Abstract
The perception of common objects elicits both visual and semantic information. We investigated the contribution of these two sources of information to object processing in a paradigm in which these sources were independently manipulated. Experimental stimuli consisted of pictorial object pairs that shared either visual or semantic features (e.g., visual: BLIMP/SHARK; semantic: GUITAR/TRUMPET). Visual and semantic similarity were independent, with visually matched items coming from distinct semantic categories and semantically matched items sharing few perceptual features. Objects were presented in pairs, with one object serving as a prime for the second target object. In two behavioral experiments, subjects either verbally named the target or responded "match/non-match" to a word label presented following the target display. Behavioral priming effects in both the picture-naming and label-matching tasks were calculated to compare the impact of similarity in the visual versus semantic conditions, relative to a neutral control condition. Results were consistent across task modality, with both tasks yielding negative priming effects in the visually congruent condition (e.g., correctly saying "shark" or correctly responding that the label "SHARK" matched when a BLIMP prime was followed by a SHARK target). Thus, sequentially viewing pictorial stimuli with a high degree of visual feature overlap led to task interference. Interestingly, these results differ from previous findings of positive priming for perceptually matched word pairs (Schreuder et al., 1984). We interpret our negative priming effect as interference stemming from activation of multiple associated labels upon viewing pictures of the objects. Since both tasks required responses at the semantic level, the presence of shared visual features in the absence of the typical semantic co-occurrence lead to interference. These findings indicate that insofar as category labels and their associated semantics impact visual object processing, these effects may actually be more significant than the perceptual similarities that arise between visual objects.
Meeting abstract presented at VSS 2012