Contentopic mapping in ventral and dorsal association cortex: the topographical organization of manipulable object information

Abstract

Understanding how object information is neurally organized is fundamental to unravel object recognition1-4. The best-known neural organizational principle of information is topographical mapping of specific dimensions. Such maps have been shown mainly for sensorimotor information within sensorimotor cortices (e.g., retinotopy)5-9. Thus, here we ask whether there are topographic maps - by analogy, contentopic maps - for mid-level object-related dimensions. We used functional magnetic resonance imaging and population receptive field analysis7 to measure tuning of neural populations to selected manipulable object-related dimensions. Here we show maps in dorsal and ventral occipital cortex that code for the score of each object on each target dimension in a linear progression following a particular direction along the cortical surface. Maps for each dimension are distinct, and consistent across individuals. Thus, object information is coded in multiple topographical maps - i.e., contentopic maps. These contentopic maps refer to intermediate level visual and visuomotor representations, and are potentially computed from the grouping of lower-level information through non-linear transformation following gestalt principles4,10. This shows that topography is a widespread and non-incidental strategy for the organization of information in the brain that leads to greatly reduced connectivity-related metabolic costs and fast and efficient readouts of information for stimuli discrimination11,12.