Computation, representation and multisensory integration of quantities in the human brain
Abstract:
Perception and cognition of quantities like numerosity (object number) and event timing guides human behaviour. Here I will describe our recent 7T fMRI studies of human neural quantity processing. First, I will describe a network of areas in the multiple demand network where tuned responses to visual numerosity, object size and event timing are all represented in topographic maps. Second, I will describe a similar mapping of haptic numerosity, the number of objects in the hand, in the same areas. Responses in the same brain areas are often taken as evidence of a common neural representation. But by looking at tuning for these different quantities in specific neural populations within these areas, we see they are distinct and not mapped onto a common representation. Finally, I will ask how visual numerosity is estimated in the brain. We demonstrate that the aggregate Fourier power of lab numerosity displays follows numerosity so closely that it is effectively equivalent, regardless of non-numerical features. Recent neuroimaging results show aggregate early visual responses closely follow numerosity. We show that aggregate Fourier power predicts these early visual responses more closely than numerosity, but numerosity predicts responses in the multiple demand network more closely. This close relationship to Fourier power demonstrates that animal visual systems can compute visual numerosity straightforwardly, while other computations must estimate other quantities. Once estimated, we propose that distinct and intermixed populations responding to different quantities allow our cognition to flexibly use and integrate different quantitative information for complex cognitive tasks.