A Computational Perspective on Segregated Processing and the Privileged Role of Space in Binding

Much neuroscience research supports segregation of visual processing.Using artificial neural networks (ANNs), we previously showed that segregated streams trained to either identify objects or determine the locations and orientations of objects actively retain information about both identity and space, independently and differently.Further, ANNs with two pathways, compared to a single pathway, have better performance (higher accuracy, less training time).These advantages increase with binding difficulty (i.e., increased number of objects), suggesting segregation itself may help ease the binding problem.However, if segregated-pathway networks have advantages over single-pathway networks, why are not all attributes segregated in the brain?Simulations suggest that when there are a greater number of classes or greater discrimination is needed for a given attribute, the advantage of segregated pathways is greater.The ANNs described above used a relative location map to successfully bind identity and absolute location of objects.However, the brain needs to combine many different attributes (e.g., color and identity) and any of these attributes could be used in a similar fashion to constrain the binding problem.Using visual objects with multiple attributes (any two of the following: identity, luminance, orientation, and location) with segregated-pathway networks, we tried to find the best attribute (identity, luminance, orientation or location) to constrain the binding problem.We found that when visual attributes are independent, a location map is always the best choice for constraining the binding problem.Together these findings show that a computational approach can provide insights into why there is segregation of processing (and thus a binding problem) as well as what is the best way, at least for visual recognition, to constrain the binding problem.Implications of the findings for neuroscience and modeling will be discussed.