Abstract
Does calling two things by the same name make them harder to visually discriminate? We investigated the effects of newly learned semantic and categorical information for simple visual stimuli on the efficiency of visual search. We sought to test (1) whether training subjects to label simple shapes with common words could affect search efficiency (following Smilek et al, VSS 2003), (2) whether the effects of categorization on search could be eliminated by a secondary task placing demands on verbal working memory, and (3) whether the same results could be observed with nonsense labels instead of real word labels. Subjects received 30 minutes of categorization training in which they learned to label simple shapes (e.g., oriented lines) with common category names (e.g., cats and dogs). After the training they were tested in a present/absent visual search paradigm. By counterbalancing across subjects, each target-distractor pair had the same label for half of the subjects and different labels for the other half. A category advantage was computed for each subject by subtracting the search time for different-category pairs from that for same-category pairs. When stimuli were labeled with real words, there was a trend towards a category advantage. This trend was reversed when subjects had a secondary verbal task during search, with a significant interaction between category (same or different labels) and secondary task (none or verbal interference). With nonsense labels, there was no category advantage in the baseline task and there was no effect of verbal interference. These results suggest first, that semantic information about simple visual stimuli can affect visual tasks, and second, that the interaction between category and search depends on the on-line access to language, at least for newly learned categories.
Supported by NSF and Searle Scholars grants to Lera Boroditsky.