Abstract
If you were looking for electrical outlets, your search would be guided by knowledge of where outlets are typically located. This differs from other forms of attentional guidance in visual search. You would not be guided by target features (as in a search for red among green) nor would you be guided by specific location information (as in spatial cueing experiments). Rather you would be guided by what could be called global or distributed spatial information (e.g. outlets are typically found on walls, near the floor). How does distributed spatial guidance compare to other forms of guidance? Observers viewed an unchanging loose pyramid of cubes. Each cube had three visible surfaces: the top and two sides. Cube tops did not form a coplanar surface. On each trial, Ts and Ls were distributed randomly over these surfaces and observers searched for the letter T. There were three conditions. In the No Guidance condition, observers simply searched for a T. In the Color Guidance condition, one third of items were red and observers were told that the T, if present, would be red. In the Global Location Guidance condition, observers were told that the T, if present, was on the top of a cube. The slope of the RT × Set Size function was 42 msec/item in the No Guidance condition. Color Guidance produced a slope of 18 msec/item, roughly what would be expected if observers restricted search to red items. Global Location Guidance produced a steeper slope (24 msec/item) and a substantially higher intercept (815 vs 615 msec for Color Guidance). Even in a fixed scene, global spatial information appears to take longer to become effective. Perhaps feature guidance can occur in a feed-forward manner while global spatial information requires reentrant processing.
supported by NIMH & AFOSR