Torralbo, A; Walther, DB; Chai, B; Caddigan, E; Fei-Fei, L; Beck, DM; (2013) Good exemplars of natural scene categories elicit clearer patterns than bad exemplars but not greater BOLD activity. PLOS ONE , 8 (3) , Article e58594. 10.1371/journal.pone.0058594.

Preview

PDF journal.pone.0058594.pdf Download (1MB)

Abstract

Within the range of images that we might categorize as a “beach”, for example, some will be more representative of that category than others. We used ‘good’ and ‘bad’ exemplars of six natural scene categories to confirm that human categorization is sensitive to this manipulation and explore whether brain regions previously implicated in natural scene categorization show a similar sensitivity to how well an image exemplifies a category. Participants were more accurate and faster at categorizing good exemplars of natural scenes. A classifier trained to discriminate patterns of fMRI activity associated with the viewing of our scene categories showed higher decoding accuracy for good than bad exemplars of a category in the PPA, RSC and V1. A univariate analysis, however, revealed that there was either no difference in overall BOLD signal evoked by good and bad scenes (in RSC and V1) or the signal was actually higher for bad scenes (in PPA), suggesting that good exemplars produce a qualitatively, rather than quantitatively, better pattern of activity for categorizing natural scenes. Overall, our results provide further evidence that V1, RSC and the PPA contain information relevant for natural scene categorization. Finally, image statistic analysis shows that good images in our categories produce a more discernible average image and are more similar to each other. These results are consistent with both low-level models of scene category and models in which the category is built around a prototype.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item