INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Studies of the forces employed in object manipulation using a precision grip have found that the grip forces are optimally sufficient to prevent accidental slips and yet are not so excessive as to crush a fragile object or to cause muscle fatigue (Johansson and Westling 1984a; Westling and Johansson 1984). These investigations suggested that in precision grasping, the pinch force and its rate of application were determined by the anticipated weight and friction of the object. It was also shown that the grip force was affected by the margin of safety, set by the individual and based on prior experience. A critical supporting observation was that cutaneous anesthesia of the thumb and index finger disrupted the coordination between the grip and lifting forces that was normally adapted to the friction between the object and the skin, resulting in slips especially with relatively heavier objects presenting low-friction surfaces (Johansson and Westling 1984a; Westling and Johansson 1984). Furthermore, Collins et al. (1999) showed that cutaneous afferents signaling contact with a target object play an important role in triggering and regulating the finger muscle activity during grasping. Loss of tactile sensation has also been associated with a significant increase in the overall grip force applied to the object both during grasping and lifting as well as during static holding.

Although initially the appropriate application of grasping and lifting forces to a novel object depends on feedback from cutaneous afferents of the hand, further familiarity gained through manipulatory experience contributes to the formation of a memory trace or internal model of the physical properties of the grasped object (Gordon et al. 1991, 1993; Johansson and Westling 1984b, 1988). The acquired internal model not only contributes to the preprogramming of grip and lifting forces, but also to the changes in grip forces anticipating the tangential forces on the skin caused by the changes in acceleration occurring during oscillatory movements of a hand-held object (Flanagan and Wing 1993, 1997).

It might be expected that once acquired, the internal model based on knowledge of an object's inertial and frictional properties would compensate for any loss of cutaneous sensation from local anesthesia (Jenmalm and Johansson 1997; Johansson and Westling 1984a). However, it seems that when subjects are asked to grasp a familiar object with anesthetized fingers, the grip forces were consistently excessive despite extensive prior experience with the object (Häger-Ross and Johansson 1996; Jenmalm and Johansson 1997; Johansson and Westling 1984a; Johansson et al. 1992; Westling and Johansson 1984). Certainly, the absence of a sense of friction explains why grip forces are higher during the first few lifts of a familiar object and any additional uncertainty about slips induced by changes in acceleration would further encourage the use of greater grip forces. As further evidence of the CNS's capacity to compensate for the loss of cutaneous feedback, Johansson and Westling (1984a) reported that when subjects alternatively lifted the same object with an anesthetized and un-anesthetized hand, the grip force in the insensate hand was influenced by the experience of the sensate hand. It would seem, therefore, that prior tactile experience helps to reduce excessive grip forces on subsequent trials without tactile feedback. That is, one could hypothesize that repeated practice with grasping the same object with anesthetized fingers should lead to some grip force reduction based on the internal model of the object's inertial properties. The objective of this study was to examine the capacity of an acquired internal model to compensate for the loss of normal cutaneous sensation after digital anesthesia.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Task

The subjects were seated comfortably with the right arm abducted at about 30°, the elbow flexed at about 90°, and the forearm resting on a firm, supporting surface. At a signal from the experimenter, the subjects were instructed to grasp a metal tab between the thumb and index finger and to lift it about 2.5 cm and maintain this position stationary for 4 s. Certain trial blocks were conducted without visual feedback, and although the subjects were not blindfolded, they were not able to turn the head such that peripheral vision of the arm was available. A 1.0-kHz tone signaled to the subject that the armature had been lifted to the desired height between 1.5 and 3.5 cm. At the conclusion of testing with intact sensation, anesthesia with mepivacaine of the thumb and index was achieved by a ring-block infiltration of the digital nerves around the metacarpal-interphalangeal joint at the base of each finger. Infiltration continued until all sensation in the two fingers was gone, indicated by complete insensitivity to skin contact with a camelhair brush and Semmes-Weinstein monofilaments. The loss of cutaneous sensation was repeatedly verified throughout testing. In general, the local anesthesia lasted for about 2 h, but in a few subjects, supplementary injections were required before the conclusion of testing.

Statistical analysis

In general, the data were evaluated using either t-tests for paired comparisons, correlation coefficients, or one or two-way analyses of variance. The peak force on each trial during the dynamic lifting phase was identified and averaged, and the forces and torques were averaged over the last 500 ms on each trial to obtain a mean value for the static phase.

