UC San Diego

The human mind rapidly parses high-dimensional dynamic input into useful representations. Identifying the source of relevant sensory information in time is critical to the formation of informative perceptions. One source of information to aid perceptual processing is the temporal regularity of sensory input. Knowledge of what regularities are utilized and how they are utilized is critical to understanding how perception functions in a dynamic world. The objective of this dissertation is to explore the adaptations of perception to temporal regularities in the environment. Chapters 1 and 2 use short term priming paradigms to examine the temporal segregation of perceptual information. Chapter 1 shows that inverted versus upright faces have different time courses of priming, suggesting that familiarity plays a role in temporal segregation. Furthermore, for faces, there are individual differences in priming dynamics related to perceptual identification ability. Chapter 2 examines the influence of occurrence and co-occurrence statistics in the same task. Several incremental reductions of stimulus frequency at different levels of representation all slow priming compared to higher occurring word stimuli. Word associations as a measure of co-occurrence, produce waxing and waning priming effects, with a small effect of the order of the associative relationship. Priming based on direct measures of linguistic co-occurrence produces priming dynamics that are similar to associative priming. Words which are expected to repeat provide a small increase in priming compared to priming by repeating a word. These priming effects are interpreted as the result of habituation. The findings in Chapters 1 and 2 generalize the habituation account of priming effects to faces, demonstrate its dependence on the occurrence rates of stimuli, and show that temporal co-occurrence regularities for particular items only result in small changes to the basic dynamics. Chapter 3 uses a spatial cueing paradigm to test if we adapt to diagnostic spatial-temporal relationships. Participants implicitly learn and generalize diagnostic cue-target relationships with long delays between cues and target. Diagnostic relationships are also learned for short cue-target delays when no shift of expectation is required. As a whole, these experiments find that temporal regularities pertaining to familiarity and general occurrence relationships are strongly reflected in the dynamics of perception, and that directional associations or expectations between individual stimuli deviate only slightly from general dynamics