VISUALIZATION OF SIGNIFICANT DIFFERENCES IN SOMATOTOPIC MAPS - A DISTRIBUTED T-TEST

In order to test for differences in the properties of two populations of cells within a somatotopic map, we need to be able to compare data sets in which sampled cells are randomly scattered throughout the map, and the variable being compared varies with location in the map. We c an describe cell properties as exponentially smoothed surfaces fitted to data in the plane of the map, where all data contribute to the comp utation of the value of each grid point on the surface, with weights w hich decline exponentially with distance from the grid point. Means, v ariances and Student's t values can be computed at all grid points, ke eping in mind the fact that grid points' t values are not independent of each other. We used Monte Carlo methods to demonstrate that two ran dom samples of 500 values from two populations of 100 000 values at 40 00 grid can provide a very useful picture of regions with significant differences. We recommend this procedure, or analogous approaches usin g other statistical tests, for any analysis where it is necessary to c ompare values of dependent variables when matched locations on the ind ependent axis or plane cannot be sampled in the two populations. (C) 1 997 Elsevier Science B.V.