Abstract
Visual crowding degrades feature discrimination and object recognition in the periphery, yet several studies have demonstrated continued processing specific to the crowded target (e.g. He et al., Nature, 1996; Parkes et al. Nature Neuro, 2001; but see Hancock et al, JOV, 2008). Findings of residual processing can be utilized to test mechanism(s) of crowding (for a review, see Levi, Vis. Res, 2008). We examined associative priming in flanked and unflanked conditions using a lexical decision task (LDT). In a dual-task threshold experiment, subjects viewed a flanked or unflanked word (or non-word) in the periphery and made a 2AFC word/ non-word discrimination. Second, they reported any individual letters seen in the letter string. Across the seven eccentricities tested, we found the classic pattern of crowding for both tasks. In the main experiment, a flanked or unflanked prime word preceded a semantically related or unrelated target word that appeared at fixation (e.g. target ‘WEST’ paired with prime ‘EAST’, or ‘TAKE’). The prime word appeared near Bouma's boundary, or at one of two peripheral eccentricities that had been determined to be well crowded in the threshold experiment. Following each LDT, subjects also made a word/non-word discrimination of the prime word to verify crowding on each trial. Across all trials, a comparison of reaction times between the related and unrelated conditions (Unrelated - Related) revealed an associative priming benefit in the unflanked condition and to a lesser degree in the flanked condition. However, when only incorrect trials were analyzed (a strict measure of crowding) priming was reduced in the flanked condition. Four of five subjects followed this pattern of results. These findings suggest that crowding reduces the semantic priming effect.