Poster C101, Sunday, March 25, 1:00-3:00 pm, Exhibit Hall C
Perceptual and conceptual dimensions impacting animate items in the human ventral stream
Griffin E. Koch1, Marc N. Coutanche1; 1University of Pittsburgh
The ventral stream of the human brain encodes multiple perceptual and conceptual dimensions for perceived items. Which of these dimensions impact our visual system? We report a study examining these questions of neural representations for perceived animals. We recorded brain activity during a functional Magnetic Resonance Imaging (fMRI) scan from twenty participants as they were presented with images of twelve animals. The animals were selected to vary on a number of dimensions, including taxonomic group, real-world size, and prior familiarity. We apply multivariate analysis methods, including representational similarity analysis (RSA) and machine learning classifiers, to probe the distributed patterns of neural activity evoked by these presentations. We find that an animal’s taxonomic group strongly influences how their multi-voxel activity patterns are in turn affected by other dimensions, within ventral temporal cortex. Further, we find that the reliability of neural patterns varies by conceptual dimensions. For example, images of birds showed less distinctive activity patterns than items from other categories such as insects and mammals. Additionally, we find greater dissimilarity for real-world size comparisons between large and small animals, than between similarly sized animals. Furthermore, we show that patterns of activation within the ventral temporal cortex can be a reliable predictor of dimensions such as familiarity. We examine and discuss how these findings affect existing theories of animacy and the organization of the ventral stream, as well as current views of the interaction between perceptual and conceptual neural processing.
Topic Area: PERCEPTION & ACTION: Vision