texture perception via active touch

ELSEVIER Human Movement Science 13 (1994) 301-333

HUMAN ;;,;NT

Texture perception via active touch

Barry Hughes a,*, Gunnar Jansson b

a Dept. of Psychology, University of Auckland, Private Bag 92019, Auckland, New Zealand b Uppsala Vniversitet, Uppsala, Sweden

Abstract

The importance of movement in texture perception by touch has long been appreciated, but problematic issues related to the utilization of information during active touch persist. How is the repeated demonstration of active-passive equivalence in perceptual sensitivity to felt texture to be interpreted? What does such equivalence imply about the source(s) of information used to make perceptual judgments of texture? What sort of perceptual subsystem is organized to recover texture by touch? What is the role played by the action system in this recovery? The seminal observations of David Katz and J.J. Gibson are recounted and contemporary research is described. Several unresolved issues are outlined, as are several approaches by which the availability and utilization of information within this particular perceptual-action subsystem can be empirically studied.

“ . . . the hand is not only the organ of labor, it is also the product of labor.”

Friedrich Engels

1. Introduction

The prospects of making the information in maps, graphical and picto- rial representations accessible to the blind appear tenuous. Past attempts,

* Corresponding author. E-mail: [email protected]

0167-9457/94/$07.00 0 1994 Elsevier Science B.V. All rights reserved

SSDZ 0167-9457(94)00015-7

302 B. Hughes, G. Jansson /Human Mowment Science 13 (I 994) 301-333

especially in the case of embossed maps, have often used spatially segre- gated textures to represent spatial information but such attempts have met with only partial success. The literature directed to this topic reveals a number of possible factors of importance (Lederman, 1982, Lederman and Kinch, 1979; James, 1982; Jansson, 1983, 1988, 1991, 1992; Kennedy et al., 1991; Magee and Kennedy, 1980). In our own attempts to understand and account for this problem it has seemed critical to understand more about the mechanisms, both perceptual and motor, that underlie texture perception by touch. We began by presuming that the design and production of displays that are effective for the visually impaired will emerge in part from a better understanding of the capacities and limitations of active touch in the recovery of spatial information in textured surfaces.

The modest utility of textured maps for the blind is surprising insofar as texture perception via active touch is a notably accurate naive skill. It is a commonplace occurrence that judgments of texture are made very accurately by touching; indeed such judgments may be more accurate via touch than via sight. (Is this material silk or rayon? Is this surface perfectly smooth?) Compared with the fidelity with which texture perception by touch can operate, it is puzzling that it has not proved to be a more effective mode by which other sorts of spatial information can be picked up. This puzzle has led to the identification of several lacunae in accounts of texture perception, gaps which warrant experimental investigation. In what follows we attempt to describe some of these gaps in our understanding. We especially emphasize the nature of the perceptual-motor interactions during texture perception and seek to define the outstanding issues in terms of the availability and transaction of information within the haptic system.

Some preliminary definitions of the sorts of events with which we will be concerned seem in order. Further distinctions will be discussed below. It has been recognized at least since Weber’s (1834, 1842/1978) work on two-point tactile discrimination that relative motion between a distal object and the skin surface is an advantage in the recovery of the spatial characteristics of tangible structures. Relative motion between the distal surface and the sensory surface can arise in several ways. Either the skin or the object can move relative to the stationary other, or both can move. Passive relative motion can be generated in two ways: either the surface can be moved across a stationary skin surface or the skin can be moved by an agent other than the percipient. In what follows, “movement” refers to active (not passive) movements by an actor and the “motion” refers to

B. Hughes, G. Jansson /Human Movement Science 13 (1994) 301-333 303

movement of an object, including (parts of) the actor under passive conditions. The generic term “relative motion” will be used when no distinction between active and passive versions is explicitly intended. Information in the stimulation to which relative motion gives rise, is apparently necessary for the perception of a wide variety of textural qualities, from smoothness to braille character identity.

Demonstrations of passive-active equivalence are interesting for at least two reasons. First, they contrast with the quite different phenomenological consequences of the two conditions. In essence, active touch leads to unequivocally distal attributions (i.e., to the perception of some object “out there”) even if the object itself is not readily identifiable (Gibson, 1962). Katz and others have referred to this as the objective pole of perceptual experience. In contrast, passive touch favors proximal attributions (the so-called subjective pole) but can also provoke distal attributions. Second, active-passive equivalence in texture perception implies that the perceptual system responsible for such perception requires no further information than that contained in the flux of stimulation arising at the proximal cutaneous surface, even if it is available (Loomis and Lederman, 1986). That is, any information that is generated by changes in the tensile states of the musculature or by the joint receptors during active movement exploration is not effective in the specification of texture (i.e., it does not alter perceptual sensitivity). Despite the suggestion that during active touch, movements can actually serve a perceptual function (Gibson, 19621, findings of active-passive equivalence suggest that information regarding the movements is unnecessary for perception (even if those movements are responsible for the creation of the flux of stimulation in the first place). We attempt to address this puzzle and to evaluate its consequences. We conclude that despite the widespread evidence of equivalence, the pursuit of the role of movement in the perception of texture is justified. In doing so, ancillary issues emerge and are addressed.

Although we constrain our discussion to texture perception by touch, the issues have ramifications for larger questions of theories of perception and performance, perception-action coupling, and dynamical and computational accounts of perceptual-motor organization. It has been recognized since Turvey’s (1977) provocative challenge that respect for the tight coordination between information for perception and information for action is necessary to theories of complex performance. Debate concerning the nature of information in interactive perception-action systems has never obscured the consensus that there is some sort of interaction;

304 B. Hughes, G. Jansson /Human Movement Science 13 (1994) 301-333

disagreements have arisen regarding the candidates for the means by which it is realized (e.g., whether intermediary representational systems are required), and these remain conceptually and theoretically divided (for different recent treatments, see, e.g., Cruse et al., 1990; Epstein, 1986; Gandevia and Burke, 1992; Griisser, 1986; Prinz, 1990; Scheerer, 1984; Schmidt, 1988; Viviani and Stucchi, 1992). The significance of this relationship has long been recognized in the domain of active touch - Rev&z (1950) coined the term “haptics” to emphasize it - although it has not attracted the degree of experimental and theoretical attention that visual control of action, for example, has drawn.

In recent years, however, active touch in its various manifestations has been reasserted as a perceptual-motor skill of exquisite precision and complexity. A wide variety of approaches to its study are now described in the literature. Turvey and his colleagues, for example, have developed conceptual schemes and amassed an impressive body of experimental data on aspects of dynamic perception using hand-wielded objects (see, e.g., Burton, 1993; Pagan0 et al., 1993; Solomon, 1988; Solomon and Turvey, 1988). Also emerging are more complete physiological models of the contribution of different mechanoreceptor populations to the recovery of texture. Johnson and his colleagues (e.g., Connor and Johnson, 1992; Johnson, 1983; Johnson and Hsiao, 1992; Phillips and Johnson, 1981a, b) have modelled both the mechanics of the skin under a variety of textural distortions and the spatial and temporal aspects of cutaneous discharge during texture perception (see also Loomis and Lederman, 1986; Franzen and Westman, 1991; Sherrick and Cholewiak, 1986). Drucker (1988) has demonstrated why relative motion computationally simplifies the recovery of felt texture. Johansson and Westling (1984, 1987, 1990) have offered a detailed account of how and on what basis the finger contact of objects can be modulated during active grip. And there is an ever-growing literature on computational and procedural approaches to representation, control and design in human and robotic grasp, areas which have an increasing commit- ment to ecologically valid behaviors (e.g., Goodale, 1989; Jeannerod, 1986; Klatzky et al., 1985; Klatzky et al., 1993; Klatzky and Lederman, 1990; Lederman and Klatzky, 1987, 1990; Venkataraman and Iberall, 1990).

2. Movement, motion and texture

David Katz, the phenomenologist, and J.J. Gibson, the perceptual theo- rist, each maintained a common if methodologically distinct interest in the


mechanisms of active touch. Katz pioneered the modern study of the various perceptual capacities of touch and Gibson used the various perceptual subsystems of touch to clarify his distinction between sensory and perceptual systems. While neither has had the last word on these matters, this section recounts some of their important observations and claims regarding the issues under consideration here, and identifies areas where their insights require extension. As such, it is not intended to be an exhaustive treatment of their work in the area of touch generally - Krueger (19821, for example, offers more comprehensive coverage - but of their work as it bears on texture perception.

2.1. Katz on microstructure and movement

David Katz’s study of touch culminated in his Der Aufbau der Tastwelt in 1925 (edited and translated into English by Krueger as The World of Touch, 1989; see also Krueger, 1970, 1982). Katz explored a wide variety of touch phenomena, from the perception of vibration to the localization of distal touch sources. As Krueger (Katz, 1925/1989) reminds us, though, “above all, he was concerned with texture or microstructure, that is the fine structure of the surface, rather than its shape or macrostructure” (1989, p. 8; emphasis in original) and it is Katz’s account of texture perception that is of primary interest to us here.

An important source of data already available to Katz was that of Ernst Weber, who (1834/1978, in De Tactu) offered one of the earliest references to “the great improvement of tactile acuity produced by movement of the touch-organs” (p. 51). Weber’s research on two-point thresholds showed that they diminished when relative motion was introduced and even more when active movement was possible. Weber explained the benefit of active over passive touch by arguing that in the former case the percipient has access to both perceptual (temporal onsets and spatial locations of the points) and motor information (the velocity of the hand) and, on this basis, computes the distal structure: “we notice the time that elapses between the touch of the first and the second points on the same part of the finger; and we are aware of the speed with which we move the finger; and we then measure the space between the points from the time passing between the two pricks” (p. 52).