Apparatus

The apparatus used in the present experiment was modified and improved after preliminary experiments suggested the presence of forces other than those involved with grasping and lifting. The device shown in Fig. 1 is the improved version of a similar apparatus used in an earlier study (Cadoret and Smith 1996). The one-degree-of-freedom armature was set in a compressed air bushing that allowed near frictionless movement in the vertical direction. The voltage applied to the coil provided a range of resistive forces opposing lifting. Flat rectangular grasping surfaces for the thumb and index finger were attached to a horizontal strut mounted at 90° to one end of the armature. A load cell measured the total compression or grip force between the index and thumb, and a second load cell mounted on the armature measured the vertical or lifting force on the armature.

View larger version (39K): [in this window] [in a new window]   Fig. 1. Linear motor equipped with load cells to measure the grasp and lifting forces and a 3-dimensional (3-D) force/torque sensor to measure off-axis forces and torques. Note inset showing the axes of force and torque measurement. x-axis torque is a rotational force in the sagittal plane, y-axis torque is a rotational force in the horizontal plane, and z-axis torque is a rotational force in the frontal plane.

A lightweight (9.4 g) six-axis force-and-torque sensor (Nano-6-axis force and torque transducer; ATI Industrial Automation, Garner, NC) was added to measure side to side (x axis), up and down (y axis), and pushing and pulling (z axis) forces. The three force traces were fed to a proprietary analog-to-digital converter with 16-bit precision at a conversion rate of 250 Hz, which was used to calculate the three torque components. The inset above the force transducer in Fig. 1 indicates the directions of the x, y, and z linear forces. The curved arrows indicate the planes about the x, y, and z axes of the clockwise or counterclockwise rotational forces that have been called the x, y, and z torques. With this arrangement, the vertical load cell essentially replicated the y-axis force measurement. However the length of the strut bearing the grasping tabs provided a significant lever arm and contributed to the x-axis torque shown in Fig. 6G, which will be discussed later. In contrast, the grip-force load cell measured the sum of the compression forces exerted by the thumb and index together, whereas the x-axis force output from the force and torque sensor measured the force differential between the two fingers. Taken together, these two measures made it possible to calculate the force exerted by the index and thumb separately.

Finally, two high-resolution (67 points/cm²), ultra-thin (0.1 mm), pressure-sensitive surfaces from Tekscan Pressure Measurement Systems were added to each of the finger pads to record the pressure distribution under the thumb and index finger at a frequency of 100 Hz. During stationary holding, the position of the fingers did not change, allowing the position of the center of pressure under each finger to be calculated with accuracy.

Subjects and parameters investigated

The ethics committee of the Faculty of Medicine of the Université de Montréal approved the experimental protocol, and a total of 20 right-handed subjects (17 women and 3 men) signed informed consent forms and volunteered to participate in this study. Experiment one focused on the effects of friction on grip forces applied before and after digital anesthesia. Three subjects (2 women, 1 man; ages 24-56 yr) performed 10-trial blocks of the lift-and-hold task with smooth metal and emery paper textures, with three resistive forces (0.5, 1.0, and 2.0 N), with and without visual feedback, and finally with and without digital anesthesia. Four subjects (4 women, ages 24-28 yr) performed the task with 25-trial blocks with the two textures, with a 1.5-N resistive force, and with and without visual and cutaneous feedback. The slip force, or point at which the object slipped from the grasp, was measured for each surface on a separate block of trials in which the subjects were required to first lift and hold the object and then gradually release until the resistive force caused it to slip from the grasp. The coefficient of friction was determined as the average ratio of the grip-to-load force at the moment of slip. The emery paper surface had an approximate coefficient of friction against the skin of about 1.74, whereas the smooth metal had an approximate coefficient of friction of about 0.73.

Experiment two focused on the forces exerted by the hand in a total of 10 additional subjects (9 women, 1 man; ages 24-49 yr) who were tested with a simpler protocol focusing on the effects of local anesthesia, vision, and resistive force. These subjects were first tested on all conditions with sensation intact and then later after blocking the digital nerves to the thumb and index finger with mepivacaine. Each condition of resistive force (0.5, 1.0, and 2.0 N) and visual feedback were presented in 25-trial blocks in a balanced order randomizing the sequence of resistive force and visual feedback conditions.

