Figure/Ground Segmentation via a Haptic Glance: Attributing Initial Finger Contacts to Objects or Their Supporting Surfaces.

The current study addresses the well-known "figure/ground" problem in human perception, a fundamental topic that has received surprisingly little attention from touch scientists to date. Our approach is grounded in, and directly guided by, current knowledge concerning the nature of haptic processing. Given inherent figure/ground ambiguity in natural scenes and limited sensory inputs from first contact (a "haptic glance"), we consider first whether people are even capable of differentiating figure from ground (Experiments 1 and 2). Participants were required to estimate the strength of their subjective impression that they were feeling an object (i.e., figure) as opposed to just the supporting structure (i.e., ground). Second, we propose a tripartite factor classification scheme to further assess the influence of kinetic, geometric (Experiments 1 and 2), and material (Experiment 2) factors on haptic figure/ground segmentation, complemented by more open-ended subjective responses obtained at the end of the experiment. Collectively, the results indicate that under certain conditions it is possible to segment figure from ground via a single haptic glance with a reasonable degree of certainty, and that all three factor classes influence the estimated likelihood that brief, spatially distributed fingertip contacts represent contact with an object and/or its background supporting structure.

Tactile and Haptic Illusions.

This paper surveys the research literature on robust tactile and haptic illusions. The illusions are organized into two categories. The first category relates to objects and their properties, and is further differentiated in terms of haptic processing of material versus geometric object properties. The second category relates to haptic space, and is further differentiated in terms of the observer's own body versus external space. The illusions are initially described and where possible addressed in terms of their functional properties and/or underlying neural processes. The significance of these illusions for the design of tactile and haptic displays is also discussed. We conclude by briefly considering a number of important general themes that have emerged in the materials surveyed.

Haptic Classification of Facial Identity in 2D Displays: Configural versus Feature-Based Processing.

Participants learned through feedback to haptically classify the identity of upright versus inverted versus scrambled faces depicted in simple 2D raised-line displays. We investigated whether identity classification would make use of a configural face representation, as is evidenced for vision and 3D haptic facial displays. Upright and scrambled faces produced equivalent accuracy, and both were identified more accurately than inverted faces. The mean magnitude of the haptic inversion effect for 2D facial identity was a sizable 26 percent, indicating that the upright orientation was ¿privileged¿ in the haptic representations of facial identity in these 2D displays, as with other facial modalities. However, given the effect of scrambling, we conclude that configural processing was not employed; rather, only local information about the features was used, the features being treated as oriented objects within a body-centered frame of reference. The results indicate a fundamental difference between haptic identification of 2D facial depictions and 3D faces, paralleling a corresponding difference in recognition of nonface objects.

Haptic perception: a tutorial.

This tutorial focuses on the sense of touch within the context of a fully active human observer. It is intended for graduate students and researchers outside the discipline who seek an introduction to the rapidly evolving field of human haptics. The tutorial begins with a review of peripheral sensory receptors in skin, muscles, tendons, and joints. We then describe an extensive body of research on "what" and "where" channels, the former dealing with haptic perception of objects, surfaces, and their properties, and the latter with perception of spatial layout on the skin and in external space relative to the perceiver. We conclude with a brief discussion of other significant issues in the field, including vision-touch interactions, affective touch, neural plasticity, and applications.

Haptic Processing of Facial Expressions of Emotion in 2D Raised-Line Drawings.

Participants haptically (vs. visually) classified universal facial expressions of emotion (FEEs) depicted in simple 2D raised-line displays. Experiments 1 and 2 established that haptic classification was well above chance; face-inversion effects further indicated that the upright orientation was privileged. Experiment 2 added a third condition in which the normal configuration of the upright features was spatially scrambled. Results confirmed that configural processing played a critical role, since upright FEEs were classified more accurately and confidently than either scrambled or inverted FEEs, which did not differ. Because accuracy in both scrambled and inverted conditions was above chance, feature processing also played a role, as confirmed by commonalities across confusions for upright, inverted, and scrambled faces. Experiment 3 required participants to visually and haptically assign emotional valence (positive/negative) and magnitude to upright and inverted 2-D FEE displays. While emotional magnitude could be assigned using either modality, haptic presentation led to more variable valence judgments. We also documented a new face-inversion effect for emotional valence visually, but not haptically. These results suggest emotions can be interpreted from 2-D displays presented haptically as well as visually; however, emotional impact is judged more reliably by vision than by touch. Potential applications of this work are also considered.

Haptic recognition of static and dynamic expressions of emotion in the live face.