Katz’s work on texture perception by active touch comprised two broad lines of inquiry. One was a phenomenological and theoretical consideration of the nature of touch; the other was a series of experimental investigations into the haptic discriminability of various common materials. While aware


of the work that had been done on touch, Katz (1925/1989) nonetheless considered it conceptually flawed. He criticized those approaches (including those of Weber, Helmholtz, and Katz’s mentor Miiller) which revealed a bias toward “temporal atomism” - that is, which treated a tactile perceptual event as the sum or composite of static tactile samples, as a tactile version of tachistiscopic perception. Katz intended that his work would demonstrate that processes occurring over time result in perceptual outcomes that “by no means can be conceived of as the sum of those experiences which arise in connection with motionless tactual stimuli” (Katz, 1925/1989, p. 77). Indeed, Katz argued that haptic perception was, in large part, event perception and hence required an analysis in the other direction: rather than treating and seeking to explain tactile percepts as the summation of separate static impressions, he sought to understand how the perceiver can go from the spatiotemporally dynamic proximal stimulation back to the spatially structured distal object: i.e., how “the cinematic form of the stimulus is converted to the static properties of an object” (ibid, p. 79). This persists as perhaps the most basic issue in texture perception by touch.

Despite his faith in the primacy of movement in touch, Katz was careful to distinguish instances where movement made a difference to perception from those where it did not. For example, he conceded that in judgments of hardness or softness movement played a far less critical role than in judgments of texture or elasticity. He also noted the conditions under which perceptual equivalences emerged; how, for example, judgments of hardness did not alter under sizable differences in movement speeds. When he conducted experiments on subjects’ abilities to qualitatively describe pieces of paper varying in their smoothness (Katz, 1925/1989, pp. 101-1081, he found that the judgments were almost identical under active and passive conditions, although it took actively moving hands appreciably longer than the l-2 s passively felt surfaces required for judgments, and even though subjects were able to use all five fingers rather than three or four in the passive condition. He also noted that each subject had a particular movement speed that produced the best results and that there existed a range of speeds (3-60 cm/s> beyond which subjects’ performance deteriorated. Katz did not, however, describe the relationship between the subject’s optimum speed and his or her preferred speed; that is, whether or how well a subject knew the velocity at which to move in order to recover sufficient information to render an accurate judgment. This, too, remains a potentially important and testable question.

B. Hughes, G. Jansson /Human Movement Science 13 (I 994) 301-333 307

Katz demonstrated the extent to which the proximal and the distal stimuli are distinguished in touch. He argued that the hand - and not just the skin surface, nor the cutaneous receptors, nor the muscle and joint receptors - represented the “organ” of touch. Katz’s insistence that the functioning hand could not be compartmentalized into its distinct sensory and motor roles without losing the essence of active touch stemmed from his claim that the “tactile sensitivity” of the hand is dependent not only on motion but on the active intentional movement that it was capable of producing. It is active movement, which he termed “a creative force” in this type of perception, that made the hand the organ of touch. In this respect, Katz anticipated Gibson’s conceptions of active touch and Rev&z’s (1950) description of movement giving touch a “creative and formative energy” (pp. 61-62).

2.2. Gibson on touch as a perceptual system

Gibson’s contributions to perceptual theory are well known. Gibson’s contributions were primarily to ecological optics; it may be useful to review some of what Gibson had to say regarding the world of touch. Like Katz, Gibson observed that a variety of perceptual puzzles that were typically associated with vision (such as the ability to discriminate object motion from self-movement and the perception of distal objects rather than proximal stimulation) had analogs in touch. Gibson held that theories of sensory channels, which stressed the differences between senses, would be inadequate to account for such analogous phenomena. Indeed, Gibson had serious problems with all theories of perception in which the sense organs are mere producers of sensations.

For the haptic system, as for the visual system, Gibson (1966) argued that the standard anatomical/ physiological partition of the perceptual world had been misguided, that sensory modalities could not be identified with perceptual systems. He insisted that, far from being passive relayers or transducers of stimulation, the senses are part of, indeed that they are cross-cut by, perceptual systems whose function is to actively seek information. “Active touch does not fulfil the supposed criteria for a single sense modality. Nevertheless it provides a quite definite channel of information about the external environment” (Gibson, 1962, p. 479).

Hence, rather than accept that the specificity of the different sorts of mechanoreceptors was functional (that, e.g., cutaneous receptors, muscle spindles, and joint receptors serve unique perceptual functions), Gibson

308 B. Hughes, G. Jansson /Human Movement Science 13 (I 994) 301-333

argued that “[i]n perception, nerves function vicariously” (Gibson, 1966, p. 109) and that an alternative conception was possible: perceptual subsytems specifically organized across these receptor types. According to this alternative, one could conceive of different subsystems comprising perceptual organs that were, in turn, composed of different, special-purpose combinations of mechanoreceptors. Multiple, modal stimulations would be sup- planted by singular, amodal information (Epstein, 198.5).

Among the perceptual subsystems that Gibson (1966) suggested were three that could be organized for texture perception: cutaneous touch, haptic touch and dynamic touch. Cutaneous touch is that which occurs in the absence of movement of the proximate joints or change of state of the musculature, and information would be contained in cutaneous mechanoreceptor activity only. By haptic touch, Gibson meant the spatial information contained in stimulation from both cutaneous and joint mechanoreceptors. By dynamic touch, Gibson (1966) meant the combination of spatial information extracted during haptic perception with nonspatial information from “muscular exertion” (p. 109) into a single perceptual (sub)system (see also Pagan0 et al., 1993).

Gibson (1962, 1982a, b) stressed that active touch is exploratory rather than merely receptive. In addition, movement and the spatiotemporal stimulation, Gibson (1962) insisted, are lawfully related: “variations in skin stimulation are caused by variation in . . . motor activity” (p. 477) and, therefore, specific changes in the movement velocity, for example, will instantaneously produce equally specific changes in the flux of stimulation. This is the information that the haptic system extracts, and movement, for Gibson, generated a unique source of information for the actor (see Epstein et al., 1986, for an unconventional active touch example). Gibson insisted on a related point. He argued that in the recovery of object structure by touch the movements themselves are exploratory or information-seeking and not just performatory. In this regard, Gibson (1962) made a telling series of observations on the kind of exploratory movements that are made during unsighted touch:

“the observer seems to be trying to obtain mechanical events at the skin at various places in various

combinations. The movements by which he obtains them do not ever seem to be twice the same, but

they are not aimless. If the hand is conceived of as a sense organ, he seems to be adjusting it. He

appears to be searching for information.” (p. 481)

This observation would appear to suggest that not only is movement the cause of perception, but it is among the contents of perception because it

B. Hughes, G. Jansson /Human Mor)ement Science 13 (1994) 301-333 309

produces variations in hand location, movement velocity and kinematic variables as well as changes in nonspatial, mechanical variables of the musculature, all of which contribute to the flow of information. This may be seen as an inversion of standard treatments according to which the higher-order variables in stimulation need to be “taken into account” during perception (see Epstein, 1973, for its application to vision) and where standardization of the variables simplifies or makes for more effi- cient perception. Gibson was quite explicit in his contention that during active touch the increased variation in the movement patterns changes most what is felt at the fingerpad and that, far from giving rise to confusion or ambiguity as to the structure of the explored surface, such variation actually contributes to the extraction of invariants (such as spatial structure): “a clear unchanging perception arises when the flow of sense impressions changes most” (Gibson, 1962, pp. 487-488). The familiar Gibsonian concept of the invariant can be discerned here. The information in stimulation did not have to be (indeed, was unlikely to be) contained in the simple(r) output of a single mechanoreceptor type. Instead, information was contained within the complex flux of transformations of multiple receptor types over time: “There must exist relations of separation and proportion [in stimulation] which remain invariant over time and which are specific to the object. They do not represent the shape of the object but they specify it” (ibid, p. 488). Movements have a specificational role in this regard, according to Gibson (e.g. 1982al: the active control of manual or digital exploration makes possible the isolation of an invariant exterospecific component (i.e., information specifying external objects and events) from the propriospecific component (i.e., information specifying self-movement) and it is this isolation that ensures the specification of distal events or objects.

3. Contemporary investigations of texture perception by touch

The emphasis that both Katz and Gibson placed on (al movement during active touch and on (b) the complex stimulation with higher-order, informational invariants contained within it, has not been lost on modern researchers. However, several of the larger issues to which they addressed themselves have not been pursued. We offer a brief overview of more contemporary extensions of research that Katz and Gibson may fairly be said to have motivated, concentrating on those research domains and

310 B. Hughes, G. Jansson /Human Mocement Science I3 (I 994) 301-333

attendant effects that bear most closely on our primary focus, perceptual- motor interactions in texture by touching. Loomis and Lederman (1986) provide an extensive overview of these and various other research avenues in tactile perception.