A third part of this study focused on different control conditions designed to provide a better understanding of the results. Three additional subjects provided unique data. The first was a healthy 25-yr-old man naïve to the purpose of the experiment who was asked to perform a series of lift-and-hold movements with his fingers deliberately misaligned on the grasp surface. The second was a healthy 26-yr-old woman who performed two blocks of 25 trials initially using a self-selected grip force and then later deliberately exerting a 9.0-N grip force that was similar to the mean force used by subjects with the fingers anesthetized. Finally, a third subject was a 50-yr-old woman suffering from a polysensory neuropathy of both cutaneous and proprioceptive afferents (Forget and Lamarre 1987, Simoneau et al. 1999).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiment one

In the initial study, seven subjects were asked to perform repeated grasp, lift, and hold sequences against four resistive forces (0.5, 1.0, 1.5, and 2.0 N) and two surface textures (emery paper and smooth metal). Table 1 shows the average grip forces applied during the static holding. Two 2-way analyses of variance (ANOVA) were used to test the effects of resistive force, texture, and anesthesia on peak grip force, and two additional analyses of variance were used to test the effects of resistive force, texture, and anesthesia on mean static grip force. A 2-way ANOVA of the effects of resistive force [F(1,3) = 147.696, P < 0.001] and digital anesthesia [F(1,3) = 226.143, P < 0.001] found both variables to have a significant effect on peak grip force. A similar 2-way ANOVA for the effects of texture [F(1,1) = 109.497, P < 0.001] and digital anesthesia [F(1,1) = 206.550, P < 0.001] found both variables also to have a significant effect on peak grip force. Identical 2-way ANOVAs were used to evaluate the impact of resistive force, texture, and anesthesia on mean static grip force. Resistive force [F(1,3) = 160.068, P < 0.001] and anesthesia [F(1,3) = 344.720, P < 0.001] were significant main effects, and in a separate 2-way ANOVA, texture [F(1,1) = 103.662, P < 0.001] and anesthesia [F(1,1) = 269.331, P < 0.001] were found to be significant main effects.

The mean grip forces for 10 trials lifting three opposing forces are illustrated for one subject in Fig. 2A. With the fingers anesthetized, the grip force increased substantially. However, even without cutaneous feedback, this subject applied a greater mean grip force when lifting a greater resistive force, suggesting that some awareness of the required lifting force must have been available to the subject, probably from proprioceptive feedback, since no difference was found with or without visual feedback. Paradoxically, despite substantially increased grip force after digital anesthesia, most subjects initially dropped the object due to the inappropriate coordination of the grasping and lifting forces during the first few trials, particularly with the highest resistive force. No doubt these unrecoverable slips would have occurred even more frequently if we had used resistive forces equivalent to the 400- to 800-g weights used in most similar studies.

View larger version (24K): [in this window] [in a new window]   Fig. 2. A: effect of digital anesthesia on the manipulation of 3 different resistive forces simulating object weight for a single subject. The hatched line represents the mean grip force ±SD for 20 trials with cutaneous sensation intact, and the solid line represents the 20-trial average grip force ±SD after digital anesthesia. B: effect of digital anesthesia on the manipulation of 2 conditions of surface friction for a single subject. The hatched line ±SD represents the mean grip force for 20 trials with cutaneous sensation intact, and the thick line ±SD represents the mean grip force after digital anesthesia.

The same seven subjects were asked to grasp, lift, and hold the uncovered smooth metal grasping tabs or the same tabs covered with emery paper, contacting the fingers. The local anesthesia resulted in a substantial increase in both the dynamic and static grip force for all subjects. Figure 2B shows the increase in mean grip force after digital anesthesia for a single subject lifting a 1.5-N resistance. The grip force increase was greater for the smooth metal surface than for the emery paper surface.

A 2-way ANOVA found that both surface texture [F(1,1) = 6.12, P < 0.014] and digital anesthesia [F(1,1) = 134.49, P < 0.001] significantly affected the latency between the grip force onset and the initiation of the load force. Both these observations are in agreement with earlier studies by Johansson and Westling (1984a). Figure 3A illustrates the 10-trial mean grip and load forces employed to lift a smooth metal object offering a 1.0-N resistance with and without digital anesthesia for a single subject. Figure 3B shows a single lifting trial after digital anesthesia. In both parts of Fig. 3, the traces have been aligned on the onset of grip force. Without cutaneous feedback, the onset of lifting was delayed by several hundred milliseconds.

