Abstract
In everyday visual object perception, simple shapes are often combined to form complex ones according to Gestalt grouping principles. However, how grouping between shapes is implemented at the neural level and where in the human brain different types of grouping are encoded are not fully understood. In this study, using fMRI multi-voxel pattern analysis (MVPA), we examined the neural representation of grouping by spatial connectedness and temporal synchrony in the ventral visual cortex of the human brain. Specifically, in pre-defined regions of interest, we obtained fMRI response patterns for two shapes shown either (1) spatially connected and presented simultaneously, (2) spatially disconnected and presented simultaneously, or (3) spatially disconnected and presented asynchronously. We also obtained fMRI response patterns for each shape shown alone. We then linearly combined the patterns for the individual shapes and tested the similarity between our synthesized two-shape patterns and the actual two-shape patterns using a linear classifier. Gestalt principles have argued that grouping results in the whole to be different from the sum of the parts. This would predict that the difference between the actual and the synthesized two-shape patterns would be greater for the grouped than the ungrouped shapes. With this approach, we observed grouping by spatial connectedness and temporal synchrony in multiple retinotopic areas as well as the lateral occipital region. These results demonstrate the feasibility of using fMRI MVPA to study the neural representation of Gestalt grouping.
Meeting abstract presented at VSS 2015