A comparison of pisiform and thumb grips in stiffness assessment.

Practitioners of spinal manipulative therapy frequently use tests that involve the therapist applying forces to the patient's spine. The therapist then makes judgments about the quality of the movement that results. A variety of variables are said to be judged during manual examination, such as resistance-free range of motion,[1,2] spasm,[2-4] end-feel,[2,3,5,6] resistance,[2,3,7,8] spasm-free resistance,[2,7] reactive tissue stiffness,[8] thicker tissue stiffness,[8] and stiffness of the movement.[2,3,8,9] Unfortunately, these variables have not been clearly defined.

Manipulative therapists often describe assessing "stiffness," a term that appears in the osteopathic,[10] medical,[11] chiropractic,[9] and physical therapy literature.[2,3,8] Within the physical therapy literature, the term "stiffness" has been used to describe the results of the posteroanterior (PA) central pressure test.[2,3,8] This manual examination procedure involves the application of an anteriorly directed force over the spinous process of the prone patient.[3] Maitland[3] suggests two methods for performing this test; in the first method, the thumb tips are used as the point of contact, whereas in the second method, the area of the ulnar border of the hand between the pisiform and the hook of the hamate is the contact point. With both techniques, the force is produced with arm and trunk movement, but the names given to the technique (thumb and pisiform) reflect the contact point.

Although the collection of information on spinal stiffness continues to be emphasized in the manipulative physical therapy literature,[8] there is some doubt about the reliability of clinicians' stiffness judgments. In a recent study,[12] we confirmed the observations of Matyas and Bach[13] in finding that physical therapists' judgments of stiffness have only poor-to-fair intertherapist reliability. In an attempt to develop a more reliable clinical measure of stiffness, we have conducted a series of psychophysical studies investigating the perception of stiffness. By studying the perception of stiffness, it may be possible to identify a range of factors that affect stiffness perception, which, if controlled in the clinical environment, would enhance the reliability of clinical judgments of PA stiffness.

The starting point in this research program was to first establish whether physical therapists could reliably judge stiffness when it was not confounded by concurrent information on other dimensions. To investigate this issue, we developed a mechanical apparatus with which we could present calibrated metal springs for stiffness assessment (a purely unidimensional task). We then returned to the six clinicians who had participated in the original clinical reliability study,[12] but this time with metal springs as stiffness stimuli (rather than patients with low back problems) to be judged on the same -5 to +5 scale. Even though the subjects had only 10 minutes to acquaint themselves with the middle and extreme spring stiffness values (whereas they had from 8 to 21 years of experience in rating human spines), we found much more acceptable reliabilities, with an average intraclass correlation coefficient (type 2,1) value of .60. We also calculated Pearson product-moment correlation coefficients between the known physical values of the metal stiffness stimuli and the ratings applied for each therapist and found coefficients ranging from .71 to .91, with an average of .82.[14]

Having established that at least unidimensional stiffness stimuli could be rated reliably, in a second psychophysical study,[15] we set out to quantify stiffness discriminability in terms of a traditional psychophysical measure, the Weber fraction. The Weber fraction, or discrimination threshold, describes how different two stiffness stimuli must be before one stimulus can be consistently discriminated as noticeably different from the other. In that study, we found that the Weber fraction was 11% when the pisiform grip was used to assess stiffness stimuli provided by the same mechanical apparatus used in the first psychophysical study. Our confidence in this estimate of the Weber fraction has been further enhanced by a review of studies that have evaluated the Weber friction from stiffness judgments on other materials.[16] The review identified 12 studies that investigated the stiffness discrimination threshold and noted a range of estimates for the Weber fraction from 2% to 32%, with an average value of 12.5%.[16] Having convinced ourselves that physical therapists can meaningfully judge unidimensional mechanical stiffness, we returned to investigating why this ability does not seem to be transferred to the clinical assessment of PA spinal stiffness.

