Abstract
Introduction: A cueing effect, in which valid cues lead to better performance than invalid cues, is commonly assumed be caused by limited attentional resources improving perceptability at the cued location, at a cost to the uncued location (i.e., Posner, 1980). However, an ideal observer, which assumes parallel and equal perceptability at both locations, can also predict cueing effects (Eckstein, et al., 2002). Classification images were employed in a cueing task to assess the dynamics of information use at the cued and uncued locations. Limited resource attention models might suppose a delay in information use at the uncued location (e.g., a serial ‘spotlight’), whereas a parallel model (e.g., ideal observer) would predict no such delays.
Method: Two observers participated in a yes/no contrast discrimination task of vertical Gabors (1 cpd, 1 octave bandwidth, full-width, half-height). On each trial, two flickering Gabors appeared in the upper left and right, displaced 5° horizontally and vertically from central fixation (eccentricity = 7.1°). Half the trials contained two low-contrast (pedestal) Gabors (mean peak contrast = 23.4%); the other half contained one pedestal and one signal Gabor (mean peak contrast = 39.1%), and observers judged signal presence. The stimulus duration was 272 ms, divided into 12 intervals (23 ms/interval), in which the Gabor contrasts changed randomly on each interval (Gaussian-distributed, σ = 15.6% contrast). Dark square cues (4°) indicating signal locations (when present) with 80% validity appeared around one location simultaneously and continuously with the stimulus.
Results: Both observers had modest cueing effects (valid hit rate - invalid hit rate: 0.183, 0.156). First evidence of information use at the uncued location was 92 ms, compared to 46 ms at the cued location. However, there was considerable overlap in the time courses at the two locations, and the time of peak information use did not differ (138 ms).