Sa2

Ruffini

Slow

Large l cm is asked whether the dot or ridge spacing of one surface is greater than another the response is a judgment about the surface and it provides a clue to the subject's capacity for form perception (e.g., Reference 33 and 34); if the same subject is asked whether the second surface feels rougher (or softer) than the first, the response is a description of his or her experience and it provides a clue to the subject's perception of texture.

3.2.2.1 Form Perception

The ability to discriminate object or surface features and the capacity for pattern recognition at the fingertip are the same whether the object is contacted by active touch or is applied to the passive hand.35 Form perception is affected only marginally by whether the object is stationary or moving relative to the skin; it is unaffected by scanning speed up to 40 mm/s; it is unaffected by contact force, at least over the range from 0.2-1 N; and it is affected only marginally by the heights (relief) of spatial features over a wide range of heights.36-39

Three psychophysical studies of the limit of tactile spatial acuity are illustrated in Figure 3.1. In all three studies, the element width that resulted in performance midway between chance and perfect discrimination was between 0.9 and 1.0 mm, which is close to the theoretical limit set by the density of SA1 and RA primary afferents at the fingertip.40 Acuity declines progressively from the index to the fifth finger41 and it declines progressively with age.42-44 Whether these differences in acuity are due to differences in innervation density is not known. Spatial acuity at the fingertip is the same in man and monkey.45 Spatial acuity at the lip and tongue is significantly better than at the fingertip.46-48

3.2.2.1.1 Spatial Acuity

Tactile spatial resolution of about 1 mm requires an innervation density of (at least) about one afferent per square mm and it requires that individual afferents resolve the spatial details at least as well as human subjects. Neither the PC nor the SA2 system comes close on either score.13-49 Note that the human performance illustrated

FIGURE 3.1 Human performance in gap detection (open circles), grating orientation discrimination (filled squares), and letter recognition (open squares) tasks. The abscissa represents the fundamental element width for each task, which was gap size for the gap detection task, bar width (half the grating period) for the grating orientation discrimination task, and the average bar and gap width within letters (approximately one fifth the letter height) for the letter recognition task. Threshold is defined as the element size producing performance midway between chance (50% correct for the gap and grating tasks, 1/26 for letter recognition) and perfect performance. Adapted from Johnson, K.O. and Phillips, J.R., J. Neurophysiol. 46, 1177-1191, 1981, with permission.

FIGURE 3.1 Human performance in gap detection (open circles), grating orientation discrimination (filled squares), and letter recognition (open squares) tasks. The abscissa represents the fundamental element width for each task, which was gap size for the gap detection task, bar width (half the grating period) for the grating orientation discrimination task, and the average bar and gap width within letters (approximately one fifth the letter height) for the letter recognition task. Threshold is defined as the element size producing performance midway between chance (50% correct for the gap and grating tasks, 1/26 for letter recognition) and perfect performance. Adapted from Johnson, K.O. and Phillips, J.R., J. Neurophysiol. 46, 1177-1191, 1981, with permission.

in Figure 3.1 begins to rise above chance at element sizes around 0.5 mm, which means that either the SA1 or RA system must begin to resolve spatial detail at 0.5 mm or less. Evidence that only the SA1 afferents account for the spatial resolution illustrated in Figure 3.1 comes from neurophysiological experiments in which SA1 and RA afferents were studied with the same periodic gratings used in the psychophysical experiments. SA1 responses to a periodic grating convey information about spatial structure when the groove and ridge widths are 0.5 mm wide (e.g., see Figure 3.2). When the grooves and ridges are 1 mm wide, SA1s provide a robust neural image of the stimulus. In contrast, RAs require grooves that are at least 3 mm wide before their responses begin to distinguish a grating from a flat surface; most RAs fail even to register grooves 3 mm wide.7 The RA response illustrated in Figure 3.2 was the most sensitive to spatial detail of all the RAs studied. Kops and Gardner50 obtained nearly identical results with an Optacon, a dense array of vibro-tactile probes designed as a reading aid for the blind.51 PC afferents were unable to resolve grooves that were 5 mm wide (Figure 3.2).

3.2.2.1.2 Pattern Recognition

The relationship between SA1 response properties and pattern recognition behavior in a letter recognition experiment is illustrated in Figure 3.3. In the study illustrated in Figure 3.3, Vega-Bermudez et al.35 showed that there was no detectable difference in human performance between active and passive touch and that the confusion matrix shown in Figure 3.3 is characteristic of human letter recognition performance across a wide range of stimulus conditions. Recognition behavior is highly pattern specific; recognition accuracy differs significantly between letters (ranging from 15% for the letter N to 98% for the letter I) and more than 50% of the confusions are confined to 7% of all possible confusion pairs (22 out of 325 possible confusion pairs), which are enclosed in boxes in Figure 3.3. The confusions in all but 5 of those 22 pairs are highly asymmetric (p < 0.001). Analysis of the hit rates and false positive rates suggests that this recognition behavior bears no relationship to cognitive bias. The frequency of occurrence of letters in English bears no relationship to the rates of correct responses, false positives, or total responses. Further, if the recognition behavior illustrated in Figure 3.3 was related to cognitive biases, the hit rates and false positive rates should be related but they are not.35

The responses of SA1 afferents to the same letters scanned across their receptive fields (Figure 3.3) seem to explain the recognition behavior. For example, B is rarely identified as B; instead, it is called D more often than it is called B. Conversely, D is virtually never called B (Figure 3.3 top). The reason for this response bias can be explained by the SA1 surround suppression mechanism discussed earlier, which suppresses the response to the central, horizontal bar of the B: the neural representation of the B does, in fact, resemble a D more than it does a B (see Figure 3.3). For another example, C is often called G or Q, but G and Q are almost never called C. An explanation is that many of the features that discriminate the letters are missing in the neural representations, so a lack of the features that distinguish a G or Q from a C in the representation of the C is not a strong reason to not respond G or Q. Conversely, the strong representation of the distinctive features of the G and Q make confusion with a C unlikely. The performance illustrated in Figure 3.3 is for naive subjects in their first testing session. Performance improves steadily on repeated testing.35 One explanation for this improvement is that subjects learn the idiosyncracies of the neural representations (e.g., when a subject recognizes the distinctive feature of the G in the neural representation he or she is less likely to mistake a C for a G).

The responses of typical human cutaneous afferents to Braille symbols (top row) scanned over their receptive fields are illustrated in Figure 3.4. The human SA1, RA, and PC responses to these raised-dot patterns are indistinguishable from the responses of monkey SA1, RA, and PC afferents to similar patterns.36 SA1 afferents provide a sharp, isomorphic representation of the Braille patterns, RA afferents provide a less sharply defined isomorphic representation, and PCs and SA2s provide no useful spatial information.

3.2.2.1.3 Studies with the Optacon

Psychophysical and neurophysiological studies with the Optacon provide a unique window on tactile perception in the absence of activity in the SA1 system. Neu-rophysiological studies show that the Optacon activates the RA and PC systems

0 0

Post a comment