Abstract
Perceptual load is a major determinant of how much incidental information can be processed from a given display. Here I present evidence to support this view, showing that short-term priming, observable over several trials, functionally increases processing capacity and shifts the locus of attentional selection. Thus, both early and late selection can be observed in identical displays depending on their immediate history. Two types of tasks were used. One was a pop-out task which required the discrimination of an always-present target in a display that also contained another salient feature irrelevant to the task. This arrangement allowed examination of how memory for the relevant feature affects memory for the irrelevant feature. A second type of task was a dual task where color and shape, each defining a unique target in a three-element display, both had to be discriminated. Repetition of either or both of target-defining features allowed for examination of the effect of priming on moderating task difficulty. Using different repetition structures and manipulating the salience of target-selecting features in these tasks the following results were obtained: (1) Short-term priming exists only for the segmenting feature. (2) In a dual task, chance repetition of either segmenting feature leads to improvement in performance. (3) The encoding of the irrelevant salient feature depends on the amount of short-term priming for the relevant feature: the irrelevant feature is encoded when priming is strong, it is not encoded when priming is absent. These results suggest that the difficulty of a task and consequently the amount of irrelevant information we can extract from a visual stimulus is not an inherent property of that task/stimulus. Rather, fluctuations in the amount of priming produced by repetitions of relevant segmenting features influence the amount of allocated resources, and therefore determine how much irrelevant information we can encode at any given point in time.