A critical issue when seeking to develop a more reliable clinical protocol for assessing PA spinal stiffness would be to establish whether both versions of the PA pressure test are equally efficient for collecting stiffness information. Maitland[3] suggests that this may not be the case. Maitland advises that when stronger PA pressure is required in the lumbar spine, the pisiform grip is preferred over the thumb grip because "...the physiotherapist loses the degree of feel that she should have."[3(p282)] This suggestion that there is, some loss of the ability to perceive stiffness in the lumbar spine with the thumb grip was presumably based on Maitland's clinical experience, as there have been no investigations of the relative stiffness discriminability for the two methods. There has, however, been some investigation of the effect of the method of testing stiffness stimuli on stiffness perception. Although these psychophysical studies have shown that the method of handling stimuli can affect the ability to discriminate stiffness,[17-19] none of these studies used a paradigm that simulated the clinical application of the PA pressure test; therefore, they do not help evaluate Maitland's statement.

Our study sought to investigate Maitland's suggestion by using a mechanical device to provide stiffness stimuli and signal detection theory (SDT) methods to quantify performance with the two versions of the PA central pressure test. Signal detection theory was originally proposed to account for the performance of human operators attempting to detect signals near threshold.[20] In such a case, the detection problem may be represented as making decisions as to whether an event was a true signal or a sample from the randomly fluctuating background noise ever-present in neural systems. An example of a signal detection task[21] that of the sonar operator who must make a judgment about whether the sound he or she hears is a submarine or a school of fish. One could consider the school of fish as noise that is sometimes less intense, sometimes very close to the signal provided by the submarine. In such a case, the operator can set various criteria for responding to the sound, based on the costs and payoffs associated with the various possible outcomes and the likelihood that an enemy submarine is in the area.

To characterize the operator's performance under a range of criteria, receiver operating characteristic (ROC) curves are constructed, plotting the probability of correctly identifying the signal (a hit) against the probability of falsely identifying the signal (a false alarm) for each criterion adopted by the operator. The advantage of SDT methods over classical psychophysical methods is that they provide an estimate of the subject's ability to discriminate between stimuli, which is independent of the criterion, or the subject's bias to respond in a particular way, which the subject adopts.[22] One of the important premises of the theory is that an operator's sensitivity to signals on a particular dimension is not related to his or her willingness to respond "signal."

Welford[23] has described an extension of SDT methods to the discrimination situation involving more than two signals in which, for each stimulus pair the operator is required to discriminate, the less intense stimulus can be regarded as "noise." In our study, we used the category-rating experimental method,[20] which is used to generate estimates of performance, to evaluate the subjects' performance when discriminating five stiffness stimuli with both versions of the PA central pressure test. This SDT method requires subjects to judge the magnitude of stimuli on a categorical scale, with these category boundaries serving as the criteria (with their associated hit and false-alarm rates) for constructing ROC curves. These curves were constructed to characterize the ability of the subjects to discriminate the stiffer spring from the less stiff spring for the comparison of spring I versus spring 2, spring 2 versus spring 3, and so on. For each comparison, the less stiff spring is regarded as the noise and the more stiff spring is regarded as the signal.

The primary purpose of our study was to determine whether there was any difference in performance with the two versions of the PA central pressure test, either in the subjects' ability to discriminate between stiffness stimuli when using the particular test version or in their bias to rate a stimulus as stiffer when using a particular version.

Method

Subjects

Twenty subjects took part in the study, and their characteristics are given in the Table. All subjects gave informed consent prior, to participation in the study.

Table.
Description of the 20 Subjects Who Took Part in the Study

Factor                          x[bar]    SD       Range
Height (cm)                     169.7     9.5      154-192
Weight (kg)                      69.5    12.8       53-104
Age (y)                          33.3     9.7       19-55
Physical therapy experience (y)  10.5     9.8        0-34

Stiffness Stimuli

The five stiffness stimuli were provided by a mechanical device. The point of contact for the subject's hand was on a metal lever, which was free to rotate noiselessly about a bearing at one end. The lever rested on a compression spring at the free end. The spring was mounted in the base and provided the resistance to downward movement of the lever. The stiffness of the downward movement at the point of contact could be altered by changing the position of the spring in the mounting base plate, thereby altering the effective lever arm the subject used. Prior to data collection, the device was calibrated using weights of known mass to provide the force. A dial indicator (Mitutoyo model 2050F(*)) was used to measure the resultant displacement of the point of contact. The stiffness of the movement was then determined for each of the five positions that the spring could take along the base plate. This arrangement allowed the experimenters to present five stiffness values to the subjects (11.91 N/mm, 12.54 N/mm, 13.19 N/mm, 13.86 N/mm, and 14.55 N/mm).

