Parallel distractor rejection as a binding mechanism in search

Abstract

The relatively common experimental visual search task of finding a red X amongst red O's and green X's (conjunction search) presents the visual system with a binding problem. Illusory conjunctions (ICs) of features across objects must be avoided and only features present in the same object bound together. Correct binding into unique objects by the visual system may be promoted, and ICs minimized, by inhibiting the locations of distractors possessing non-target features (e.g., Treisman and Sato, 1990). Such parallel rejection of interfering distractors leaves the target as the only item competing for selection; thus solving the binding problem. In the present article we explore the theoretical and empirical basis of this process of active distractor inhibition in search. Specific experiments that provide strong evidence for a process of active distractor inhibition in search are highlighted. In the final part of the article we consider how distractor inhibition, as defined here, may be realized at a neurophysiological level (Treisman and Sato, 1990).

Keywords: attention; conjunction search; feature binding; inhibition; visual search.

Figure 1

Accounting for conjunction search. Three influential accounts of conjunction search are depicted. In all cases search is for a vertical red bar, amongst vertical green and horizontal red bars. (A) Illustrates the basic location based cross referencing scheme that is the core of Feature integration Theory (FIT). (B) Illustrates the inhibitory revision of FIT proposed by Treisman and Sato (1990), and (C) illustrates the guided search revision proposed by Wolfe et al. (1989). (A) FIT: stimuli are decomposed into constituent features. Serial spatial selection by attention serves to recombine features based on location. (B) Feature inhibition revision of FIT: inhibition of distractors with non-target features (dotted lines) leaves target as only remaining uninhibited item. (C) Guided search revision of FIT: activation from target features is summed in an activation map. Spatial attention selects location with highest activation. Dotted lines indicate activation of activation map. Blue columns represent activation levels. Blue arrow represents direction of spatial selection by activation.

Figure 2

Varieties of conjunction search. Much of the research on selection and binding has used visual search tasks. Participants response times to find a particular target amongst distractors is compared. (A) Shows the classic “conjunction search” display, shown by Treisman and Gelade to produce inefficient search performance. (B–E) Each illustrate cases of efficient conjunction search. (A) Treisman and Gelade (1980). Inefficient color – form conjunctions. Finding the green N is difficult. (B) Wolfe et al. (1989). Efficient color – orientation conjunctions. Finding the red vertical bar is easy. (C) McLeod et al. (1988). Efficient motion – form conjunctions. Finding the moving X (arrows indicate motion) is easy. (D) Nakayama and Silverman (1986). Efficient color – depth conjunctions. Finding the front blue square is easy. (E) Duncan and Humphreys (1989). Efficient orientation – orientation conjunctions. Finding the L is easy.

Figure 3

Illustration of Kim and Cave (1995) probe-dot study of conjunction search. Target is a red square. Distractors share either shape (green square), color (red circle) or neither feature (green circle) with the target.

Figure 4

Illustration of the Dent et al. (2012) paradigm. Arrows next to items indicate oscillatory motion. Arrows underneath stimuli panels indicate the passage of time.