3.1. Perception of textural roughness

Roughness/smoothness is the primary perceptual dimension of textured surfaces and has been the subject of substantial experimental study, begin- ning with Katz (192.5/1989; see also Meenes and Zigler, 1923). Modern focus on textural roughness has primarily been on the roles of active and passive movements in its perception and on its mechanoreceptive basis. Together, these lines of investigation suggest that sufficient for a complete account of texture perception by touch are the spatiotemporal events occurring at the interface of object surface and skin surface - the events that arise via any sort of relative motion - and their transmission by cutaneous afferents. We examine this popular view and critically assess its strength and weaknesses.

Among contemporary investigations, Susan Lederman’s work has been seminal in the area of perception of surface roughness (see, e.g., Leder- man, 1974, 1981, 1982, 1983; Lederman and Abbott, 1981; Taylor and Lederman, 1975; Taylor et al., 1973). It is this work that has suggested most clearly that, phenomenological differences notwithstanding, the perceptual discriminability or scaling of roughness is equivalent under active and passive conditions. Using machined surfaces comprising grooves and ridges, each of various widths, Lederman (1981) found that roughness estimates are a function of the structure of the surface being felt: more particularly, estimates increase with increases in spatial period, and even more particularly they increase with the width of the grooves but not the ridges passed over by the fingers. Increased finger force applied at the surfaces was also found to produce larger estimates of roughness. However, less than 1% of estimate variance was accounted for by the mode of scanning. Active touch and passive touch lead to equivalent estimates. Moreover, large increases in the relative velocity between surface and fingerpad (i.e., from 10 to 250 mm/s) did little to influence roughness estimates in either mode of perception (Lederman, 1974, 1983).

Lamb (1983) also failed to find any evidence of differences between active and passive touch in the textural discriminability of raised dot surfaces. While the passive condition was entailed by having each subject’s


fingerpad in contact with a textured drum rotating at a constant velocity in a single direction, active exploration was observed to take a variety of general patterns and kinematic profiles. Despite the large differences in the spatiotemporal events created at the skin surface by these two conditions, perceptual sensitivity was almost identical. Heller (1989) also found no evidence of active-passive differences in judgments of textural smoothness.

Recognizing that different velocity profiles in the two conditions might confound matters, Lederman (1981, 1983) sought to equate the proximal events by (1) using sinusoidal passive movements (i.e., the surface was moved sinusoidally under a restrained fingerpad) and then (2) using an external clock to regulate mean velocities of the active movements in an equivalent manner (Lederman, personal communication, 1990). Even in this case, no evidence was found for a superiority of one mode over the other, strengthening claims of the negligibility of any information derived from movement (as opposed to that derivable from mere relative motion) in the recovery of roughness (Loomis and Lederman, 1986).

Psychophysical research on texture by touch (and pattern perception, for that matter) reveals very few instances in which differences between the perceptual consequences of active and passive movement differ (but see, e.g., Magee and Kennedy, 1980). This suggests that while texture or roughness discriminability is motion-dependent, it is not movement-dependent. This conclusion has been endorsed by detailed investigations of the various cutaneous receptor discharge patterns under covaried conditions of surface structure and motion. Goodwin and Morley (1987a), for example, noted that “apparently, the only requirement is movement between the grating and the skin, and the discharge of the cutaneous mechanoreceptor populations contains sufficient information for the subject to extract details about the spatial characteristics of the surface that are independent of the movement characteristics” (p. 2168).

3.2. Cutaneous contributions to texture perception

Katz and his contemporaries, particularly von Frey, had disputes about whether the perception of texture and of vibration were achieved by the same sense or whether there was a separate sense for each (see Katz, 1925/1989; Krueger, 1982). Contemporary work has done much to account for the variety of sensations (pressure, vibration, etc.) to which touch gives rise, and more importantly has begun to identify the degree to which the cutaneous receptors serve complementary roles during texture perception.


Roughness estimates reveal different equivalences under several disparate physical conditions: in addition to those of active and passive conditions and with various relative motion velocities, Taylor and Leder- man (1975) reported also that estimates of roughness are not affected by lubrication of the surfaces, a finding that has suggested (a) that roughness is determined by spatial structure and not by the coefficient of friction, and (b) that those mechanoreceptors sensitive to vibration, friction’s cutaneous manifestation, are not responsible for such estimates, at least with the large textures they used. Roughness can also be estimated when the surfaces are embossed with large elements the size of braille dots (i.e., 0.4-0.5 mm) and when the elements are microscopic etchings at the scale of microns (e.g., LaMotte and Srinivasan, 1991). It is widely agreed that quite different mechanoreceptor afferents are involved in these two instances, probably slowly adapting (or SA) afferents in the former but rapidly adapting (RA) afferents in the latter. Further, as Katz (1925/1989) demonstrated, roughness perception is possible when the perceiver uses a rigid object such as a stick to explore a surface. In such cases, vibration is the relevant energy form and Pacinian afferents are the active transducers of vibration. In other words, roughness does not seem to be tied to a single cutaneous receptor system, implying that the information for roughness is higher-order than the cutaneous discharge of any group of receptors. It is not clear among physiologists that a coding scheme that accounts for velocity-independent coding, for example, in one receptor population will necessarily apply to another (Connor et al., 1990; Johnson and Hsiao, 19921.

Moreover, Connor et al. (1990) have claimed that, contrary to previous assessments, roughness is not a monotonic function of spatial period. Using raised dot displays within which spatial period could be varied in two dimensions rather than one (as, e.g., in Lederman’s experiments), they found that perceived roughness followed an inverted-U function: estimates were maximal at a spatial period of approximately 3 mm and declined toward both smaller (1.3 mm) and larger periods (6.2 mm). This inverted-U function was observed with dots of varying height (0.5-1.2 mm), with higher dots led to larger estimates of roughness. The significance of this result lies, again, in the fact that identical estimates of roughness can emerge for surfaces that provoke quite different neural responses.

Goodwin and Morley (Goodwin and Morley, 1987a, b; Morley and Goodwin, 1987) conducted systematic parametric investigations (involving covariations in spatial period, peak velocity and spatiotemporal frequency) of the responses of three different cutaneous receptor populations - slowly


adapting (SA), rapidly adapting (RA), and Pacinian (PC> - in the macaque monkey to sinusoidal passive motion of grooved textures. They found, inter alia, that for any given period (1.5 mm, 2.0 mm, or 3.0 mm>, the responses of the SA population were invariant over large differences in peak velocity (30 mm/s to 240 mm/s), but systematically increased as the period increased. In general, when one parameter is fixed, the output of the SA population was a function of the spatial period of the surface irrespective of the spatiotemporal change that may be superimposed on it as a consequence of relative motion between surface and fingerpad (see also Phillips and Johnson, 1981a). Morley and Goodwin (1987) also found evidence of phase-locking of the receptor populations to grating ridges during the central portion of a complete scanning cycle. Since the velocities of the texture gratings were always according to a sinusoidal profile, they peaked at 90” and 270” of the cycle. Within 42” of these locations, peak velocity of the grating was within 6.6% of its maximum and it was here that phase- locking was observed. Darian-Smith and Oke (1980) had earlier demonstrated the phase-locking of afferent discharge to the temporal frequency of texture elements as they crossed the fingerpad at constant velocities such that the different receptor types were differentially effective at different ranges within the dynamic range. Phase-locking, Morley and Goodwin (1987) suggested, contributed additional information in the form of unique discharge patterns that could be useful in situations where different textures can give rise to the same mean discharge rate. However, it has since been shown by Connor and Johnson (1992) and Sathian et al. (1989) that, contrary to other sensations whose magnitude can be estimated, perceived roughness does not appear to be based on the mean firing rate of the primary afferents or any simple combination of rates of discharge across afferents, but on higher-order patterns of discharge. Connor et al. (1990) found that trying to tie an intensive coding scheme (e.g., mean afferent firing rates) to an intensive perceptual dimension (roughness) was unsuc- cessful: mean SA firing rate and perceived roughness, for example, were poorly correlated. Connor and Johnson (1992; see also Connor et al., 1990) correlated psychophysical data on roughness estimates in humans using active sinusoidal sweeping movements with the discharge patterns of SA and RA afferents in monkeys under passive, constant velocity conditions. No simple coding mechanism of mean discharge rate seemed to relate spatial structure of their displays with perceived roughness. Instead, a more complex hypothesis was suggested, that “roughness perception is propor- tional to the mean firing rate in a population of central receptors sensitive


to spatial variation in the activity of SA afferent fibers” (Cormor and Johnson, 1992, p. 3425; emphasis added).

3.3. ” Texplora tion ” in other domains

The perception of texture is, of course, not restricted to the estimation of roughness. We briefly describe three other areas in which texture by touch has been examined. They are bimodal (visual-tactual) texture perception, braille reading and the perception of texture gradients. Collectively and separately they reveal effects that are either unexpected given the downplayed role of movement in received accounts of texture perception or are beyond the narrower focus of active and passive touch. From the literature on these topics, we highlight issues that are specifically related to the role of information extraction from texture and that may inform future experimentation.

3.3.1. Bimodal perception The nature of the interactive contributions of vision and touch in

perception has also been the subject of substantial research (for a recent review, see Warren and Rossano, 1991). The emerging consensus appears to be that there is no fixed or even general dominance relationship between the two perceptual modalities, such that vision contributes more or less to perceptual judgments of texture in general or roughness in particular. While dominance can be demonstrated to exist in a given task, such as texture perception, or implied to exist by greater accuracy, briefer decision latencies, or other indices of perceptual efficiency (e.g., Heller, 1982; Jones and O’Neil, 1985), it can also be reversed by task conditions or demands (e.g., Klatzky et al., 1987; Lederman et al., 1986).

