Normalization models of contextualized touch

Abstract

Bimanual touch may require combining what is felt on the hands with where the hands are located in space. The computations supporting bimanual touch are poorly understood. We found that tactile cue combination patterns and their sensitivity to the locations of the hands differed according to the attended stimulus feature. These idiosyncratic perceptual patterns can be explained by distinct cue combination models that each involve divisive normalization, a canonical computation.