The five stiffness stimuli generated for the study were chosen so that they would be similar to PA stiffness values that may be observed in the thoracic or lumbar spine. Data for spinal PA stiffness were extracted from three studies by Lee and colleagues,[24-26] who reported that the mean PA stiffness was 13.6 N/mm (SD = 3.7) at T-4, 14.5 N/mm (SD = 3.5) at T-5, and 14.02 N/mm (SD=4.02) at L-3.

An important feature of a discrimination study is that the stimuli used be somewhat confusable, so that misclassifications will occur and the effects of factors thought to influence discrimination can be assessed. In contrast, in a reliability study, the stimuli are chosen to represent the range across which measurements are normally made. By a trial and error process, we selected a set of five stiffness stimuli where adjacent stimuli were difficult, though not impossible, to discriminate.

Scaling Method

Subjects made judgments of the perceived magnittide of the stiffness stimuli on a 1 to 5, low to high stiffness scale. Prior to each data collection session, the subjects familiarized themselves with the five stimuli they would be judging in the trial by pushing on the stimuli using both versions of the PA pressure test. As the subjects pushed on the lever for each stimulus, they were informed of its position in the series. Subjects were requested to push on the upper and lower levels of stiffness five times, they descended and ascended through the series of stiffness levels, and then they were allowed to practice the extreme stiffness levels once again. This exposure to the series of stimuli took approximately 5 minutes.

Data Collection

Data collection took place immediately after the introductory session. Subjects were requested to push down on the stiffness device only once or twice using either the thumb grip or pisiform grip, as indicated by the investigator (CM), and to then make judgments of the stiffness magnitude. The number reported was then recorded. Within one testing session, the five stiffness levels were presented on 20 occasions, with the order of presentation of the five stimuli and two grip types randomized. Each data collection session took approximately 15 minutes. During a testing session, subjects were given a 2-minute break halfway through the session. All subject wore goggles during data collection so that they could not see the spring position. Subjects returned for a second session so that each subject tested each of the five stimuli 40 times (20 times with each grip). Two subjects could not attend the second data collection session. Data for these subjects are based on one session only.

Data Analysis

The number of times a subject used a response category for each stimulus level and grip type was determined through simple collation of the responses on the raw data sheets. These data were cast into a 10X5 confusion matrix, with rows 1 to 5 containing the responses for the occasions when the thumb grip was used for stiffness stimuli 1 to 5 and rows 6 to 10 containing the responses when the pisiform grip was used for stimuli I to 5. Columns I to 5 represent the five response categories. If any two rows of this matrix are isolated, the cell entries can be converted to cumulative probabilities and an ROC curve drawn. The proportion of the total area of the graph beneath the ROC curve, P(A), is a measure of a subject's ability to distinguish between the stimuli corresponding to the two rows[20] and will be 0.5 if two stimuli are judged as being the same. In this way, measurements of stimulus discriminability were obtained for the four stimulus comparisons (ie, I versus 2, 2 versus 3, 3 versus 4, and 4 versus 5) for each grip type.

We evaluated the bias due to test type in two ways. The first measure of bias was obtained by comparing the group mean of the category ratings of perceived stiffness for each grip type and stimulus level. If no bias was present, the perceived magnitude of a given stimulus should be the same for both grip types.