The search for a general perceptual hierarchy in texture perception has given way to more detailed consideration of the specific informational requirements of certain tasks and how different perceptual systems may contribute (or fail to contribute) effective information to the perceptual judgments required. It seems likely that future experimental studies of bimodal texture perception will be - or ought to be (Warren and Rossano, 1991) - concerned as much as possible with the specific issues of informational sources and content. The relevance of this topic can be illustrated by considering a particular experimental finding and its possible implication(s) for the nature of information in active touch.

B. Hughes, G. Jansson /Human Mocement Science 13 (I 994) 30X-333 315

Heller (1982) required subjects to identify the “smoothest” of three samples of sandpaper by either visual or haptic inspection, or by the simultaneous use of both perceptual systems. He found in one experiment that performance in the bimodal condition was equivalent to haptic perception, and that the former was superior to that in the visual condition. In a second experiment, he found perceptual accuracy in the bimodal condition to be greater than either unimodal condition alone, which in turn were not different. In all cases correct response probabilities were far higher than chance.

Hence, it could be concluded that texture perception by active touch is improved when visual information is concurrently available, but this raises the question of what information vision provides in such situations, a question made more mysterious because, as Heller (1982) notes, there is a tendency in such cases for subjects to periodically avert the eyes from the surfaces while exploring (see also Heller, 1993). To determine whether the information that the visual system provides under bimodal conditions is (or is not) information about texture, Heller contrived a situation in which the subjects could see everything but the textured surface: the display box, their hands, and their movements. Would this version of the bimodal condition sustain perceptual accuracy at previously high levels, or would the subject’s performance approach the lower level of haptic exploration alone?

The data revealed that the beneficial informational contribution of vision exists even when information did not include the textured surface itself. Perceptual access to the exploring hand, it would appear, was sufficient to maintain the bimodal superiority, or as Heller (1982) concluded: “the elimination of visual texture does not impair performance for bimodal conditions when visual regulation is still provided” (p. 343). What, precisely, is the nature of this “visual regulation”? Is information from the visual system regulating the movements by being information that is otherwise unavailable, e.g., about the spatial constraints (distance to an edge) within which the hand is working ? Is it regulating the movements by offering a visual target for the hand movements and thereby permitting the exploratory movements to be “programmed” in a manner that is qualitatively different and quantitatively more effective than when engaged in nonvisually guided scanning? Is vision providing information about what is “out there” or what the hand is doing there?

Given the claim that active touch creates a flux of stimulation that has two components, the exterospecific and the propriospecific (Gibson, 1962,

316 B. Hughes, G. Jansson /Human Mocement Science 13 (1994) 301-333

1982a), it is not clear which component visual information provides in cases of bimodal superiority. Nor has it been determined, more generally, how information specifying the same textured surfaces but extracted in two perceptual systems is utilized.

3.3.2. Reading braille Under most circumstances, unlike roughness perception, braille reading

efficiency is greater via active than passive touch (e.g., Foulke, 1991; Grunwald, 1966; Heller, 1985). Foulke (1991) has attributed this superiority to active touch’s “supply of more neural data via the excitation of muscle, ligament, and joint receptors [and to the control of] the higher mental processes that guide the search for the information encoded in braille” (p. 224). That is, individual differences in braille reading speed are attributable to the efficiency of “higher” sensory, memorial, and linguistic capacities. However, it has become apparent that perceptual-motor coordination is also critical.

There is now evidence that the fastest braille readers use both hands to read and, more significantly, use both hands not in a “conjoint” fashion where the index fingers are placed close together and scan the text together. Instead, the hands are coordinated in a “disjoint” manner: the left index finger scans from the left to some point near the middle of a line at which point it is joined by the right hand and conjoint reading occurs, and then the right finger completes the line. As it does so, the left finger is moved to the next line, and begins scanning it as well (Bertelson et al., 1985; Mousty and Bertelson, 1985). Hence, the best braille readers can be found to be deploying a complex form of perceptual-motor coordination: scanning two adjacent passages of text on different lines simultaneously and covering the same text successively during the rest of the time. Indeed, the faster readers tend to minimize the conjoint phase in favor of disjoint coordination. Not only do both hands participate in the simultaneous pick-up of text information, it is possible for them to do this with different passages of text. (It is apparently not the case that during disjoint reading the fingers are out of phase with respect to braille cells; a large portion of this scanning entails both fingers being over braille characters simultaneously; cf. Bertelson and Mousty, 1989; Millar, 1987). Bertelson et al. (1985) also reported that scanning by a wide variety of readers was “generally continuous, with little variation in speed” (p. 238) and their kinematic analyses support this description. The velocities were not perfectly con-


stant, however. During stoppages, return movements and regressions, large rates of change in velocity were also be observed.

Braille reading, too, raises questions about the nature of perceptual-action relations. To what extent are these relations unique to braille? Does the uniformity of the spatial relations among textured characters constrain active touch in ways that are distinct from texture perception of other types? To what degree (and how) is the information extracted during reading used to modulate the action system (to alter or stabilize the speed of the movements, for example) and to what degree (and how) is it modulated by it? Are the spatiotemporal patterns of stimulation generated when reading used to control hand speed (and if so under what circumstances)? Does hand speed determine pick-up of the information in the braille cells (and if so under what circumstances)? What is the status of the observed variations in velocity during reading: are they a source of error that, given some threshold, the perceptual system needs to (but perhaps does not) overcome?

The complex pattern of coordination evidenced by expert braille readers also begs for analysis in terms of the perception-action relations. How does such coordination emerge: via natural, individually idiosyncratic bimanual organization or via something more akin to a learned reorganization? Is the apparent benefit of bimanual movement during reading due only to the increase in information that it makes available or does the self-organizing character of bimanual movements (e.g., Beek et al., 1992, Kay et al., 1987, Kelso and Schiiner, 1988) also afford its own benefits, by minimizing interlimb velocity differences or stabilizing interlimb phase differences, for example?

3.3.3. The perception of texture gradients With the exception of braille textures, experimental textures have typi-

cally been precisely uniform, such as etched grooves, or stochastically distributed, as in sandpaper. To our knowledge, only one other study of texture gradients has been reported: Connor et al.‘s (1990) use of linear texture gradients (embossed dots) in order to assess roughness perception. As such, we have considered our investigations to be preliminary. Even so, they are suggestive of more complexity in texture perception than ho- mogenuous textures might lead one to expect (a complete report of which is forthcoming; Jansson et al., 1994).

A surface that is slanted in depth produces an optical texture gradient:

318 B. Hughes, G. Jansson /Human Mocemenr Science 13 (1994) 301-333

The spatial density of textural elements is progressively greater and their shape compressed in the direction of the slant and to a degree commensu- rate with the slant. Raised (approximately 0.4 mm) texture element depic- tions of these surfaces (without shape compression but in both dot and gridline formats) have been presented to blindfolded sighted subjects. In one set of experiments, subjects were asked to determine the direction of the gradient and then to either estimate the magnitude of the gradient, or to estimate the corresponding depicted slant. Each of these tasks can be done reasonably accurately, although (a) estimates of texture gradients are slightly more accurate than (translated) optical slants, (b) subjects do not seem capable of reliably discriminating gradients that differ by (the equivalent of) slants of approximately 5-10” (Jansson and Hughes, 1991a, b) and (c) they benefit greatly from prior verbal descriptions coupled with haptic exposure to the range of surfaces (e.g., Holmes, 1993). When we (Jansson and Hughes, 1990) used the photographic versions of our textures and had subjects use an Optacon to scan virtually textured surfaces, ’ estimates of gradient magnitude were only slightly less accurate (the exponents of best-fitting functions were flatter), this despite the probability that the mechanoreceptors activated by the vibrating pins of the Optacon array were likely to have been different from those mechanoreceptors activated by embossed textures (see above).

Subjects in these experiments were not constrained in their movements except to the use of one fingerpad at a time. The exploratory movements, recorded on videotape, revealed two different exploratory phases, according to the task: one when the direction of the gradient was being sought and another when its magnitude or slant was to be estimated. The explorations seemed to fall into two basic types for each phase. The recovery of the gradient direction typically entailed either slow movements around the perimeter of the surface or rapid, cyclical sweeps across the lateral extent of the surface (with mean velocities in the range of 6-10 cm/s). Attempts to recover the slant magnitude also included these cyclical sweeps, but also revealed other exploratory patterns. One common pattern entailed compar- ing textural density at the surface edges by rapidly scanning at one end,

’ The Optacon-II (Telesensory Systems, Inc) is an optical-to-vibrotactile transduction system. 2D

(5 x 20 cell) optical images are picked up by a small hand-held camera and projected to an array upon

which a contralateral fingerpad is placed. The pins of the array vibrate at 235 Hz. As used here, it produced the same textural surface but indirectly in the form of structured patterns of vibrotactile

stimulation rather than directly as tangible dots or lines.

B. Hughes, G. Jansson /Human Mocement Science 13 (1994) 301-333 319

lifting the finger to other edge and scanning rapidly there, followed (presumably) by an estimation of the difference and therefore magnitude of the intervening gradient. A subsequent comparison of surfaces whose gradients were occluded was conducted: one surface type had its margins inaccessible to touch but the interior available; the other had its interior obscured but its margins available for sequential comparison. Perceptual performance (gradient magnitude estimation) was superior in the latter case.

