Abstract
During a brief visual glance, multiple stimuli often compete for representation, leading to degraded encoding of stimuli's identities, particularly when these appear outside the focus of visual attention. Similarity among stimuli plays an important role in competition, as reduced encoding precision is typically observed with increased stimulus similarity. Similarity, however, can be perceived or judged according to different stimulus properties and at different processing levels - a featural (low) or a conceptual (high) level of representation. The present study tried to tease apart perceptual from conceptual (e.g., categorical) similarity factors, by holding constant each of these dimensions while assessing their unique contribution to stimulus competition. Using a "dilution" paradigm, we measured the extent to which neutral stimuli diluted (i.e., attenuated or eliminated) distractor interference effects in a flanker task, as a function of diluter-distractor similarity. Target and distractor were both composed of digital letters (H,T), while diluter stimuli were selected from one of two conceptual categories - letters or digits. Critically, the diluters were identical in their low-level features across both categories (e.g., the digital letter E vs. the "mirrored" digital number 3, digital P vs. digital 9, etc.). When holding constant category and manipulating diluters' visual appearance, a rather robust perceptual manipulation (color) was required to reduce diluter-distractor competition. However, when controlling for perceptual similarity, a significantly larger dilution effect (indicating competition) was observed among the letter- than the digit-diluter condition, suggesting an additional strong inter-category source of competition. This latter effect, however, was obtained only when the different categories were presented in separate blocks, not when mixed within the same block. Taken together, our findings suggest that competition occurs at both featural and categorical levels, yet processing unattended categorical information (i.e., of diluter stimuli) may strongly rely on continuous task-related strategies, expectancies, and/or recent experience (e.g., trial-to-trial priming).
Meeting abstract presented at VSS 2018