 1, with all items near fixation. In Exps 2 & 3, all set sizes were intermixed with items permitted to appear throughout the 24 by 24 deg field. Results: Efficiency of conjunction search is very different for small and large set sizes. For conjunction search, average small set size slopes were steep (29, 34, & 47 ms/item for Exps 1, 2, &3), while large set size slopes were shallower (10, 13, 23 ms/item). Average slopes did not differ between search tasks for small set sizes. As usual, they did for larger set sizes. Ratios of target-absent to target-present slopes varied widely and changed dramatically over small set sizes. More typical ratios between 2:1 and 3:1 were seen for larger set sizes. Conclusions: The data strongly indicate that the mechanisms governing search are not the same at the start as they are later in the search. In particular, the similar inefficiency of all these searches at small set sizes suggests that guidance by features like color and orientation does not begin to have its effects until 4 or more items have been selected.
NIH-NEI and AFOSR