Abstract
How multiple distinctive objects may be represented in the human brain is largely unknown. A multi-object display can be characterized by three features: The identities of the objects, the global layout of the objects, and the precise relationship between adjacent objects. Using fMRI MVPA, here we investigated how these features of a multi-object display may be represented in the human brain by examining responses from topographically defined early visual areas (V1-V4), object processing regions in higher ventral regions (LO and pFs) and parietal regions (inferior IPS and superior IPS). The experimental displays were created by manipulating the locations of three artificial objects or three novel shapes on a donut-shaped background. In order to examine layout representation independent of location representation, while keeping layout the same, the three objects rotated around the donut within a block of trials. We also included a control condition to assess the contribution of spatial envelope to spatial layout representation. Participants viewed the displays and performed an orthogonal size change detection task. In V4 and pFs, we observed successful object identity decoding. In LO, we observed not only object identity decoding, but also decoding for the global layout of the objects and the precise pairing between objects. Importantly, the global layout decoded in LO could not be attributed to the representation of the spatial envelope of the multi-object display. In contrast, spatial envelope contributed significantly to global layout decoding in inferior IPS and superior IPS. These results show that all three features characterizing a multi-object display are represented in LO, suggesting that this brain region likely plays an important role mediating the perception and representation of multi-object displays.
Meeting abstract presented at VSS 2018