These observations suggest that the sorts of exploration undertaken depend strongly on the information sought, that subjects will organize movement patterns that offer the best prospects for such recovery, and that some subjects will (seemingly) resort to abrupt, sequential comparisons of textures rather than attempt to extract gradients from direct and continuous exploration. Gibson (1966) characterized this later sort of activity as “semi-intellectual” (p. 12.5) rather than perceptual but it is, nonetheless, indicative that subjects find the task simpler by moving in certain ways and that continuous contact, even while in motion, is not always a preferred exploratory strategy.

We have also conducted experiments using texture gradients of other forms. In one experiment we presented two adjacent gradients (each 5 cm wide) representing flat grids slanted in depth. The interior, adjacent edges of the panels were depicted as either both farther away or both closer (so the surfaces were quasi-symmetrical about the surface midpoint). The subjects’ task was to explore the surfaces using cyclical lateral sweeps, at any speed they chose and for no longer than 10 s. They were then to report which gradient, the left or the right, was larger. One or both surfaces were slanted and the critical independent variable was the difference in gradient between the two. No shape information available that could be used by the subjects to estimate slant or gradient was available. Differences could range between 0” and 40” (for purposes of exposition, we refer to the gradient in terms of depicted angular slant in depth). This task proved extremely difficult: only the maximum slant differences were reliably discriminated. Subjects’ performance revealed nothing of the renowned fidelity of touch; nor indeed did it conform with the patterns of cutaneous discharge that Connor et al. (1990) have been able to measure with their gradients.

A second experiment used a different sort of gradient. Subjects were asked to explore (in the same manner as above) a textured grid surface that, with equal probability, either did or did not contain a density


discontinuity (an abrupt step-change in spatial period). The discontinuity took up 25% of a 10 cm-wide surface and its magnitude and locations within the surface (and hence also within the scanning cycle) were varied. In this task, subjects proved remarkably capable. Estimates of perceptual sensitivity (8) revealed that only the smallest discontinuity could not be reliably detected, subjects rarely reported a discontinuity in its absence, and sensitivity was not affected by where on the surface, or within the scanning cycle, the discontinuity appeared.

These experiments highlight both the ease and difficulty that can be associated with texture perception by touch. As such, they provoke more questions than they answer. We have, however, tentative evidence that abrupt changes in texture are more easily recovered than are smooth gradients, a prospect we are currently pursuing explicitly. A large number of questions regarding the recovery of gradients remain to be addressed. We describe possible structural variants below but note here that texture gradients can also be multidimensional. That is, variations in the spatial period of texture elements can be supplemented and covaried with other microstructural gradations: in element size, height, and shape, for example.

4. Issues and lacunae

We have reviewed the primary antecedents to contemporary research in texture perception by touch and summarized some relevant data from this later period. In this section we concentrate on several problematic issues that have emerged from this review and that appear in need of evaluation.

The first is to ask why active touch does not give more frequent evidence of being a more accurate (or at least a differentially accurate) perceptual subsystem than its passive analog. How justifiable is the emerging consensus of the negligible role of movement information in texture perception? Given Gibson’s categorization of the perceptual (sub)systems of touch, a second issue emerges, namely, how exploratory touch is to be characterized. This boils down to the need to determine the precise influence of movement information to perception, and on perception. The answer to this question has implications for the contents of theoretical accounts and models of texture perception. We next consider the issue of perceptual constancy in texture perception and how this might be examined by using covariations in spatial structure of surfaces and constraints on the exploratory patterns used in their recovery.


4.1. Active touch and passive touch: When (and why) are they perceptually equivalent?

There are several bones of contention surrounding the contrast of active and passive scanning and the interpretation of the data that emerge from such comparisons. As noted, no detectable differences between active and passive perceptual conditions have been found in roughness/ smoothness estimation (e.g., Heller, 1989; Lederman, 19811, texture discrimination (e.g., Lamb, 1983), and pattern identification (e.g., Vega-Bermudez et al., 1991). During braille reading, on the other hand, movement velocity and interlimb coordination appears to be very important, and subjects use various sorts of movements to recover different distal information during gradient texploration. An account that incorporates both instances and predicts where movement information would influence perception seems desirable.

There are other issues to consider before accepting the consequences of active-passive equivalence. For example, the problems associated with failures to locate statistically significant differences are real, and while they have been both anticipated and defended in the literature, they persist as an issue of concern. But this is not the only problem. Comparisons of active and passive touch are complicated also by the fact that, under normal circumstances, active touch is characterized by repetitive, bidirectional, cyclical sweeps with irregular if grossly similar velocity profiles. By contrast, passive touch is often created by constant unidirectional velocity displays, such as textured rotating drums against which the finger is applied with some stable force (e.g., Vega-Bermudez et al., 1991; Lamb, 1983). Such differences suggest that an important problem is the potential confounding of variables (indeed, combination of confounds are in principle possible). Could a constant relative velocity contribute to an elevation in discriminability? Were the contact forces equated throughout and if not with what consequence? Do directional changes in one condition but not in the other have any bearing on perceptual discriminability? Equating the kinematics in the two conditions would seem an important control in such investigations, but efforts to do so remain problematic.

If active movements are typically sinusoidal in velocity, arranging for a sinusoidally moving display against a static fingerpad (such as Lederman, 1981, employed) is one important control. However, insofar as this still does not guarantee idealized replication of the two kinematic profiles, it can be judged only a partial solution. Whereas an appropriate passive

322 B. Hughes, G. Jansson /Human Mooement Science 13 (1994) 301-333

controller could produce perfectly sinusoidal, and perfectly repeatable, velocity patterns, the same precise pattern is never repeated actively. Hence, to equate the relative motion in the two conditions would entail having the passive condition reflect all of the kinematic variations, large and small, that are inherent in a given active scanning pattern. To have, for each active pattern, a faithful kinematic equivalent that could be passively perceived is possible in principle but difficult to create experimentally. For this reason, primarily, we have sought a strategy that focuses on active movements and exploration only. We argue that if systematic (dynamical) movement variation produces systematic perceptual variations, the role of information in movement cannot be dismissed. In the event that suitable active-passive conditions, which equate the cutaneous events in detail, can be contrived, of course, a clearer picture of the contribution of movement information to texture perception would be possible.

4.2. What kind of touch is organized for texploration?

As noted above, Gibson (1966) referred to three perceptual subsystems that are particularly relevant to texture perception by touch. Cutaneous touch occurs without concurrent movement of the joints or length change of the musculature; haptic touch is the combination of cutaneous information with information from joint receptors into another (and different) perceptual subsystem; and dynamic touch refers to the covariation of spatial information via haptic perception and nonspatial information from muscular exertion within a different system again. Accordingly, passive touch of the sort that occurs when an experimenter moves a surface across a subject’s restrained fingerpad would be cutaneous touch in Gibson’s scheme, as would be the instance of static surface against static skin. Passive touch of the sort that occurs when an external agent (e.g., an experimenter) moves a perceiver’s (otherwise) restrained fingerpad across a surface would be haptic touch, irrespective of whether the surface was moving too. As Gibson described it, dynamic touch would encompass all those situations where the subject had to organize exploratory action of his or her own to produce information, and would be of mechanoreceptive and muscular origin. Which brings us to the crux of the problem: is texploration by touch an instance of dynamic touch or of haptic touch? It is not entirely clear from Gibson’s writings where texploration would fit into this scheme (the distinctions described above - at least those between haptic and dynamic touch - do not appear in Gibson (19621, where indeed the title refers only to “active” touch).


Gibson (1966, p. 112) offered the example of the man who can identify the texture of the ground before him by using a stick as an example of haptic touch. Presuming the holder of the stick was capable of wielding it, why would this not be an instance of dynamic touch? Gibson’s (1966, p. 128) preferred examples of dynamic touch were wielded objects: cups, sticks, objects whose weight had to be discriminated, objects whose con- tainers could be shaken for information as to content. What distinguishes the sources of effective information generated by wielding an object (for whatever purpose) from those generated by exploring by touch a textured surface (so as to determine its roughness, softness, elasticity, or microstructure)?

Turvey and his colleagues (e.g., Pagan0 and Turvey, 1993, Pagan0 et al., 1993; Solomon, 1988, Solomon and Turvey, 1988) have considered this issue more than most and their use of the term dynamic touch, which they take from Gibson, is slightly clearer but not entirely so. For example, the object being manipulated in dynamic touch “affects the tensile states of the muscles, tendons and ligaments [which] are engendered by holding and manipulating” (Pagan0 and Turvey, 1993, pp. 125-126) whereas an example of haptic touch is of “the hands enveloping an object and sweeping . . .

freely over its surfaces” (ibid, p. 126). Elsewhere, they specify that “dynamic touch is different from cutaneous and haptic touch in that the muscular exertion involved in producing a given distribution of mechanical energy is relevant to the properties being perceived” (Pagan0 et al., 1993, p. 43). This is entirely comprehensible in the case of wielding a rod in order to estimate reaching distance. But does it not also encompass the sort of activity - the sliding, the pushing, the shearing, the pressing, the controlled slipping, and so on - that is the essence of texploration? We would argue (provisionally) that it does: that texploration is dynamic touch insofar as what is perceived is determined in part by the action that is applied to it. Given that this runs counter to the implications of active-passive equivalence, we contend that empirical work will be critical in making this determination.