The second measure of bias was obtained for each individual by calculating P(A)s from ROC curves constructed using the rows from the confusion matrix corresponding to the same stimulus (ie, when the stimulus was tested with the thumb grip and the pisiform grip). As mentioned previously, P(A) will be 0.5 if two stimuli arejudged the samc; thus, a measure of bias can be obtained by subtracting 0.5 from the P(a)s obtained. A IP(A) - 0.51 value of 0 represents no bias (ie, when the same stimulus isjudged with the two versions of the PA pressure test, on average the stimulus should feel the same with both grips). The maximum value of IP(A) - 0.51 is 0.5, and in such a situation one version of the test will always make the stimulus feel stiffer in comparison to when it was judged with the other grip type.

The stimulus comparison was a repeated-measures variable (ie, all subjccts experienced all four stimulus comparisons), so contrasts for trend(27) were written across the four stiffness comparisons to test for the presence of different trend components relating stimulus discriminability, as measured by P(A), to stimulus comparison. An example of such a trend component would be if it became harder to discriminate between stimuli as absolute stiffness increased. This test was done within an analysis-of-variance framework, which also examined the effects of grip type on stimulus discriminability. Because P(A) values tend to be negatively skewed in their distribution, they were transformed by the suggested [2 arcsin-[sqaure root]P(A)]/[pi] conversion prior to analysis.[28,29]) A similar analysis of trend also was performed on the perceived magnitude data to ascertain whether any bias to rate a stimulus as stiffer induced by a particular grip type became greater or lesser as stimulus stiffness increased.

Results

The mean and 95% confidence intervals for P(A) values for the four stimuli comparisons and two grip types are shown in Figure 1. Inspection of the graph suggests that stimulus discriminability is very similar across all the four pair-wise stimulus comparisons and across both grip types. This similarity was confirmed by the analysis of variance, which showed that there was no difference in stimulus discriminability between the two grip types (F=1.16; df=1,19). Further, the analysis of trend showed that there was no difference in stimulus discriminability across the four stimulus comparisons in terms of linear or quadratic trends F= <1.00; df= 1,19). Inspection of Figure I reveals that in all cases the P(A) 95% confidence intervals do not include 0.5. Thus, it can be concluded that the subjects are able to discriminate the stiffness stimuli even though the intervals between stimuli were quite small (of the order of 5%), but that neither the grip they adopted nor the overall magnitude of the stiffness stimuli influenced their ability to discriminate.

The group mean values for perceived stiffness magnitude are shown in Figure 2. For each of the five stimuli, the perceived magnitude was greater when tested with the thumb-grip method that when tested with the pisiform-grip version. The proportionate increase in perceived magnitude resulting from using the thumb grip ranged from 33% for stimulus I to 16% for stimulus 4. Analysis of variance showed a difference in perceived magnitude between the two grip types (F= 42.06; df = 1,19). The analysis of trend showed the expected linear increase in perceived stiffness magnitude as physical stiffness increased (F=511.4; df=1,19), but there was no interaction between linear trend and grip type (F= 1.86; df = 1,19).

The extent of the PA pressure illusion can be seen clearly from the histogram depicted in Figure 3. Only 1 subject out of the 20 tested did not display a bias toward rating the stimuli as stiffer when using the thumb grip.

Discussion

The results of our study demonstrate that the two versions of the PA pressure test do not provide equivalent information on stiffness. Although the two versions of the test result in equal stimulus discriminability, the use of two grip types produces an illusion such that when the same stimulus is judged with the thumb-grip method, it usually is judged as markedly stiffer than when judged with the pisiform-grip method. This finding of a substantial bias associated with the thumb-grip version of the PA central pressure test is consistent with the clinical observation of Maitland,[3] who warned that a degree of feel is lost with the thumb-grip version of the test. This "loss of feel" appears to be a bias in perceived magnitude, not reduced stimulus discriminability.

We must emphasize that these results may apply only to the stiffness values used in the study. Although these values were chosen to be typical of the mean PA stiffness values observed in the thoracic or lumbar spine, further research across a larger range of stiffness stimuli is needed, given that PA stiffness values may range from 5.4 to 27 N/mm in human subjects.[24-26] We also must emphasize that although this study demonstrated equal stiffness discriminability for the two grip types, it did not address the discriminability for other factors that manipulative physical therapists may judge with the test (eg, "spasm-free resistance"[7]).

