The Function of Sensory Maps

Empirical evidence has revealed that the behavior of neural cells may be correlated with salient features of the host’s environment. On the assumption of there being an internal entity which treats such behavior as a stand-in for the relevant feature, it is possible to interpret the functional role of these cells in representational terms. But it is puzzling why they are often organized in array-like structures, as in the case of the lineorientation columns found in the visual cortex of the cat. If the functional role of individual cells is representational, what then is the functional role of the array-like structure? The puzzle deepens in the case of those arraylike structures found in primary sensory areas. These so-called sensory maps would seem to be in the same architectural category as other arraylike structures. But their functional role is doubly in doubt. Applying a representational interpretation in this case involves the supposition that the system is engaged in representing a part of itself, which seems counterintuitive if direct access can be achieved at the same cost. For purposes of resolving these two puzzles, an informational analysis of the sensory task is set out. This is used to analyze the architectural requirements entailed in the implementation of optimal sensory processing. It is then demonstrated that array-like structures offer informational and resource advantages in both sensory and non-sensory contexts.