View larger version (18K): [in this window] [in a new window]   Fig. 3. Effect of digital anesthesia on the synchronization of grasping and lifting. Initiation of the lifting or load force is delayed by more than 100 ms after digital anesthesia. Surface = smooth metal, resistive force = 1.0 N.

Experiment two

DIGITAL ANESTHESIA AND THE PREPROGRAMMING OF GRIP AND LIFTING FORCES. The grip forces used to lift three resistances of 0.5, 1.0, and 2.0 N were tested in 10 additional subjects. As in experiment one, the onset of lifting was significantly delayed without cutaneous feedback. A 2-way ANOVA indicated that both resistive force [F(1,2) = 231.00, P < 0.001] and anesthesia [F(1,2) = 783.25, P < 0.001] were significant factors in determining the static grip force. The interaction was not significant. Table 2 shows that the static grip forces after anesthesia were significantly greater than with intact sensation. The smaller load forces of 0.5 and 1.0 N produced force and torque patterns similar to the 2.0-N load force, although they were of smaller magnitude. The patterns were the same both with intact tactile sensation and after digital anesthesia, and for this reason, they will not be described in further detail.

View larger version (35K): [in this window] [in a new window]   Fig. 4. A: force traces from a series of 5 consecutive lifts for each of the 3 resistive forces (0.5, 1.0, and 2.0 N) for a single subject with intact cutaneous sensation. Movement, grip, and load forces and their derivatives are shown on the left. Grip and load force correlation and phase planes for the grip and load force rates are shown on the right. B: comparable force traces from a series of 5 consecutive lifts for each of the 3 resistive forces (0.5, 1.0, and 2.0 N) for the same subject after digital anesthesia are shown on the right.

DIGITAL ANESTHESIA AND ADDITIONAL FORCES AND TORQUES. In general, with tactile sensation intact, all subjects applied forces relatively efficiently during lifting and holding with minimal force expenditure in directions other than grasping or lifting (i.e., relatively small off-axis forces). The single exception was the x-axis torque noted in METHODS (seen in Fig. 6G) that represented a mechanical characteristic of the apparatus due to the 6.4- to 7.0-cm lever arm, dependent on the position of the fingers rather than a feature of grip force application by the subject. This x-axis torque was unaffected by digital anesthesia in all subjects as shown by a 2-way ANOVA [F(1,2) = 0.22, P > 0.64].

View larger version (7K): [in this window] [in a new window]   Fig. 5. A: the mean pressure exerted by the thumb and index for a single subject with sensation intact. B: an overall increase in finger pressure after digital anesthesia, as well as an increased imbalance between the 2 finger forces.

View larger version (20K): [in this window] [in a new window]   Fig. 6. Effect of digital anesthesia on lifting a 2.0 N resistive force for the same subject shown in Fig. 5. The thin line represents the mean grip force for 20 trials with cutaneous sensation intact, and the thick line represents the 20-trial average grip force after digital anesthesia. A-C: displacement, grip force, and load force. D-F: linear forces along the x, y, and z axes indicated in Fig. 1. An upward deflection indicated increased thumb pressure in D, the lifting force in E, and a pulling force in F. G-I: torques around the x, y, and z axes and clockwise torque are shown as a positive deflection.

FINGER MISALIGNMENT AND Y- AND Z-AXES TORQUES. Tekscan high-resolution, pressure-sensitive surfaces were used to provide evidence of the misalignment of the fingertips after digital anesthesia. Figure 7 illustrates a three-dimensional (3-D) reconstruction of the area of contact pressure under the thumb and index finger during 500 ms of static holding of a 2.0-N resistive load during three trials with tactile sensation intact and after digital anesthesia. The upper portion of Fig. 7 illustrates the finger-pressure cones exerted prior to the injection of local anesthesia during three separate trials. The increased finger pressure after digital anesthesia is indicated by the greater height of these pressure cones (Fig. 7, middle). Software algorithms from Tekscan were used to determine the precise horizontal and vertical coordinates of the center of pressure for each finger during stationary holding. The difference between the centers of pressure of the thumb and index finger in the horizontal and vertical planes determined the lever arm contributing to the y-axis and z-axis torques, respectively.