The results of our study add to a body of knowledge from the existing psychophysical literature that identifies a range of factors that affect stiffness perception. Factors such as stiffness magnitude,[19,30-32] the method of handling stiffness stimuli,[17-19] surrounding stimuli in the series and stimuli experienced in the past,[18,33,34] stimuli dimensions,[18] temperature,[18,35] and patterns of attention[35] have all been shown to affect stiffness perception with materials other than backs. As yet, this information has not been assimilated into the manipulative therapy body of knowledge. This is unfortunate because the psychophysical literature suggests many approaches to refining the clinical protocol for rating stiffness that may well enhance the reliability of stiffness judgments. In 1952, for example, Sheppard and Scott-Blair[18] showed that testing a stiffness stimulus twice provides better discrimination than a test with three trials, which was in turn better than testing for only one trial.

Until recently, little consideration of the source of the sensory signals responsible for the perception of stiffness had taken place. One of the first proposals was that of jones and Hunter,[31] who speculated that because the mechanical parameter of stiffness is a function of force and displacement, the sensafion of stiffness would be derived from the integration of sensory signals conveying force and displacement information. Because sensations of force are primarily derived from the corollary discharge of the descending efferent command[36] and sensations of displacement are primarily derived from signals arising from the receptors in the muscle spindle,[37] Jones and Hunter emphasized both muscle spindles and corollary discharge as sources of sensory signals for stiffness perception. Recent work by Srinivasan and Lamotte[38] and Tan et al,[39] however, has shown that under certain situations, subjects also can judge stiffness from cues on total work performed and from cutaneous cues such as surface pressure/contact area. These two studies demonstrate that there are multiple means of computing and discriminating the stiffness of compliant objects and that both the conditions of stiffness assessment and the surface of the object are important determinants of which cues are used, and therefore of the resolution of the system.

The bias associated with the use of the thumb-tips method was both large and consistently observed. Of the 20 subjects tested, all except I subject exhibited a bias, which was on average of the order of 25%. This ability to consistently generate quite substantial illusions of perceived stiffness magnitude has relevance for the study of the perception of stiffness. Neurophysiologists have made use of illusions involving distortions in the perception of limb position, movement, and weight to reveal the sensory processes involved in proprioception.[40] In the 1950s and 1960s, for example, joint receptors were thought to be the principal receptors in kinesthesia.[37] Evidence for the role of muscle spindles in kinesthesia, however, arose when neurophysiologists[41] noted powerful illusions of both limb position and movement induced by vibration of muscles. A similar approach may help reveal the mechanisms involved in the perception of stiffness. This information also would have relevance outside of manipulative physical therapy because many everyday manual tasks (eg, judging the ripeness of fruit) require individuals to judge stiffness.

Presuming that the bias is due to some difference in performance of the test, two potential explanations for the bias are apparent. The first explanation is that the bias is a result of a difference in the area of contact between the two versions of the test (ie, with the thumb-grip version the area of contact is smaller and thus the contact pressure is greater), If cutaneous pressure is involved in signaling stiffness, as suggested by Srinivasan and LaMotte,[38] this provides an explanation for the bias. The second suggestion is that with the thumb-tips method of PA pressure testing, muscles of the thumb and hand are required to work as stabilizers for performance of the task, whereas this is not the case with the pisiform method. Kilbreath and Gandevia[42] have shown that the perceived heaviness of a weight lifted by one muscle is increased when another related muscle also is required to work. These authors suggested that the bias in perceived heaviness could be due to a failure of the central nervous system to completely partition the destination of motor commands. If the sensation of stiffness is similar to the sensation of heaviness and is mediated by the magnitude of central motor commands,[31] then this difference in the performance of the task would predict the observed bias in perceived stiffness. Although both explanations are plausible, they await investigation.

