Abstract
Introduction: Two predominant attention tasks, visual search and cueing, often have been treated separately with different attentional theories for the effects found in each task (visual search-set size, cueing-validity). An ideal observer/SDT approach suggests that these two effects may be modeled with a purely selective attentional mechanism with equal discriminability at all locations (e.g., Palmer, et al., VR, 2000; Eckstein, et al., JOV, 2002). In this study we show that an ideal observer model with one selective attention mechanism (weighting information by cue validity) can appropriately predict both set size and cueing effects in the same task. Method: 2 observers participated in a cued visual search of a Gaussian blob (sd=11′, contrast = +6.25% on a +7.81% contrast pedestal) in image noise (Gaussian, sd=2.73 cd/m2). The signal appeared on ½y the trials in 1 of 8 locations (eccentricity=6°), with observers making a yes/no judgment on signal presence. Effective set size (N) was set at 2, 3, 5, or 8 by precues (500ms) at the possible signal locations. There were 1 primary cue and N-1 secondary cues. On signal present trials, the primary cue always had 50% validity across set size; otherwise, the signal appeared randomly in one of the secondary cue locations. Thus, each single secondary cue had less validity (.5/(N-1)) with increasing set size. There were 2 stimulus durations: long (1s), with eye movements, and short, with durations roughly matched to the 1st saccade processing time in the other (long) duration condition (1st saccade latency - 80ms dead time). Results: The ideal observer predicted both cueing and set size effects for this task; also, the predicted effects fit well with the human observers' with the long duration. Observers' performance for the shorter duration and for 1st saccade localization (long duration only) were predicted less well, suggesting that shorter stimulus durations hampered optimal decisions for both perceptual judgments and saccades in this task.