The hypothesis that texploration is three- rather than two-dimensional exploration may also be considered. As Katz (192.5/1989) observed, and as others (e.g., Lederman, 1974) have confirmed (see above), what is perceived during 2D texploration can be influenced by the force applied in the third dimension (i.e., by pressure exerted orthogonally to the texture surface). A surface feels rougher if the force applied during contact is increased, even if no movement actually occurs normal to the surface. Our


own investigations into texture gradient perception suggest, as well, that force is rarely uniform during scanning and that subjects will modulate it during exploration. The application of graded forces in combination with graded hand velocities in texploration would be an additional reason for treating texploration as dynamic touch.

Questions related to the characterization of texploration are not idle. If one considers texploration to be an instance of haptic, but not dynamic, touch then certain sources of information (all those deriving from muscular exertion and tendinous strain) would not need to be an integral part of any model of it. Models of texture perception could be restricted to consideration of the information in cutaneous discharge, which while not straightfor- ward, would not be complicated by having to incorporate mechanical variables. Connor and Johnson (1992) have begun to develop one such model of cutaneous and neural coding.

Conversely, if texploration is an instance of dynamic touch then certain important and, as Gibson anticipated, different sorts of explanatory con- cepts are going to have to be invoked: accounts of dynamic touch are eventually going to necessitate explanations of how the information from three disparate sources is united in a single subsystem. The flux of stimulation from which such information is to be extracted would include nonspatial, mechanical variables generated by the musculature as part of an exploratory act. As Turvey and his colleagues have detailed, wielding objects yields information in both geometrical and (bio)mechanical stimulation. Consequently their modelling of this perceptual performance has incorporated mechanical, not spatial, parameters. Modelling texploration as haptic touch (or even cutaneous touch) would perhaps be a simpler enterprise; it is not clear that it would be a more accurate one as a result. On the other hand, mistakenly modelling haptic touch as dynamic touch would entail seeking informational invariants in the wrong place and using inappropriate tools.

4.3. Perceptual constancy in texture perception

One of the phenomena to which texploration gives rise, and upon which investigators since Katz (1925/1989) have remarked, is the perceptual stability or constancy across encounters with a given surface. This is not, strictly, the same thing as the active-passive equivalence referred to above: by constancy we refer to the perception of an invariant distal surface structure in the face of highly variable proximal conditions of exposure to it (see also Lederman, 1982.)

B. Hughes, G. Jansson /Human Mocement Science 13 (1994) 301-333 325

In our experience, the patterns of movements that subjects generate can be categorized in a few ways (see above), such as rapid, low-force scanning or slow, high-force scanning but within these general categories it would be very difficult to locate kinematically identical patterns. That is, scanning movements during active touch have a degree of stereotopy but are rarely stereotyped at the kinematic level. How is this variation to be understood? Is it the unintentional by-product of an imprecise motor controller or perhaps the result of unexpected changes in the frictional coefficient that perturbs the trajectory of the exploring finger or hand? Is kinematic variance treatable as error that has to be tolerated or noise to be corrected for (with some degree of success) by the perceptual system to which it is linked? To the degree that texploration is not stereotyped, an adequate model of it would contain an account of how variable exposures to a surface (i.e., movements that differ in kinematic detail; cutaneous events that differ continuously in the spatiotemporal structure) can give rise to invariant estimates of textural structure.

If, during texploration, subjects engage in the perceptual-motor exploration of the type Gibson (1962) described for the recovery of shape, one might find subjects intentionally varying their movements because such kinematic variation serves to enrich the spatiotemporal flux and this in turn eases the extraction of spatial invariants. Others (e.g., Solomon and Turvey, 1988) have found subjects, in judging reachability with hand-held rods, to reveal rapid perceptual-motor learning, trying many types of movements but soon locating a consistent and effective style of wielding, termed “style-stability” (Solomon, 1988, p. 220). Do comparable searches for style stability in the action system emerge in texture perception, and if so what are their characteristics? Is the search for style-stability, to the extent it is evident, determined by the geometry of the surfaces being explored, by task demands, or by the effectiveness of the information to which they give rise?

If the precise manner in which the movements are made is not the critical aspect of the perception of texture, as follows from claims that texploration serves only to produce and not modulate the flux of cutaneous stimulation, it is unclear why such stability would arise in texture perception. One might as easily expect to find scanning movements limited in their kinematic consistency to a “sensory window” constrained in size by the operating characteristics of the mechanoreceptors (factors might include avoidance of adaptation, phase-locking of different receptor dis- charges, spatiotemporal resolution, etc.). If velocity, or its rates of change, are parameters that have to be “taken into account” by the system, as


might be expected by an evaluative signal system (e.g., feedback compara- tors), or by a computational model, one might also expect to find subjects attempting to move at a constant velocity for relatively large portions of a scanning cycle. Computing spatial structure would be computationally simplified (in principle, if not biologically) if there were no spatiotemporal variation in the discharge of cutaneous receptors that had to be separated from any spatiotemporal variation owing to changing velocity.

4.4. From roughness perception to the recouery of microstructure

While textural roughness/smoothness perception is among the most frequent (and hence ecologically significant) uses to which active touch is put, it is but one perceptual dimension to which textured surfaces give rise. Connor and Johnson’s (1992; Connor et al., 1990) findings that roughness estimates cannot be based on any simple spatial or temporal intensity scale, but only on a more complex spatial-difference basis, which is “above” the level of individual mechanoreceptors, leads to their claim that textural roughness is an intensive continuum without an intensive neural coding scheme, and emphasizes that the structural determinants of roughness perception are yet to be completely identified (Johnson and Hsiao, 1992; Loomis and Lederman, 1986). Moreover, estimates of roughness are relatively simple tasks in the realm of touch, certainly far less demanding than the recovery of microstructure during braille reading or raised-line map exploration. The extension of the consensus of active-passive equivalence (with the implications this has for the role of action in texture perception) from roughness to texploration is not warranted and should not be pre- sumed.

One useful extension might be to experimentally manipulate both the spatial structure of the textured surfaces and the structure of the movements used in exploring them. We have begun exploring this by using surfaces whose elements are structured in different ways but which all contain spatial gradients of various types and magnitudes (i.e., the spatial period of texture elements is not constant within a given surface). Our rationale has been that observation of the consequences for perception of such structural, spatial gradients considered in combination with, and not separate from, the movements over them could yield valuable information about texture perception generally and the issues outlined above more specifically. Since the movements during active touch are predominantly graded - in velocity most apparently, but also in force - the covariation of


the structure and motion can lead to insights into the organization of a texploration system.

One important question is whether, given some textured surface contain- ing a gradient or discontinuity (and no other perceptual access except that via active touch), the spatial structure attracts a preferred scanning style, say, in frequency. What, if any, changes in scanning frequency are exhibited by changes in the geometry of the discontinuity relative to the surface; i.e., in its type, magnitude, or location? Given that a subject exhibits a “preferred” frequency for a given surface, what is the consequence of driving the frequency from its preferred rate, in either direction? Indeed, does the preferred frequency correspond to an ideal frequency; i.e., are subjects most accurate when moving at a self-determined frequency? Does a perceptual constancy emerge, and if so, over what range of spatial and movement parameters?

Evidence of preferred frequencies, evidence of a high correlation between preferred frequencies and perceptual accuracy, and evidence that departures from preferred frequencies diminish perceptual accuracy, would collectively speak to the mutual dependence - indeed, the integration - of the perception and action subsystems during texploration. It would, we argue, suggest that, far from being irrelevant to texture perception, self- organized movement is integral to texture perception. Given that adequate controls could be contrived, such a paradigm might also be applied to the active-passive question. We would anticipate that prior experimental support for the hypothesis that a preferred scanning frequency coincides with the ideal one would lead in turn to an expectation of a superiority of the active over passive version.

5. Summary and concluding remarks

The term haptics reflects the long-held importance that has been at- tached to the directness of the interaction of movement and perception during touch. However, as we hope to have suggested, certain conceptual and empirical puzzles persist. There is now an emerging consensus that active and passive texture perception offer equivalent perceptual information. This, in turn, suggests that the movements a person makes while in physical contact with a surface have no informational bearing on the perception that ensues. We have been concerned that such a conclusion is premature, that the generalizability of this remains in question, and that how and when such equivalences extend remains an empirical question.


This later issue is not unconnected to the Gibsonian characterization of perceptual subsystems in touch. Determining which of the potential sources of information offered by texploration are actually effective will constrain models of texture perception. If texture perception is based on cutaneous discharge only, the role of movement is simply to permit the discharge to take place (and hence, passive motion works just as well) and information regarding action patterns is superfluous. On the other hand, if texploration is a form of dynamic touch - if there is information for perception in the exploratory actions themselves - not only is the consensus of active-passive equivalence in jeopardy, but quite different models will likely ensue. We have suggested that there are means by which these questions can be addressed empirically and that different accounts of texture perception might make quite different predictions regarding the organization of the exploratory action system and the perceptual sensitivity of the system given a particular organization. The outcome of such work would not only have ramifications for theories of information utilization in perceptual-motor activities, but would also serve as the basis for the emergence of more effective systems for those such as the visually impaired for whom active touch remains a critical source of spatial information.

Acknowledgements

Preparation of this manuscript was supported in part by grants from the New Zealand Lottery Science Board and the University of Auckland Research Committee to Barry Hughes and from the Swedish Council for Planning and Coordination of Research, the Swedish Council for Social Research, and the Swedish Work Environment Fund to Gunnar Jansson. Our experimental work referred to in this article has been carried out in compliance with guidelines set by the University of Auckland’s Human Subjects Ethics Committee and has benefitted from the contributions of Emily Holmes and Lars Backstrom. The constructive critical comments of Peter Beek, Piet van Wieringen, Morton Heller, and those of an anony- mous reviewer in particular are gratefully acknowledged.

