DORSAL HORN SPATIAL REPRESENTATION OF SIMPLE CUTANEOUS STIMULI

A model of lamina III-IV dorsal horn cell receptive fields (RFs) has b een developed to visualize the spatial patterns of cells activated by light touch stimuli. Low-threshold mechanoreceptive fields (RFs) of 55 1 dorsal horn neurons recorded in anesthetized cats were characterized by location of RF center in cylindrical coordinates, area, length/wid th ratio, and orientation of long axis. Best-fitting ellipses overlapp ed actual RFs by 90%. Exponentially smoothed mean and variance surface s were estimated for these five variables, on a grid of 40 points medi olaterally by 20/segment rostrocaudally in dorsal horn segments L-4-S- 1. The variations of model RF location, area, and length/width ratio w ith map location were all similar to previous observations. When ellip tical RFs were simulated at the locations of the original cells, the R Fs of real and simulated cells overlapped by 64%. The densities of cel l representations of skin points on the hindlimb were represented as p seudocolor contour plots on dorsal view maps, and segmental representa tions were plotted on the standard views of the leg. Overlap of modele d and real segmental representations was at the 84% level. Simulated a nd observed RFs had similar relations between area and length/width ra tio and location on the hindlimb: r(A) = 0.52; r(L/W) = 0.56. Although the representation of simple stimuli was orderly, and there was clear ly only one somatotopic map of the skin, the representation of a singl e point often was not a single cluster of active neurons. When two-poi nt stimuli were simulated, there usually was no fractionation of respo nse zones or addition of new zones. Variation of stimulus size (area o f skin contacted) produced less variation of representation size (numb er of cells responding) than movement of stimuli from one location to another. We conclude that stimulus features are preserved poorly in th eir dorsal horn spatial representation and that discrimination mechani sms that depend on detection of such features in the spatial represent ation would be unreliable.