View larger version (47K): [in this window] [in a new window]   Fig. 7. A 3-D reconstruction of the pressure distributions exerted by the thumb and index finger during steady-state holding of a 2.0-N resistive force on 3 trials with tactile sensation intact (top row), after digital anesthesia (middle row), and by a patient with no conscious cutaneous or proprioceptive sensation (bottom row).

TACTILE ANESTHESIA AND FINGER ALIGNMENT. The horizontal and vertical misalignment of the two centers of pressure was noted for all subjects before and after digital anesthesia. Correlation coefficients were calculated between the alignment errors and the mean steady-state y- and z-axes torques for the first and last five trials in each condition, with and without the digital anesthesia. Table 4 shows the correlations between the horizontal alignment distance and the y-axis torque and the vertical alignment distance and the z-axis torque. These correlations ranged from r = 0.14 (n = 20, not significant) to r = 0.95 (n = 20, P < 0.001). However, a significant correlation was found for at least one of the two axes for all 10 subjects, indicating that, in general, the finger misalignment significantly contributed to the y- and z-axes torques. The y- and z-axes torques together would have produced a resultant torque (T_r) on the fingers calculated as
Table 4 also shows that the mean correlation between this resultant torque and grip force for all subjects was generally quite strong (r = 0.79; range, 0.46-0.98).

View larger version (12K): [in this window] [in a new window]   Fig. 8. Mean (20 trials) offset distance between the fingers for each of the 10 subjects with tactile sensation intact and after digital anesthesia. For 7 of 10 subjects, the difference was statistically significant (*P < 0.05).

EFFECT OF DELIBERATELY MISALIGNED FINGER PRESSURE. To substantiate the causal link between finger alignment and the y- and z-axes torques, we asked an additional naïve male subject with cutaneous sensation intact to perform a series of 10 lifts against a 2.0-N resistive force with the fingers aligned and then deliberately misaligned by about 20 mm in either the horizontal or vertical directions. Figure 9A shows that when the fingers were accurately aligned, there was practically no z-axis torque, and the y-axis torque was rather small. In contrast, Fig. 9B shows that when the fingers were misaligned in the vertical direction, a significant z-axis torque appeared with the same small amount of y-axis torque. Figure 9C shows that misalignment in the horizontal direction caused a significant y-axis torque with negligible z-axis torque. Although not illustrated, the deliberate finger misalignment was also accompanied by a significant increase in the grip force.

View larger version (17K): [in this window] [in a new window]   Fig. 9. A: torques from a single subject generated by grasping and lifting a 2.0-N resistive force with the fingers aligned (10 trials), whereas B and C show the deliberate misalignment of the fingers in either the vertical or horizontal axis (10 trials). Note that in A, there was a negligible y-axis torque (left) when the finger were horizontally well aligned, but in B, vertical misalignment by approximately 20 mm produced a substantial z-axis torque (right). In contrast, C shows a negligible z-axis torque when the fingers were vertically well aligned (right), but a horizontal misalignment of approximately 17 mm produced a substantial y-axis torque (left).

EFFECT OF EXCESSIVE GRIP FORCE. A second control was carried out to determine what effect excessive grip force would have by itself. A naïve female subject with cutaneous sensation intact was asked to perform a series of 10 lifts against a 2.0-N resistive force using the force of her own choosing, and then a second series of 10 lift-and-hold trials using a 9.0-N grip force that was about the average force used by subjects with anesthetized fingers. The mean misalignment distance under the control condition was 1.83 mm compared with 2.27 mm using excessive grip force, but an F test indicated that this difference was not statistically significant. However the y-axis torque increased from an average of 19.0 to 119.3 N-mm [F(1,18) = 37.134, P < 0.001], and the z-axis torque increased from an average of 3. 3 to 13.0 N-mm [F(1,18) = 9.798, P < 0.001].

