Authors:

Duncan, J. & Humphreys, G.

Reference:

Year of publication:

1992

CBU number:

2744

Abstract:

Recently Treisman (1991) described a series of visual search studies aiming to test between feature integration theory and an alternative (Duncan & Humphreys, 1989) according to which feature and conjunction search are basically similar. We point out that the latter account has two distinct levels: first, a summary of search findings in terms of stimulus similarities, and second, a theory of how visual attention is brought to bear upon relevant objects. Working at the first level, Treisman found that, even when similarities were calibrated and controlled, conjunction search was much harder than feature search. The theory, however, can only really be tested at the second level, since the first is an approximation. An account of the findings is developed at this level, based on the two processes of input-template matching and spreading suppression. New data show that, when both these factors are controlled, feature and conjunction search are equally difficult. Finally we consider possibilities for unification of the alternative views.