Clinical Implications

The results of our study demonstrate that standardization of the testing method is a prerequisite of a reliable protocol for rating PA stiffness in the clinic. Our results suggest that therapists should ensure that they use the same version of the PA central pressure test each time they examine a patient's spine. In the thoracolumbar region, where both versions of the test are possible, therapists who switch from one technique to the other may produce misleading estimates of PA stiffness that interfere with clinical decision making. We suggest that when recording the results of the PA pressure test, clinicians should note which version of the test was used. Given that the two versions of the test have equal stiffness discriminability, clinicians should adopt the version of the test that is more comfortable for the patient or themselves.

Conclusion

The results of this investigation demonstrate that the two methods of the PA central pressure test, although equally sensitive to changes in physical stiffness, do not provide similar information on stiffness to the therapist. Although the two tests have equal stimulus discriminability, the thumbtips method produces a substantial bias, making stimuli appear stiffer than when tested with the pisiform-grip method.

[Figures 1 to 3: ILLUSTRATION OMITTED]

(*) Mitutoyo Corp, 31-19 Shiba5-chome, Minato-ku, Tokvo 108, Japan.

References

[1] Magarey M. Examination and assessment in spinal joint dysfunction. In: Grieve G, ed. Modern Manual Therapy of the Vertebral Column. Edinburgh, Scotland: Churchill Livingstone Ltd; 1984:481-497. [2 ] Grieve G. Mobilisation of the Spine: Votes on Examination, Assessment and Clinical Method. 4th ed. Edinburgh, Scotland: Churchill Livingstone Ltd; 1984. [3] Maitland GD. Vertebral Manipulation. 5th ed. London, England: Butterworth Co (Publishers) Ltd; 1986. [4] Corrigan B, Maitland GD. Practical Orthopaedic Medicine. London, England; Butterworth & Co (Publishers) Ltd; 1983. [5] Cyriax J. Textbook of Orthopaedic Medicine, Volume 1: Diagnosis of Soft lissue Lesions. 7th ed. London, England: Bailliere Tindall; 1978. [6] Fave L, Wiles M. Manual examination of the spine. In: Haideman S, ed. Principles and Practice of Chiroptactic. East Norwalk, Conn: Appleton & Lange; 1992:301-318. [7] Hickling J, Maitland G. Abnormalities in passive movement: diagrammatic representation. Physiotherapy. 1970;56:105-114. [8] Jull G, Treleaven J, Versace G. Manual examination: Is pain provocation a major cue for spinal dysfunction? Australian Journal of Physiotherapy. 1994; 40:159-165. [9] Nathan M, Keller T. Measurement and analysis of the in vivo posteroanterior impulse response of the human thoracolumbar spine: a feasibility study. J Manipulative Physiol Ther. 1994; 17:431- 441. [10] Stoddard A. Manual of Osteapathic Practice. 2nd ed. London, England: Century Hutchinson Publishing Ltd; 1983. [11] Mennell JM. Back Pain: Diagnosis and Treatment Using Manipulative Techniques. Boston, Mass: Little, Brown and Co; 1960. [12] Maher C, Adams R. Reliability of pain and stiffness assessments in clinical manual lumbar spine examination. Phys Ther. 1994;74:801-811. [13] Matyas T, Bach T. The reliability of selected techniques in clinical arthrometrics. Australian Journal of Physiotherapy. 1985;31:175-199. [14] Maher C, Adams R. Is the clinical concept of spinal stiffness multidimensional? Phys Ther. 1995;75:854-864. [15] Maher C, Adams R. A psychophysical evaluation of manual stiffness discrimination. Australian Journal of Physiotherapy. 1995;41:161-167. [16] Maher C. Perception of stiffness in manipulative physiotherapy. Physiotherapy Theory and Practice. 1995;11:35-44. [17] Scott-Blair GNO, Coppen FMV. The subjective judgements of the elastic and plastic properties of soft bodies: the "differential thresholds" for viscosities and compression moduli. Proceedings of the Royal Society. 