If humans can detect the wealth of tactile and haptic information potentially available in live facial expressions of emotion (FEEs), they should be capable of haptically recognizing the six universal expressions of emotion (anger, disgust, fear, happiness, sadness, and surprise) at levels well above chance. We tested this hypothesis in the experiments reported here. With minimal training, subjects' overall mean accuracy was 51% for static FEEs (Experiment 1) and 74% for dynamic FEEs (Experiment 2). All FEEs except static fear were successfully recognized above the chance level of 16.7%. Complementing these findings, overall confidence and information transmission were higher for dynamic than for corresponding static faces. Our performance measures (accuracy and confidence ratings, plus response latency in Experiment 2 only) confirmed that happiness, sadness, and surprise were all highly recognizable, and anger, disgust, and fear less so.

The material-weight illusion revisited.

Experiment 1 documents modality effects on the material-weight illusion for a low-mass object set (58.5 g). These modality effects indicate that the material-weight illusion is principally a haptically derived phenomenon: Haptically accessed material cues were both sufficient and necessary for full-strength illusions, whereas visually accessed material cues were only sufficient to generate moderate-strength illusions. In contrast, when a high-mass object set (357 g) was presented under the same modality conditions, no illusions were generated. The mass-dependent characteristic of this illusion is considered to be a consequence of differing grip forces. Experiment 2 demonstrates that the enforcement of a firm grip abolishes the low-mass material-weight illusion. Experiment 3 documents that a firm grip also diminishes perceptual differentiation of actual mass differences. Several possible explanations of the consequences of increasing grip force are considered.

Manipulation with no or partial vision.

The present research investigated the role of vision in closed- and open-loop processing during manipulation. In Experiment 1, participants performed common manipulatory tasks with 100% accuracy in less than 1 s without vision. In Experiment 2, the effects of extensive practice of a peg-in-hole task were examined within 4 functionally significant stages of manipulation. Performance was consistently faster with than without vision in the prereach, grasp, and transport + insert stages; reverse effects were observed during the reach stage. In Experiment 3, the effects of practice with partial vision were examined: Participants initially learned the peg-in-hole task with full vision and then transferred to learning the same task with vision available only during 1 functional stage. Overall, performance was fastest when vision was limited to the prereach and reach stages.

Tactile roughness perception with a rigid link interposed between skin and surface.

Subjects made roughness judgments of textured surfaces made of raised elements, while holding stick-like probes or through a rigid sheath mounted on the fingertip. These rigid links, which impose vibratory coding of roughness, were compared with the finger (bare or covered with a compliant glove), using magnitude-estimation and roughness differentiation tasks. All end effectors led to an increasing function relating subjective roughness magnitude to surface interelement spacing, and all produced above-chance roughness discrimination. Although discrimination was best with the finger, rigid links produced greater perceived roughness for the smoothest stimuli. A peak in the magnitude-estimation functions for the small probe and a transition from calling more sparsely spaced surfaces rougher to calling them smoother were predictable from the size of the contact area. The results indicate the potential viability of vibratory coding of roughness through a rigid link and have implications for teleoperation and virtual-reality systems.

The influence of visual illusions on grasp position.

Visual size illusions have been shown to affect perceived object size but not the aperture of the hand when reaching to those same objects. Thus, vision for perception is said to be dissociated from vision for action. The present study examines the effect of visual-position and visual-shape illusions on both the visually perceived center of an object and the position of a grasp on that object when a balanced lift is required. The results for both experiments show that although the illusions influence both the perceived and the grasped estimates of the center position, the grasp position is more veridical. This partial dissociation is discussed in terms of its implications for streams of visual processing.

Anticipating load torques produced by voluntary movements.

The stability of an object held between the finger and thumb depends on friction developed by grip force, normal to the contact surfaces, to overcome tangential load force. Previous research has shown that in lifting an object, grip force rises with the increase in gravitational load force as the hand takes the weight and that in moving an object, grip force is adjusted to meet movement-induced inertial load force. Those results demonstrated the anticipatory nature of coordination of grip force with load force. Whether grip force anticipates load torque was studied in this research. When participants were constrained to use grasp points where the grasp axis was manifestly distant from object center of mass, it was found that they made grip force adjustments in anticipation of load torques that tended to destabilize an object as a result of lifting or moving it. These adjustments imply use of information about object center of mass in movement planning.

The golf-ball illusion: evidence for top-down processing in weight perception.

Theories of weight illusions have traditionally emphasised either the primary contribution of low-level sensory cues or the role of expectation based on knowledge and past experience. Current models of weight illusions lean quite strongly towards sensory-based interpretations. The current experiment raises a problem for such approaches by generating a weight illusion that is difficult to explain other than by the participants' knowledge. Golfers (who expect a weight difference between ball types) reliably judged practice golf balls to weigh more than real golf balls of the same weight. In contrast, non-golfers (who expect no weight difference between ball types) judged practice and real balls of equal weigh to weight the same. Furthermore, within the group of golfers, those who expected the weights of the two ball types to be the most discrepant prior to lifting tended to report the strongest illusions subsequent to lifting. Because there is no low-level sensory cue between ball types that on its own would signal a weight difference, the current finding suggests that there is a top-down component to weight perception that is based on experience with specific objects.

