The tactile perception of stimulus orientation

Abstract

Studies of the visual system suggest that, at an early stage of form processing, a stimulus is represented as a set of contours and that a critical feature of these local contours is their orientation. Here, we characterize the ability of human observers to identify or discriminate the orientation of bars and edges presented to the distal fingerpad. The experiments were performed using a 400-probe stimulator that allowed us to flexibly deliver stimuli across a wide range of conditions. Orientation thresholds, approximately 20° on average, varied only slightly across modes of stimulus presentation (scanned or indented), stimulus amplitudes, scanning speeds, and different stimulus types (bars or edges). The tactile orientation acuity was found to be poorer than its visual counterpart for stimuli of similar aspect ratio, contrast, and size. This result stands in contrast to the equivalent spatial acuity of the two systems (at the limit set by peripheral innervation density) and to the results of studies of tactile and visual letter recognition, which show that the two modalities yield comparable performance when stimuli are scaled appropriately.

• Acuity
• discrimination
• orientation
• psychophysics
• somatosensory
• tactile

View correction statement:

Corrigendum

Notes

Notes

1.  Convergence is intended in the sense of non-parallel. The lines never actually converge to a single point.

2.  This analysis does assume that subjects use similar criteria and have comparable sensitivity, neither of which seems to be the case (Figure 5); d′s derived from pooled data tend to be underestimated relative to those obtained when the averaging is done after the computation of the d′ (compare pooled sensitivity vs mean sensitivity). However, at relatively low mean sensitivity levels (including d′ = 1.35), the difference between mean and pooled sensitivity is small, even if subjects’ criteria and sensitivities vary over a wide range (Macmillan and Creelman 1991).

3.  Unlike in the discrimination experiment, all the subjects yielded some proportion of false alarms so we were able to compute angular thresholds for individual subjects. We show the pooled sensitivity in the bottom panel of Figure 7 because some subjects yielded perfect performance for large Δφ s.