Abstract
Estimating the size of a space is intuitively central to our daily interactions, for example when deciding whether or not to take a crowded elevator. Here, we examined how neural areas respond to scenes that parametrically vary in the volume of depicted space. Observers were shown blocks of indoor scene categories and performed a one-back repetition task while undergoing whole brain imaging in a 3T fMRI scanner. The 18 scene categories varied in the size of depicted space on a 6 point log scale, from small and confined spaces such as closets and showers, to expansive areas such as concert halls and sports arenas. Using a regions-of-interest approach, we found that activity in the lateral occipital complex (LOC) systematically decreased as the size of space increased, showing a preference for smaller spaces (r=-.64, p<.01). On the other hand, activity in the parahippocampal place area (PPA) did not change as the size of space varied: this region responded equally strongly to all types of scenes regardless of the volume of the space (r=.14, p.1). We further examined the multivoxel pattern activity in the PPA using a linear support vector machine. Voxel patterns in the PPA classified the six different volumes of space well above chance (39% performance with leave-one-block-out cross-validation, chance level being 17%). Importantly, most classification errors were found across scenes that were close in size (within 1-2 scales), and not across scenes that were further in size (within 4-5 scales). Similar results were found in LOC (36% classification performance). These data suggest that scene volume information is coded in a distributed manner over a range of areas in the ventral visual pathway, consistent with the general idea that understanding the size of a space can influence a wide range of our interactions and daily navigation through the world.
Funded by NSF CAREER award to A.O. (IIS 0546262). We thank the Athinoula A. Martinos Imaging Center at McGovern Institute for Brain Research, MIT for help with fMRI data acquisition. SP and TK contributed equally to this work.