1. Textures formed by periodic dot arrays are defined by the dot density, spacing, and angular orientation with respect to the direction of motion. In this report we evaluate the effects of the dot density (intensive cues) and arrangement (spatial cues) on the ability of human subjects to discriminate texture patterns scanned across an OPTACON tactile stimulator that selective stimulates rapidly adapting cutaneous mechanoreceptors. We compared dot arrays arranged on the index finger in specific patterns (horizontal, vertical, diamond, up diagonal, or down diagonal orientation) and spaced 4.8, 7.2, or 9.6 mm apart (high, medium, and low density) with the use of a two-alternative forced-choice protocol. 2. Textures are well discriminated when their elements are tightly spaced along one axis and widely spaced on all other axes. Humans distinguish textures that differ only in orientation with mean accuracy of 75% at low density and 65% at medium density, but discriminate high-density textures poorly (mean accuracy = 48%). Accuracy is related to the angular disparity between patterns, and to similarity of spacing and orientation along major and minor axes of the arrays. Vertical and horizontal patterns are more accurately distinguished than the oblique ones, and diamond arrays are the least well discriminated. Diagonal and diamond textures are often confounded, and the up and down diagonal patterns are confused with each other particularly as the texture density rises. The preference for the vertical and horizontal patterns may relate to an interaction between the orientation axis of the texture and its direction of motion across the skin. 3. Intensive cues provided by the total number of applied stimuli supplement the spatial cues inherent to the pattern orientation, because textures that differ in both spacing and orientation are discriminated better than those that differ only in orientation or spacing. Mean accuracy ranges from 96% for comparisons of high- and low-density textures, which differ in the total number of dots by a factor of 2, to 80% when medium-density patterns are compared with high- or low-density textures. 4. Textures that differ in density but not in orientation are less well discriminated than those of different orientation. For example, 82% of patterns that differ in both density and orientation are distinguished correctly in pairings of low- and medium-density textures, whereas those that differ only in density are discriminated correctly on 45% of trials. Subjects seem to use spatial rather than intensive cues when discriminating patterns of similar density, suggesting that the similarity of form (the spatial arrangement of the closely spaced dots) is more readily apparent than small differences in spacing along the axis of motion. 5. Subjects are most most successful in differentiating texture patterns when they are able to mentally picture the orientation and spacing of the pattern. We found a strong correlation between the subjects' ability to discriminate textures of a given spacing and their ability to identify the specific texture by matching it to the appropriate visual representation. Subjects are able to identify correctly all five orientations at low and medium densities, with mean accuracy of 76%, but recognize only the vertical arrays when high-density patterns are presented. The ability to image the textures is noteworthy, because subjects received no feedback about performance. 6. Spatial imaging of textures appears limited by the diameter of cutaneous receptive fields on the hand. We propose that the structural axis of a regular texture array results from perceptual linkage of adjacent elements along one principal axis by continuous bands of neural activity when their spacing is smaller than the receptive field diameter.(ABSTRACT TRUNCATED)