Relative availability of surface and object properties during early haptic processing.

How the relative order in which 4 property classes of haptically perceived surfaces becomes available for processing after initial contact was studied. The classes included material, abrupt-surface discontinuity, relative orientation, and continuous 3-D surface contour properties. Relative accessibility was evaluated by using the slopes of haptic search functions obtained with a modified version of A. Treisman's (A. Treisman & S. Gormican, 1988) visual pop-out paradigm; the y0 intercepts were used to confirm and fine-tune order of accessibility. Target and distractors differed markedly in terms of their value on a single dimension. The results of 15 experiments show that coarse intensive discriminations are haptically processed early on. In marked contrast, most spatially encoded dimensions become accessible relatively later, sometimes considerably so.

Effortful touch with minimum movement: revisited.

The ecological static moment-torque model proposed by C. Carello, P. Fitzpatrick, I. Domaniewicz, T.C. Chan, and M.T. Turvey (1992) does not uniquely explain the perception of rod length by static holding. Guided by a mechanical analysis of the gravitational forces and torques produced in the hand as it statically holds rods of different lengths and materials at different orientations, we offer 2 additional theoretical explanations, the force-torque and weight-percept models. Experiment 1 demonstrates that all 3 models predict perceived rod length with considerable success. Experiment 2 provides clear experimental support for the force-torque and weight-percept models over the static moment-torque model. Experiment 3 pits the former 2 models against each other. Current results favor the weight-percept model. Implications for theories of haptic weight perception and design of a new tactile sensor are also considered.

Cognitive salience of haptic object properties: role of modality-encoding bias.

The influence of modality-encoding bias on the relative importance ('cognitive salience') of object shape, size, and material, with the last determined by weight and thermal variations, was examined. Experiment 1 confirmed that for these stimulus objects all five properties were very accessible haptically, as measured by the time to identify the property level of each designated property; however, observers were still generally faster for geometric than material properties. In experiment 2, the influence of modality-encoding bias on cognitive salience was assessed by using a task involving free sorting by similarity. As predicted, modality-encoding bias strongly influenced cognitive salience. Observers favoured sorting by material under haptic- bias instructions, and three-dimensional geometric properties (especially shape) under visual-bias instructions. Videotaped hand movements indicated that modality-encoding biases reflect long-term knowledge of the relative speed and precision of manual exploration patterns, rather than exploration of the current set of objects.

Identifying objects from a haptic glance.

Subjects identified common objects under conditions of a "haptic glance," a brief haptic exposure that placed severe spatial and temporal constraints on stimulus processing. They received no advance cue, a superordinate-level name as cue, or a superordinate and basic-level name as cue. The objects varied in size relative to the fingertip and in the most diagnostic attribute, either texture or shape. The data suggest that object recognition can occur when global volumetric primitives cannot directly be extracted. Even with no cue, confusion errors resembled the target object and indicated extraction of material and local shape information, which was sufficient to provide accuracy above 20%. Performance improved with cuing, and the effect of exposure duration was observed primarily with minimal cuing, indicating compensatory effects of top-down processing.

Effects of Object Texture on Precontact Movement Time in Human Prehension.

Using kinematic data in a precision-grip reaching task, Weir, MacKenzie, Marteniuk, and Cargoe (1991) concluded that prior to contact with an object, its texture does not affect the course of grasping. The present study used their task of reaching for and lifting a slippery-, normal- (polished metal), or rough-surfaced dowel. This occurred under the original, blocked condition, in which textures were held constant within a series of trials, and under a new, randomized condition, in which textures varied randomly from trial to trial. Performance was also examined over more extended periods of practice. Reaction time and precontact movement time were directly measured. In contrast to the results of Weir et al., 1991, reaching for the slippery dowel resulted in slower movement time. This effect was found both early and late in practice for the randomized condition; it was found only in late practice for the blocked condition. These effects can be attributed to the greater geometric and dynamic precision required for lifting a slippery object.

The adaptive control of prehension.

This paper provides a brief overview of the adaptive control of prehension. Several primary issues for research in the field are considered. One pertains to the stages of prehensile activity, with particular concern for the kinematic control of precontact phases, and the coordination of load and grip forces following contact. Others relate to the basis of anticipatory and sensory-guided (via visual, proprioceptive, tactile, and haptic feedback) control of prehension. The nature of internal representations of actions is also briefly considered.