Abstract
Previous functional Magnetic Resonance Imaging (fMRI) research using multi-voxel pattern analysis (MVPA) has identified the parahippocampal place area (PPA) as a brain region responsible for discriminating natural scene categories and determined that color photographs (which contain color, texture, and structure) and contour line drawings (which contain just structure) yield equivalent category-specific patterns of activation in PPA (Walther et al., 2009; 2011). While these findings suggest PPA primarily uses structural information to represent scene category, no research has used line drawings to investigate the role of structural information in the formation of neural representations for individual scenes within a category. In this study, we seek to explore neural representations for scenes by investigating the representational similarity of individual color photographs and line drawings depicting natural scenes in PPA. We used an event-related fMRI design to measure repetition suppression between pairs of color photographs and line drawings. In two pure conditions, the same color photograph or the same line drawing was shown for both presentations. In two mixed conditions, either a color photograph or line drawing was presented first while an image from the opposing image modality depicting the same scene was presented second. Results show robust repetition suppression is only present in PPA for pure pairs, indicating representational differences between individual color photographs and line drawings depicting the same scenes. While MVPA evidence indicates that the structure of a scene is sufficient for PPA to encode scene category, the current study provides evidence that PPA encodes an array of information from individual scenes, including color, texture, and structure. This research expands scientific understanding of the neural processes involved with perception of complex environments by identifying which dimensions of information are included in neural representations of individual scenes.
Meeting abstract presented at VSS 2013