well, that it fails to activate the SA1 system, and that the PC system cannot account for spatial pattern recognition performance with the Optacon;52,53 therefore, the many psychophysical studies employing the Optacon54 are studies of the sensory capacity provided by the RA system. The human ability to resolve spatial patterns with the

Optacon is exactly that which would be predicted based on the RA responses illustrated in Figure 3.2. For example, humans cannot discriminate the orientation of an Optacon grating pattern until the grooves in the grating exceed 5 mm width.50 This implies that the SA1 system is solely responsible for the limits of resolution illustrated in Figure 3.1. Curvature Perception

Combined psychophysical and neurophysiological studies of curvature perception provide evidence of the neural mechanisms of form perception not based on the limits of spatial acuity.55-58 These studies by Goodwin, Wheat, and their colleagues show that estimates of curvature are unaffected by changes in contact area and force and, conversely, estimates of force are unaffected by changes in curvature. This latter finding is particularly surprising considering that SA1 firing rates are strongly affected by curvature.58-59-60 These psychophysical observations (that curvature perception is unaffected by changes in contact area or force) suggest that the spatial profile of neural activity in one or more of the afferent populations is used for the perception of curvature and that a different neural code (e.g., total discharge rate) is used for the perception of force. Only the SA1 population response provides a veridical representation of curvature that can account for the psychophysical obser-vations.58-61 The SA1 population responses to a wide range of curvatures are shown in Figure 3.5. RAs respond poorly to such stimuli and provide no signal that might account for the ability of humans to discriminate curvature.58,62,63 Texture Perception

Our knowledge of texture perception and its neural mechanisms has changed dramatically in the last decade. A major step is the demonstration that texture perception involves two strong dimensions, roughness and softness, and a weaker third dimension described as something like stickiness. Multidimensional scaling studies have shown that texture perception includes soft-hard and smooth-rough as independent perceptual dimensions, surface hardness and roughness can occur in almost any combination, and that they account for most or all of texture perception.64,65 A third,

FIGURE 3.2 (previous pg.)Responses of SA1, RA, and PC afferents to a grating pressed into the skin. The grating is shown in cross section beneath each response profile. The bars are 0.5 mm wide; the grooves are deeper than illustrated (2.0 mm deep) and are 0.5, 0.5, 0.75, 1.0, 1.5, 2.0, 3.0, and 5.0 mm wide. The responses displayed in each profile were obtained by indenting the skin to a depth of 1 mm, holding the indentation for 1 second, raising the grating, and then moving it laterally by 0.2 mm before the next indentation. The horizontal dimension of the response profile represents the location of the center of the receptive field relative to the grating; for example, the left peak in the SA1 response profile (approximately 95 imp/s) occurred when the center of the SA1 receptive field was directly beneath the left edge of the grating. The RA illustrated here was the most sensitive to the spatial structure of the grating of all RAs studied. Some RAs barely registered the presence of the 5 mm gap even though they responded vigorously at all grating positions. Adapted from Phillips, J.R. and Johnson, K.O., J. Neurophysiol. 46, 1192-1203, 1981, with permission.

FIGURE 3.3 Confusion matrix of responses obtained from humans in a letter recognition task (top) and responses of a monkey SA1 afferent to the same letter stimuli (A-Z), which approximate the neural images of letter stimuli conveyed to the brain (bottom). The confusion matrix is derived from the pooled results of 64 subjects who performed either the active or passive letter identification task. The letters were raised 0.5 mm above the background and were 6 mm high. Matrix entries represent the frequencies of all possible responses to each letter (e.g., the letter A was called N on 8% of presentations). The numbers in the bottom row represent column sums. The numbers in the right-most column represent the number of presentations of each letter. Boxes around entries represent letter pairs whose mean confusion rates exceed 8%. For example, the mean confusion rate for B and G is 8% because G is called B on 11% of trials and B is called G on 5% of trails. The neural image (bottom) was derived from action potentials recorded from a single SA1 afferent fiber in a monkey. The stimuli consisted of the same embossed letters as in the letter recognition task scanned repeatedly from right to left across the receptive field of the neuron (equivalent to finger motion from left to right). Each black tick in the raster represents the occurrence of an action potential (see legend of Figure 3.4 for details). Adapted from Vega-Bermudez, F., Johnson, K.O., and Hsiao, S.S., J. Neurophysiol. 65, 531-546, 1991, with permission.

0 0

Post a comment