Abstract
Humans perceive a coherent visual world across time and space. To update incoming sensory information while maintaining a stable representation of the environment, it may be critical to utilize spatial-temporal continuation and/or shared local elements across views. Previous studies have suggested different roles of the parahippocampal place area (PPA) and the retrosphenial cortex (RSC) in scene processing, and the engagement of the PPA in ensemble processing. Using fMRI, we examined how the PPA and the RSC might integrate information by relying on continuous spatial-temporal sequence of information flow, or mere shared elements in a scene. We used scene images that were divided into three segments, with 66% overlap between the first and second segments and 33% overlap between the first and the third segments (Park & Chun, 2009). Across four conditions, participants (N=20) viewed identical segments for three times, three completely different scenes, or three segments of a scene either in sequential or displaced orders. We found that bilateral PPA not only showed significantly stronger activations for different than identical scenes (p’s<.0001), but also for different scenes than three segments of the same scenes (p’s<.005). For bilateral RSC, similar response amplitudes were observed for different scenes and three segments of the same scenes (p’s=.087-.57), which were stronger than that for identical scenes (p’s<.001). More importantly, using univariate and multivariate analyses, the sequence order of the three segments did not affect the response amplitude in either the PPA or the RSC (p’s>.52), nor reveal above-chance decoding accuracy in both regions (p’s>.86). The results suggest that the PPA, but not the RSC, distinguishes between same vs. different environments, despite differences across views of the same scene. Although the PPA may not be sensitive to the sequence of information flow, it appears to integrate segments with shared elements to form a coherent representation.