Target specificity improves search, but how universal is the benefit?

Abstract

Pictorial cues generally produce stronger search performance relative to categorical cues. We asked how universal is the benefit of a pictorial cue? To test this, we trained observers to categorize sinusoidal gratings in which categories were distinguished by spatial frequency or orientation. Next, participants completed a search task in which targets were pictorially and categorically cued. Measures of target and distractor processing that primarily rely on foveal processing showed universal benefits; however, the benefit was larger in the orientation condition. Importantly, an index of the direction of spatial attention (i.e., target guidance) showed that the orientation condition produced a pictorial benefit but the spatial frequency condition did not. Experiment 2 replicated the spatial frequency results and also included conditions that increased the discriminability, lowered the spatial frequencies, or both increased the discriminability and decreased the spatial frequencies of the categories. We found that only categories utilizing lower spatial frequencies produced a pictorial guidance benefit. This demonstrates that pictorial cues do not universally improve search performance above categorical cues; it depends on the features that distinguish the categories. Additionally, the increased discriminability condition improved guidance but failed to produce a pictorial benefit, suggesting an interesting disassociation between the amount of target guidance and the existence of a pictorial benefit. Given that perception is known to progress from low/coarse to high/fine spatial frequencies, this suggests that the pictorial guidance benefit acts on early low-spatial frequency processing only, but foveal object recognition processes utilize both early and late spatial frequency processing.

Keywords: Categorical; Guidance; Pictorial; Target representation; Visual search.

Figure 1.

Category distributions for the two category learning conditions. The Orientation condition is plotted on the left (A) and the spatial frequency condition is plotted on the right (B). Each coordinate pair indicates frequency and orientation values for a grating stimulus, which are illustrated by two examples from each condition.

Figure 2

Figure 2a. Sequence for the categorization task. Figure 2b. Sequence for the search task. In this example, the category-defining dimension is orientation. Half of all trials presented a categorical cue and the other half presented a pictorial cue. The search target is in the lower left-hand side of the search array.

Figure 3.

The percentage of trials in which the target item was the first object fixated in Experiment 1. Error bars indicate SEM. The black line represents chance (25%).

Figure 4.

Mean time spent dwelling on targets and distractors in Experiment 1. Error bars indicate SEM.

Figure 5.

a. This graph shows the two categorical distributions for the low discriminability conditions. b. This graph shows the two categorical distributions for the high discriminability condition. The displayed values are transformed according to the formulas described in Experiment 1.

Figure 6.

The means for guidance in Experiment 2. Error bars indicate SEM. The black bar represents chance (25%).

Figure 7.

Mean time spent fixating targets and distractors in Experiment 2. Error bars indicate SEM.