Two processing modes in visual search

Abstract

Treisman's Feature Integration Theory and Cave and Wolfe's Guided Search Model are currently the two most prominent and widely accepted theories of visual search. Each accounts for data in a variety of paradigms but it appears that the principal theoretical position of each, that individual attributes of objects independently influence the efficiency of visual search for their conjunctions, needs qualification. This paper proposes two modes of visual search, one in which attributes act independently to guide focal attention and another in which attributes interact in guiding focal attention. Evidence for attribute interaction is found in three different experiments. In experiment 1, search for conjunctions of color and size were faster than for either single attribute alone and faster than could be predicted from a statistical facilitation from two independent contributions to visual search efficiency. Experiment 2 showed how field effects (target-distractor similarity and distractor heterogeneity) differ depending on whether attributes are predicted to interact or act independently. Experiment 3 further examined the effect of target-distractor similarity, showing a difference from predictions made by models of attribute independence.