The roughness of standardized surfaces (embossed dot arrays or gratings) was compared by scanning them with the little finger of either hand while the ulnar nerve was cooled unilaterally at the elbow; both hands remained warm. Across-hand comparison of roughness showed that a given surface felt smoother on the cooled side. When the surfaces were adjusted to feel equally rough, that on the cooled side would normally have felt appreciably rougher. The effect of the nerve cooling on axonal conduction was monitored during the psychophysical experiments by stimulating the ulnar nerve above the cooled region and recording the EMG of an ulnar-innervated hand muscle. During cooling, large myelinated axons remained unblocked, but prolongation of their absolute refractory period to 5–10 msec left them unable to transmit trains of impulses at high frequencies (Wedensky inhibition). By varying the length of nerve cooled and the cooling temperature, it was shown that the perceptual effects were not due to an increase in the normal temporal dispersion of impulses transmitted by different-sized afferents. The effect of increasing the absolute refractory period on the signals from the various types of cutaneous afferent was modeled mathematically, using earlier human single-fiber responses to dot arrays. It is concluded that the reduction of perceived roughness with nerve cooling is due to Wedensky inhibition, and that the percept of roughness is related to the local contrast in the afferent spatiotemporal image.