GRASPING LIFTING AND HOLDING IN A PATIENT WITH A POLYSENSORY NEUROPATHY. We compared the grasping and holding performance of healthy subjects after digital anesthesia with a 50-yr-old female patient, G. L., suffering from a polysensory neuropathy involving the afferent fibers over the entire body below the nose. The patient has a complete loss of touch, pressure, and kinesthesia in the limbs, neck, and trunk, and her condition has been stable for more than 20 yr (Forget and Lamarre 1987, Simoneau et al. 1999). G. L. was asked to perform the grasp, lift, and hold task, both with and without visual control. In the condition without vision, she was allowed to view the position of her hand and the manipulandum between trials. G. L. appeared to use a single default grip force since object fragility was not a constraint. Her grip force was about the same for the three resistive forces and two surface textures. She increased her grip force further when performing the task without visual feedback. Figure 7 (bottom) shows the 3-D reconstruction of the pressure cones under the thumb and index finger exerted by G. L. during static holding on three grasping trials under visual control. The mean offset distance of the center of pressure was greater than the mean distance in control subjects with intact cutaneous sensation and was similar to the mean offset distance in subjects after digital anesthesia. Figure 7 also shows that G. L. had a significant force imbalance favoring the index finger. In addition, G. L. also had an unusually large finger skin contact area, and this behavior may have reflected a compensatory strategy to cope with losing the sense of friction on the fingers by spreading the contact over a wider area of semi-compliant skin to increase the friction with the grasped object.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the first experiment we were able to replicate several observations from earlier studies of object manipulation after digital anesthesia. Specifically we confirmed that anesthetizing the thumb and index fingers disrupted the preprogramming of grip and load forces. The grip force onset and the initiation of the lifting force were temporally dissociated, resulting in a significant increase in the duration of the preloading phase (Johansson and Westling 1984a; Westling and Johansson 1984). We also confirmed that the capacity to optimally adapt grip forces to the simulated weight and surface friction of a manipulated object was severely impaired by the loss of cutaneous sensation. However the most striking effect shown by all subjects in this study was the use of vastly excessive grip forces during both lifting and holding, which had also been reported in many previous studies (Häger-Ross and Johansson 1996; Jenmalm and Johansson 1997; Johansson and Westling 1984a; Johansson et al. 1992; Westling and Johansson 1984).

There are probably several reasons for the grip force increases after digital anesthesia. First, local anesthesia reduces the rate of sweating and lowers the skin-object friction, which is known to influence grip force (Edin et al. 1992; Johansson and Westling 1984b; Smith et al. 1997). However, reduced friction due to dryness of the skin was only one factor contributing to the higher sustained grip forces after digital anesthesia. A second factor was the inadvertent and unconscious application of forces and torques, which would have required increased grip force to prevent the fingers from slipping on the grasping tabs.

Despite the fact that the subjects were fully familiarized with the modest range of resistive forces offered by the apparatus, after digital anesthesia, they nonetheless applied excessive grip forces without apparent adaptation on over 25 trials. Moreover, the anesthesia produced a significant disturbance in the correlation between the grip and load forces and the preprogramming of the grip and load force rates without any apparent tendency toward improvement. The anesthetized subjects showed some grip force modulation with different resistive forces presumably on the basis of proprioceptive feedback, whereas in contrast, the patient with the polysensory neuropathy showed no modulation of grip force with the various resistive forces. These results are surprising considering the several studies demonstrating anticipatory grip force control driven from an internal model of the object (Flanagan and Wing 1993, 1997) or from motor memories (Gordon et al. 1993). On the basis of these studies, one might have predicted a significant degree of compensation as a result of the experience acquired with the normal tactile sensation prior to the anesthesia. However, the results of this study indicate that the preprogramming of the grip and load force rates was severely disrupted by the local anesthesia. Instead, the data from this study suggest that the internal model requires either continuous tonic excitation from cutaneous receptors, or at the very least, frequent intermittent reiteration to function optimally.

Jenmalm and Johansson (1997) also found that changing the angle of flat grasping surfaces from parallel to various positive and negative cambers substantially changed the grip forces in close correlation with the net tangential-to-normal force ratio on the skin. In this case, vision of the grasping surfaces was sufficient to produce some compensation for the loss of tactile sensation from digital anesthesia, probably because the tapered surfaces provided a recognizable visual stimulus associated with either a substantial increase or decrease in required grip force. The authors noted, however, that despite some adaptation, the subjects with anesthetized fingers invariably used "considerably stronger horizontal forces " than the same subjects tested with normal digit sensitivity.

With cutaneous sensation intact, all subjects in the present study generated only very small forces in directions other than those intended for grasping and lifting. After digital anesthesia, the x- and z-axis linear forces arising from the force imbalance between the fingers and from pushing or pulling were significantly increased. In addition, significant off-axis torques appeared in the horizontal (y axis) and frontal (z axis) planes. However, it is important to emphasize that some of these extraneous forces might not have occurred on a freely moving object because they would have resulted in movement of the object that the subject could have corrected visually. Yet, because of the stationary nature of the apparatus in this study and coupled with the lack of pressure sensation from the fingers, substantial linear forces occurred during grasping about which the subjects were apparently unaware.

