Attentional capacity limit for visual search causes spatial neglect in normal observers.

When observers simultaneously monitor several positions in the visual field, distracting stimuli have a devastating effect on the ability to discriminate between similar shapes. For example, the minimum tilt necessary for an observer to discriminate between a clockwise and anticlockwise tilt has been shown to increase with the square root of the number of untilted distractors. Here we show that these rapid visual searches remain inefficient even with extended practice. Moreover, each of our observers performed particularly poorly when uncued targets appeared in certain idiosyncratic positions, as though he or she neglected to process part of the visual field. This type of neglect is not commensurate with the popular 'max rule' strategy, in which observers simply report the direction of the largest apparent tilt. Nor is it consistent with tilt averaging. It is, however, consistent with an attentional effect in which both the signal and the noise from neglected positions are decreased, leaving the local signal/noise ratio constant. We show that our data can be well fit by models in which discriminations are based on a combination of these locally weighted, noisy signals.