Abstract
Introduction: Visual features are the general building blocks for attention, perception, and working memory. But it remains largely unknown whether they work interchangeably in various tasks. For example, could it be possible that colors work especially well in visual search, but bar orientations work especially well in change detection? Or are they always "general" for all tasks? Here, I explore the factors which can quantitatively predict all the differences they make in various paradigms. Method: I tried to combine the strengths of experimental and correlational approaches in a novel way by developing an individual-item differences analysis which reveals the relations between the ways in which an item is processed in multiple tasks. To reflect the wide range of the visual features and the tasks they are used in, I included 16 stimulus types and 8 tasks. A large sample size (410) ensured that all eight tasks had a reliability (Cronbach's α) of no less than 0.975, allowing the factors to be precisely determined. Results: Three orthogonal factors were identified which correspond respectively to featural strength (i.e., how close a stimulus is to a basic feature), visual strength (i.e., visual quality of the stimulus), and spatial strength (i.e., how well a stimulus can be represented as a spatial structure). Featural strength helped substantially in all the tasks but moderately less so in perceptual discrimination; visual strength helped substantially in low-level tasks but not in high-level tasks; and spatial strength helped change detection but hindered ensemble matching and visual search. Conclusion: Jointly, these three factors explained 96.4% of all the variances of the eight tasks, making it clear that they account for almost everything about the roles of these 16 stimulus types in these eight tasks.
Meeting abstract presented at VSS 2016