Multidimensional-expertise space: Multidimensional scaling changes after expertise training with objects.

Perceptual experts are defined by their ability to make fast and accurate identifications of objects within their domain of expertise at specific levels of categorization (Tanaka & Taylor, 1991). Researchers have previously applied a multidimensional space (MDS) framework to better understand the psychological similarities of members of the same object or face category. For example, difficulties in recognizing other-race faces have been attributed to a more densely clustered MDS space of other-race versus own-race faces (e.g., Byatt & Rhodes, 2004; Papesh & Goldinger, 2010). Here, we used MDS analyses to examine whether 9 hours of expertise training with computer-generated objects influenced density of clustering and whether this differed for an untrained control group. Adults completed similarity ratings and training with two "families" of objects, one trained at a basic level (multiple exemplars of 10 species all labeled "Other") and one trained at a subordinate level (multiple exemplars of 10 species, labeled "A" through "J"). MDS plots were constructed based on dissimilarity ratings between object species before and after training. For the pre-training group, and for an untrained control group, MDS results show two distinct clusters of species that fall in line with the two family distinctions. After subordinate-level training, the density of clusters changed along two hypothesized dimensions, family membership and feature distinctiveness. In contrast, basic-level training only led to changes along the dimension of family membership. These results suggest that basic-level and subordinate-level expertise training leads to qualitatively different changes in the density and specificity of psychological object representations. Meeting abstract presented at VSS 2015.