References

Beek, P.J., M.T. Turvey and R.C. Schmidt, 1992. Autonomous and nonautonomous dynamics of

coordinated rhythmic movements. Ecological Psychology 4, 65-95.

B. Hughes, G. Jansson /Human Mouement Science 13 (1994) 301-333 329

Bertelson, P. and P. Mousty, 1989. Simultaneous reading of braille with the two hands: Reply to Millar

(1987). Cortex 25, 495-498.

Bertelson, P., P. Mousty and G. D’Alimonte, 1985. A study of braille reading: 2. Patterns of hand

activity in one-handed and two-handed reading. Quarterly Journal of Experimental Psychology 37A,

235-256.

Burton, G., 1993. Non-neural extensions of haptic sensitivity. Ecological Psychology 5, 105-124.

Burton, G. and M.T. Turvey, 1990. Perceiving the lengths of rods that are held but not wielded.

Ecological Psychology 2, 295-324.

Connor, C.E., S.S. Hsiao, J.R. Phillips and K.O. Johnson, 1990. Tactile roughness: Neural codes that

account for psychophysical magnitude estimates. Journal of Neuroscience 10, 3823-3836.

Connor, C.E. and K.O. Johnson, 1992. Neural coding of tactile texture: Comparison of spatial and

temporal mechanisms for roughness perception. Journal of Neuroscience 12, 3414-3426.

Cruse, H., J. Dean, H. Heuer and R.A. Schmidt, 1990. ‘Utilization of sensory information for motor

control’. In: 0. Neumann and W. Prinz (Eds.1, Relationships between perception and action (pp.

43-79). Berlin: Springer-Verlag.

Darian-Smith, I. and L.E. Oke, 1980. Peripheral neural representation of the spatial frequency of a

grating moving across the monkey’s finger pad. Journal of Physiology (London) 309, 117-133.

Drucker, S.M., 1988. ‘Texture from touch’. In: W. Richards (Ed.), Natural computation (pp. 422-429).

Cambridge, MA: MIT Press.

Epstein, W., 1973. The process of taking into account in visual perception. Perception 2, 267-285.

Epstein, W., 1985. ‘Transmodal perception and amodal information’. In: D. Warren and E. Strelow

(Eds.1, Electronic spatial sensing for the blind (pp. 421-430). Dordrecht: Martinus Nijhoff.

Epstein, W., 1986. Contrasting conceptions of perception and action. Acta Psychologica 63, 103-115.

Epstein, W., B. Hughes, S. Schneider and P. Bach-y-Rita, 1986. Is anything out there? A study of distal attribution in response to vibrotactile stimulation. Perception 15, 275-284.

Foulke, E., 1991. ‘Braille’. In: M.A. Heller and W. Schiff (Eds.), The psychology of touch (pp. 219-233).

Hillsdale, NJ: Erlbaum.

Franzen, 0. and J. Westman, 1991. Information processing in the somatosensory system. London:

Macmillan.

Gandevia, SC. and D. Burke, 1992. Does the nervous system depend on kinesthetic information to

control natural limb movements? Behavioral and Brain Sciences 15, 614-632.

Gibson, J.J., 1962. Observations on active touch. Psychological Review 69, 477-491.

Gibson, J.J., 1966. The senses considered as perceptual systems. London: George Allen and Unwin.

Gibson, J.J. 1982a. ‘Notes on action’. In: E. Reed and R. Jones (Eds.), Reasons for realism (pp.

385-392). Hillsdale, NJ: Erlbaum.

Gibson, J.J., 1982b. ‘The uses of proprioception and the detection of propriospecific information’. In:

E. Reed and R. Jones (Eds.1, Reasons for realism (pp. 164-170). Hillsdale, NJ: Erlbaum.

Goodale, M.A. (Ed.), 1989. Vision and action: The control of grasping. Norwood, NJ: Ablex.

Goodwin, A.W., K.T. John, K. Sathian and I. Darian-Smith, 1989. Spatial and temporal factors determining afferent fiber responses to a grating moving sinusoidally over the monkey’s fingerpad. Journal of Neuroscience 9, 1280-1293.

Goodwin, A.W. and J.W. Morley, 1987a. Sinusoidal movement of a grating across the monkey’s

fingerpad: Effect of contact angle and force of the grating on afferent fiber responses. Journal of Neuroscience 7, 2168-2180.

Goodwin, A.W. and J.W. Morley, 1987b. Sinusoidal movement of a grating across the monkey’s fingerpad: Representation of grating and movement features in afferent fiber responses. Journal of Neuroscience 7, 2192-2202.

Grunwald, A.P., 1966. A braille-reading machine. Science 154, 144-146.

Griisser, O.-J., 1986. Interaction of efferent and afferent signals in visual perception: A history of ideas and experimental paradigms. Acta Psychologica 63, 3-21.


Heller, M.A., 1982. Visual and tactual texture perception: Intersensory cooperation. Perception and Psychophysics 31, 339-344.

Heller, M.A., 1985. Tactual perception of embossed morse code and braille: The alliance of vision and

touch. Perception 14, 563-570.

Heller, M., 1989. Texture perception in sighted and blind observers. Perception and Psychophysics 45,

49-54.

Heller, M., 1993. Influence of visual guidance on braille recognition: Low lighting also helps touch.

Perception and Psychophysics 54, 675-681.

Holmes, E., 1993. Tactile texture gradients: Representing the third dimension in pictures for the blind.

Unpublished manuscript, Department of Psychology, Oxford University.

James, G.A., 1982. ‘Mobility maps’. In: W. Schiff and E. Foulke (Eds.), Tactual perception: A

Sourcebook (pp. 334-363). Cambridge: Cambridge University Press.

Jansson, G., 1983. ‘Tactual maps as a challenge for perception research’. In: J. Wiedel (Ed.),

Proceedings of the First International Symposium on Maps and Graphics for the Visually Handi-

capped (pp. 68-74). Washington, DC: Association of American Geographers.

Jansson, G., 1988. ‘What are the problems for pictures for the blind and what can be done to solve

them?’ In: C.W.M. Magnee, F.J.M. Vlaskamp, M. Soede and B. Butcher (Eds.), Man-machine

interfaces, graphics and practical applications (pp. 18-24). London: Royal National Institute for the

Blind. Jansson, G., 1991. ‘Possibilities and limitations of tactile computer graphics for the blind’. In: Y.

Queinnec and F. Daniellou (Eds.), Designing for everyone, Vol 2 (pp. 1616-1618). London: Taylor

and Francis.

Jansson, G., 1992. ‘3-D perception from tactile computer displays’. In: W. Zagler (Ed.), Computers for

Handicapped Persons: Proceedings of the 3rd International Conference on Computers for Handi-

capped Persons (pp. 233-237). Vienna.

Jansson, G. and B. Hughes, 1990. Virtual textures and active vibrotactile perception. Unpublished manuscript, Department of Psychology, University of Uppsala.

Jansson, G. and B. Hughes, 1991a. Judgments of tactile texture gradients. Journal of the Acoustical

Society of America 89, 2004.

Jansson, G. and B. Hughes, 1991b. Tactual perception of 3D scenes in 2D pictures. Proceedings of the

Sixth International Conference on Perception and Action (p. 34). Amsterdam, August.

Jansson, G., E. Holmes and B. Hughes, 1994. Haptic perception of texture gradients. Manuscript in

preparation. Jeannerod, M., 1986. The formation of finger grip during prehension. Behavioral and Brain Research

19, 99-116. Johansson, R.S. and G. Westling, 1984. ‘Influences of cutaneous input on the motor coordination

during precision manipulation’. In: C. von Euler and 0. Franz&t (Eds.), Somatosensory mechanisms (pp. 249-260). New York: Macmillan.

Johansson, R.S. and G. Westling, 1987. Signals in tactile afferents from the fingers elicting motor

responses during precision grip. Experimental Brain Research 66, 141-154.

Johansson, R.S. and G. Westling, 1990. ‘Tactile afferent signals in the control of precision grip’. In: M.

Jeannerod (Ed.), Attention and performance XIII (pp. 677-713). Hillsdale, NJ: Erlbaum. Johnson, K.O., 1983. Neural mechanisms of tactual form and texture discrimination. Federation

Proceedings 42, 2542-2547. Johnson, K.O. and S.S. Hsiao, 1992. Neural mechanisms of tactual form and texture perception. Annual

Review of Neuroscience 15, 227-250.

Johnson, K.O. and J.R. Phillips, 1981. Tactile spatial resolution. I. Two-point discrimination, gap

detection, grating resolution and letter recognition. Journal of Neurophysiology 46, 1177-1191. Jones, B. and S. O’Neill, 1985. Combining vision and touch in texture perception. Perception and

Psychophysics 37, 66-72.

