Cortical maps: where theory meets experiments

Abstract

Primary visual cortex (V1) has remarkably systematic functional maps. One commonly used class of computational models proposes that such maps are generated by a mechanism that projects the multiple dimensions of neuronal responses smoothly onto the two dimensions of cortex. In this issue of Neuron, Mriganka Sur and colleagues find a close match between such model predictions and measurements from ferret V1.

Figure 1. Schematic of Maps of Spatial Location, Orientation, Ocular Dominance, and Spatial Frequency over a Patch of Primary Visual Cortex

Neurons at the cortical position marked in red would be tuned to the particular stimulus location, orientation, ocular dominance, and spatial frequency identified by the dotted line leading up from the red spot. Neurons in the green spot would be tuned to the different set of values indicated by its dotted line. The key gives color codes identifying tunings for the different stimulus features.

Figure 2. “Ice Cube” Model for the Joint Map of Orientation and Ocular Dominance on V1 Compared to a Model with Very Poor Coverage

(Left) Ice cube model: schematic maps of orientation and ocular dominance, intersecting each other at right angles, superimposed on a map of spatial location. The black lines on the spatial map indicate the boundaries of each repeating “module.” A shift of one module corresponds to a shift to a neighboring (partially overlapping) spatial position. Since every stripe of orientation crosses all stripes of ocular dominance equally, within a module, and every stripe of ocular dominance crosses every stripe of orientation equally, each module has a complete set of all possible combinations of values of orientation and ocular dominance. Thus, coverage is maximally uniform. (Right) If orientation and ocular dominance were to run parallel to each other, any given value of orientation would be tied to only a given value of ocular dominance. Thus, V1 would represent only a small set of combinations of orientation and ocular dominance would be very sparse.