After local anesthesia, tangential forces arose because the pinch forces applied by each finger were no longer applied in a perfectly perpendicular manner between two completely aligned centers of pressure. Instead, each finger applied a vector that was partially tangential to the grasping surface, and together they created a tangential torque. This misapplication of the pinch meant that the measured grip force was closer to the slip point because of a substantial increase in the net tangential force and torque on the fingers.

The fixed nature of the manipulandum in this study compared with the "free-standing" characteristic of test objects used in most other studies of digital anesthesia raises an important question about the generality of the results of this study. One advantage of our apparatus is that it allowed us to demonstrate excessive force and misdirected torques with much lower loads (i.e., 0.5-2.0 N) than previous studies. These data imply that manipulating a freely moving object with anesthetized fingers would result in substantial tilting and rotating of the object although the extraneous linear forces in the x and z axes observed with our fixed manipulandum would be easily corrected by visual feedback. In a companion paper (Augurelle et al. 2003), it was found that, although subjects adapted relatively well to holding an object stationary against gravity, when the object was repeatedly accelerated and decelerated in oscillatory movements, it was frequently dropped, presumably as a result of tilting beyond a critical point. Why then, was this tilting not reported in earlier published accounts of the effects of finger anesthesia (Cole and Abbs 1988, Johansson and Westling 1984a)? Possibly because the centers of gravity of the objects used by both Cole and Abbs (1988) and Johansson and Westling (1984a) were so far below the misaligned fingers that an enormous torque would have been required to tilt the object to any appreciable degree. Also, the objects in both these studies were not truly free-standing but constrained either by cables (Cole and Abbs 1988) or by small holes cut into a table top (Johansson and Westling 1984a). Similarly, Edin et al. (1992) reported significant object tilting due to different coefficients of friction beneath each finger, but the degree of tilt had to be calculated because it was "barely noticeable, " given the low center of gravity.

In this study, vision would only have been useful to insure the proper alignment of the fingers. However, both the healthy subjects after digital anesthesia and the deafferented patient with years of practice in developing visual compensatory strategies failed to use visual feedback to correct the finger alignment in precision grasping, perhaps because the alignment discrepancy was too small to be visually apparent. How the loss of touch sensation at the fingertips leads to this misalignment of the fingertips during pinching is still not entirely clear. Correct finger alignment is partly a proprioceptive function that should be served by muscle spindles and tendon organs. Edin (1992) has shown that in hairy skin, SAI and SAII skin afferents are highly sensitive to stretch and may also play an important role in the kinesthesia of the hand. However, as anyone who has attempted the manipulation of small objects wearing gloves can attest, accurate placement of the fingers is clearly affected by the loss of skin sensation.

The high density of skin mechanoreceptors in the fingertips probably provides an accurate image of the pressure vectors generated by the object in contact with the grasping digits. Some form of associative learning could establish a link between the direction of pressure on the fingertips when handling familiar objects and the necessary opposition force required from the intrinsic hand muscles. Nevertheless as noted above, memory representations or internal models cannot anticipate either the location or the direction of the pressure vectors without the presence of skin afferent input.

An important issue arising from this study is whether the increased grip force after digital anesthesia is the cause or the effect of inefficient grasping. Increased grip force certainly contributed causally and significantly to the off-axis forces and torques in this study. However, the increased grip force itself may have been a response to the increased slipperiness of the skin and the inadvertent increased tangential loading on the skin. The subject with intact sensation grasping with deliberately excessive grip force demonstrated that even with a small lever arm, large unintended torques can occur, and this effect is exacerbated by any misalignment of the fingers. We would therefore speculate that the increased grip force after digital anesthesia results not only from the loss of the sense of pressure on the skin, but also from the inability to estimate the direction and magnitude of tangential force vectors and consequent failure to appropriately direct the apposition forces of the thumb and index finger. Certainly cutaneous afferents play an important role in adapting grip and lifting forces to surface friction and object weight, but our results also suggest that, in addition, cutaneous mechanoreceptors make an important contribution to guiding the direction of pinch force vectors effectively.