1939;128B: 109-125. [18] Sheppard D, Scott-Blair GW. Subjective assessments of firmness. Qjexp Psychol 1952;4:101-108. [19] Mioche L, Auroy P, Lepefit J, Compagnon D. Oral perception of hardness in viscoelastic products. J Texture Studies. 199 1;22:333-347. [20] Green D, Swets J. Signal Detection Theory and Psychaphysics. 1988 reprint edition. Los Altos, Calif: Peninsula Publishing; 1988. [21] McBurney D, Collings V. Introduction to Sensation/Perception. Englewood Cliffs, NJ: Prentice-Hall Inc; 1977. [22] Egan J, Clarke F. Psychophysics and signal detection. In: Sidowski J, ed. Exp mental Methods and Instrumentation in Psychology. New York, NY. McGraw-Hill Book Co; 1966:209-246. [23] Welford A. Skilled Performance: Perceptual and Motor Skills. Glenview, Ill: Scott Foresman & Co; 1976. [24] Lee M, Latimer J, Maher C. Manipulation: investigation of a proposed mechanism. Clin Biomech. 1993;8:302-306. [25] Lee M, Esler M, Mildren J, Herbert R. Effect of extensor muscle activation on Lhe response to lumbar postcroanterior forces. Clin Biomech. 1993;8:115-119. [26] Lee M, Lau H, Lau T. Sagittal-plane rotation of the pelvis during lumbar posteroanterior loading. J Manipulative Physiol Ther. 1994;1 7:149-155. [27] Winer B. Statistical Principles in Experimental Design. New York, NY: McGraw-Hill Inc; 1971. [28] McNicol D. A Primer of Signal Detection Theory. London, England: George Allen & Unwin; 1972. [29] Adams R. Intervening stimulus effects on category judgements of duration. Perception and Psychaphysics. 1977;21:527-534. [30] Coppen F. The differential threshold for the subjective judgement of the elastic and plastic properties of soft bodies. Br J Psychol. 1942;32:231-247. [31] Jones LA, Hunter IW. A perceptual analysis of stiffness. Exp Brain Res. 1990;79:150-156. [32] Harper R. Psychological and psychophysical aspects of studies of craftsmanship in dairying. Brjpsychol. 1952;28:1-63. Monograph. [33] Sheppard D. Subjective assessments of firmness: the use of a rating scale. Q.J Exp Psychol. 1953;5:1-9. [34] Sheppard D. The importance of psychophysical errors in subjective judgements of firmness of Cheshire cheese. J Daily Res. 1952; 19:348-355. [35] Scott-Blair GIA. The subjective assessment of firmness. Laboratory Practice. 1954;3:3-9. [36] Jones LA. Perception of force and weight: theory and research. Psychol Bull. 1986;100:29-42. [37] Proske U, Schaible H, Schmidt R. joint receptors and kinaesthesia. Fxp Brain Res. 1988;72:219-224. [38] Srinivasan M, Lamotte R. Tactual discrimination of softness. J Neurophysiol. 1995;73:88-101. [39] Tan H, Durlach N, Beauregard G, Srinivasan M. Manual discrimination of compliance using active pinch grasp: tbe roles of force and work cues. Perception and Psychophysics. 1995;57:495-510. [40] Jones IA. Motor illusions: What do they reveal about proprioception? Psychol Bull. 1988;103:72-86. [41] Goodwin G, McCloskev D, Mathews P. The contribution of muscle afferents to kinaesthesia shown by vibration induced illusions of movement and by the effects of paralysing joint afferents. Brain. 1972;95:705-748. [42] Kilbreath SL, Gandevia SC. Independent control of the digits: changes in perceived heaviness over a wide range of force. Exp Brain Res. 1992;91:539-542.

C Maher, PT, is Lecturer, School of Physiotherapy, Faculty of Health Sciences, The University of Sydney, East St, POB 170, Lidcombe, New South Wales, Australia 2141 ([e-mail]). Address all correspondence to Mr Maher.

R Adams, PhD, is Senior Lecturer, School of Physiotherapy, Faculty of Health Sciences, The University of Sydney.

This study was approved by The University of Sydney Human Ethics Committee.

This article was submitted December 29, 1994, and was accepted October 4, 1995.