B. Hughes, G. Janson /Human Movement Science 13 (1994) 301-333 331

Katz, D., 1925/1989. The world of touch. (Translated and edited by L.E. Krueger). Hillsdale, NJ:

Erlbaum. Kay, B.A., J.A.S. Kelso, E.L. Saltzman and G. Schiiner, 1987. Space-time behavior of single and

bimanual rhythmical movements: Data and and limit cycle model. Journal of Experimental Psychol-

ogy: Human Perception and Performance 13, 178-192. Kelso, J.A.S. and G. Schiiner, 1988. Self-organization of coordinative movement patterns. Human

Movement Science 7, 27-46. Kennedy, J.M., P. Gabias and A. Nicholls, 1991. ‘Tactile pictures’. In: M.A. Heller and W. Schiff (Eds.),

The psychology of touch (pp. 263-299). Hillsdale, NJ: Erlbaum.

Klatzky, R.L and S. Lederman, 1990. ‘Intelligent exploration by the human hand’. In: ST. Venkatara-

man and T. Iberall (Eds.), Dexterous robot hands (pp. 66-81). New York: Springer-Verlag.

Klatzky, R.L., S. Lederman and V.A. Metzger, 1985. Identifying objects by touch: An ‘expert system’.

Perception and Psychophysics 37, 299-302.

Klatzky, R.L., S. Lederman and C. Reed, 1987. There’s more to touch than meets the eye: The salience

of object attributes for haptics with and without vision. Journal of Experimental Psychology:

General 116, 356-369. Klatzky, R.L., S. Lederman and C. Reed, 1989. Haptic integration of object properties: Texture,

hardness and planar contour. Journal of Experimental Psychology: Human Perception and Perfor-

mance 15, 45-57.

Klatzky, R.L., J. Pellegrino, B.P. McCloskey and S.J. Lederman, 1993. Cognitive representations of

functional interactions with objects. Memory and Cognition 21, 294-303.

Krueger, L., 1970. David Katz’s Der Aufbau der Tastwelt [The world of touch]: A synopsis. Perception

and Psychophysics 7, 337-341. Krueger, L. 1982. ‘Tactual perception in historical perspective: David Katz’s world of touch’. In: W.

Schiff and E. Foulke (Eds.), Tactual perception: A sourcebook (pp. l-54). Cambridge: Cambridge

University Press.

Lamb, G.D., 1983. Tactile discrimination of textured surfaces: Psychophysical performance measure-

ments in humans. Journal of Physiology (London) 338, 551-565.

LaMotte, R.H. and M.A. Srinivasan, 1991. ‘Surface microgeometry: Neural encoding and perception’.

In: 0. Franzen and J. Westman (Eds.), Information processing in the somatosensoty system (pp.

49-58). London: Macmillan.

Lederman, S.J., 1974. Tactile roughness of grooved surfaces: The touching process and effects of

macro- and microsurface structure. Perception and Psychophysics 16, 385-395. Lederman, S.J., 1981. The perception of surface roughness by active and passive touch. Bulletin of the

Psychonomic Society 18, 253-255.

Lederman, S.J., 1982. ‘The perception of texture by touch’. In: W. Schiff and E. Foulke (Eds.), Tactual

perception: A sourcebook (pp. 130-167). Cambridge: Cambridge University Press.

Lederman, S.J., 1983. Tactual roughness perception: Spatial and temporal determinants. Canadian

Journal of Psychology 37, 498-511.

Lederman, S.J. and S.G. Abbott, 1981. Texture perception: Studies of intersensory organization using a

discrepancy paradigm and visual versus tactual psychophysics. Journal of Experimental Psychology:

Human Perception and Performance 7, 902-915.

Lederman, S.J. and D.H. Kinch, 1979. Texture in tactual maps and graphics for the visually handi-

capped. Journal of Visual Impairment and Blindness 73, 217-227.

Lederman, S.J. and R.L. Klatzky, 1987. Hand movements: A window into haptic object recognition. Cognitive Psychology 19, 342-368.

Lederman, S.J. and R.L. KIatzky, 1990. Haptic classification of common objects: Knowledge-driven exploration. Cognitive Psychology 22, 421-459.

Lederman, S.J., G. Thorne and B. Jones, 1986. Perception of texture by vision and touch: Multidimen-

sionality of intersensory integration. Journal of Experimental Psychology: Human Perception and Performance 12, 169-180.


Loomis, J.M. and S.J. Lederman, 1986. ‘Tactual perception’. In: K. Boff, L. Kaufman and J. Thomas (Eds.), Handbook of human perception and performance, Vol 1 (pp 31.01-31.41). New York: Wiley.

Magee, L.E. and J.M. Kennedy, 1980. Exploring pictures tactually. Nature 283, 287-288.

Meenes, M. and M.J. Zigler, 1923. An experimental study of the perceptions of roughness and

smoothness. American Journal of Psychology 34, 542-549.

Millar, S., 1987. The perceptual ‘window’ in two-handed braille: Do the left and right hands process

text simultaneously? Cortex 23, 11 l-122.

Morley, J.W. and A.W. Goodwin, 1987. Sinusoidal movement of a grating across the monkey’s

fingerpad: Temporal patterns of afferent fiber responses. Journal of Neuroscience 7, 2181-2191.

Morley, J.W., A.W. Goodwin and I. Darian-Smith, 1983. Tactile discrimination of gratings. Experimen-

tal Brain Research 49, 291-299.

Mousty, P. and P. Bertelson, 1985. A study of braille reading: 1. Reading speed as a function of hand

useage and context. Quarterly Journal of Experimental Psychology 37A, 217-233.

Pagano, C.C., P. Fitzpatrick and M.T. Turvey, 1993. Tensorial basis to the constancy of perceived object

extent over variations of dynamic touch. Perception and Psychophysics 54, 43-54.

Pagano, C.C. and M.T. Turvey, 1993. Perceiving by dynamic touch the distances reachable with

irregular objects. Ecological Psychology 5, 125-151. Phillips, J.R. and K.O. Johnson, 1981a. Tactile spatial resolution. II. Neural representation of bars,

edges and gratings in monkey primary afferents. Journal of Neurophysiology 46, 1192-1203. Phillips, J.R. and K.O. Johnson, 1981b. Tactile spatial resolution. III. A continuum mechanics model of

skin predicting mechanoreceptor responses to bars, edges and gratings. Journal of Neurophysiology

46, 1204-1225.

Prinz, W., 1990. ‘A common coding approach to perception and action’. In: 0. Neumann and W. Prinz

(Eds.), Relationships between perception and action (pp. 167-201). Berlin: Springer-Verlag. R&&z, G., 1950. The psychology and art of the blind. (H.A. Wolff, trans.) London: Longmans, Green

and Co. Sathian, K., A.W. Goodwin, K.T. John and I. Darian-Smith, 1989. Perceived roughness of a grating:

Correlation with responses of mechanoreceptive afferents innervating the monkey’s fingerpad.

Journal of Neuroscience 9, 1280-1293.

Scheerer, E., 1984. ‘Motor theories of cognitive structure: A historical review’. In: W. Prinz and A.F.

Sanders (Eds.), Cognition and motor processes (pp. 77-98). Berlin: Springer-Verlag. Schmidt, R.A., 1988. ‘Motor and action perspectives on motor behaviour’. In: O.G. Meijer and K. Roth

(Eds.), Complex movement behaviour: ‘The’ motor-action controversy (pp. 3-44). Amsterdam:

North-Holland. Sherrick, C.E. and R.W. Cholewiak, 1986. ‘Cutaneous sensitivity’. In: K. Boff, L. Kaufman and J.

Thomas (Eds.), Handbook of human perception and performance, Vol 1 (pp 12.1-12.58). New

York: Wiley.

Solomon, H.Y., 1988. Movement-produced invariants in haptic explorations: An example of a self-

organizing, information-driven, intentional system. Human Movement Science 7, 201-224. Solomon, H.Y. and M.T. Turvey, 1988. Haptically perceiving the distances reachable with hand-held

objects, Journal of Experimental Psychology: Human Perception and Performance 14, 404-427.

Taylor, M.M. and S.J. Lederman, 1975. Tactile roughness of grooved surfaces: A model and the effect

of friction. Perception and Psychophysics 17, 23-36. Taylor, M.M., S.J. Lederman and R.H. Gibson, 1973. ‘Tactual perception of texture’. In: E.C.

Carterette and M.P. Friedman (Eds.), Handbook of perception, Vol III: Biology of perceptual

systems (pp. 251-272). New York: Academic Press. Turvey, M.T., 1977. ‘Preliminaries to a theory of action with respect to vision’. In R. Shaw and J.

Bransford (Eds.), Perceiving, acting and knowing: Toward an ecological psychology (pp. 211-265).

Hillsdale, NJ: Erlbaum. Vega-Bermudez, F., K.O. Johnson and S.S. Hsiao, 1991. Human tactile pattern recognition: Active


versus passive touch, velocity effects and patterns of confusion. Journal of Neurophysiology 65,

531-546. Venkataraman, S.T. and T. Iberall (Eds.), 1990. Dexterous robot hands. New York: Springer-Verlag.

Viviani, P. and N. Stucchi, 1992. ‘Motor-perceptual interactions’. In: G.E. Stelmach and J. Requin

(Eds.), Tutorials in motor behavior II (pp. 229-248). Amsterdam: North-Holland. Warren, D.H. and M.J. Rossano, 1991. ‘Intermodality relations: Vision and touch’. In: M.A. Heller and

W. Schiff (Eds.), The psychology of touch (pp. 119-137). Hillsdale, NJ: Erlbaum. Weber, E.H., 1978. The sense of touch [De Tactu (1834) (H.E. Ross, trans.) and Der Tastsinn (1842)

(D.J. Murray, trans.)]. New York: Academic Press.

texture perception via active touch

Documents