Jan 24, 1995 - the fingertip forces in precision grip tasks involving re- straint of objects ...... Knibestol M, Vallbo A (1970) Single unit analysis of mechanore-.
Exp Brain Res (1996) 110:131-141
0Springer-Verlag 1996
.
Charlotte Hager-Ross Roland S. Johansson
Nondigital afferent input in reactive control of fingertip forces during precision grip
Received: 24 January 1995 1 Accepted: 15 December 1995
Abstract Sensory inputs from the digits are important in initiating and scaling automatic reactive grip responses that help prevent frictional slips when grasped objects are subjected to destabilizing load forces. In the present study we analyzed the contribution to grip-force control from mechanoreceptors located proximal to the digits when subjects held a small manipulandum between the tips of the thumb and index finger. Loads of various controlled amplitudes and rates were delivered tangential to the grip surfaces at unpredictable times. Grip forces (normal to the grip surfaces) and the position of the manipulandum were recorded. In addition, movements of hand and arm segments were assessed by recording the position of markers placed at critical points. Subjects performed test series during normal digital sensibility and during local anesthesia of the index finger and thumb. To grade the size of movements of tissues proximal to the digits caused by the loadings, three different conditions of arm and hand support were used; (1) in the hand-support condition the subjects used the three ulna fingers to grasp a vertical dowel support and the forearm was supported in a vacuum cast; (2) in the forearm-support condition only the forearm was supported; finally, (3) in the no-support condition the arm was free. With normal digital sensibility the size of the movements proximal to the digits had small effects on the grip-force control. In contrast, the grip control was markedly influenced by the extent of such movements during digital anesthesia. The poorest control was observed in the hand-support condition, allowing essentially only digital movements. The grip responses were either absent or attenuated, with greatly prolonged onset latencies. In the forearm and no-support conditions, when marked wrist movements took place, both the frequency and the strength of grip-force responses were higher, and the grip response latencies were shorter. However, the performance never approached norC. Hager-Ross (W) . R.S. Johansson Department of Physiology, Umeb University, s-90187 UmeH. Sweden: Fax: +46-90-16 66 83
mal. It is concluded that sensory inputs from the digits are dominant in reactive grip control. However, nondigital sensory input may be used for some grip control during impaired digital sensibility. Furthermore, the quality of the control during impaired sensibility depends on the extent of movements evoked by the load in the distal, unanesthetized parts of the arm. The origin of these useful sensory signals is discussed.
Key words Grasp force . Tactile sensibility . Proprioception . Motor control . Human
Introduction Mott and Sherrington (1895) noted that with spared distal innervation of the hand there was little functional impairment in the forelimb, even if the rest of the arm was deafferented. Since then, the general importance of digital sensibility in the control of precise finger movements has been repetitively documented (e.g., Denny-Brown 1966; McCloskey et al. 1983; Moberg 1962; Twitchell 1954). However, not until recently have we began to understand the specific sensorimotor control mechanisms that are involved. During precision grip tasks, for instance, signals in digital afferents that reflect important mechanical events at the digit-object interface elicit strong, purposeful automatic motor consequences. These are contingent on the specific characteristics of the encoded mechanical events at the hand-object interface and on the goal, the context, and the phase of the task (for recent reviews see Johansson and Cole 1994; Johansson and Edin 1993). Grasp stability is maintained by reactive control of the fingertip forces in precision grip tasks involving restraint of objects subjected to destabilizing external load forces (Cole and Abbs 1988; Johansson and Westling 1988; Johansson et al. 1992a, b; Jones and Hunter 1992). The grip forces, normal to the contact surfaces, are automatically regulated to the amplitude and rate with which loads are imposed tangential to the grip surfaces (Johans-
son et al. 1992a, b). Moreover, the grip forces are scaled by the frictional conditions at the digit-object interface (Cole and Johansson 1993). During load ramps consisting of a sustained force increase, a strong grip force increase (essentially a brief force-rate pulse) is automatically elicited after a short delay (Johansson et al. 1992a). This "catch-up" response accounts for a quick restoration of an adequate grip force to load-force ratio to prevent frictional slips. The grip force then continues to increase, seemingly tracking the load-force change to maintain an adequate grip force safety margin against slips (cf. Cole and Johansson 1993). Recent microneurographic recordings have demonstrated that cutaneous receptors are sensitive to the tangential forces imposed on the skin by an applied load and therefore could be responsible for both the initiation and the scaling of the grip responses (Macefield et al. 1996). In contrast, receptors in the intrinsic and extrinsic hand muscles do not respond early enough to allow them to play such a role (Macefield and Johansson 1996). Accordingly, following anesthetic block of the digital nerves, the grip response is both delayed and attenuated in some subjects and totally abolished in others (Johansson et al. 1992c; see also Cole and Abbs 1988). However, in these previous experiments with digital anesthesia, the hand was supported in a manner that minimized load-related movements of structures proximal to the anesthetized digits (Johansson et al. 1992~).This presumably resulted in a particularly poor activation of nondigital afferents (i.e., afferents originating in muscles, joints, tendons, and skin proximal to the digits). Therefore, even if these afferents are less sensitive to load forces than the tactile afferents, they may still be used to mediate grip responses during mechanically less restricted conditions. In the present study we analyzed gripforce control during various hand and forearm support conditions designed to control the extent of load-related movements in segments of the hand and arm proximal to the digits. A brief account of these results has appeared in abstract form (Hager et al. 1989).
Materials and methods Subjects, apparatus, and general procedure Nine healthy right handed subjects participated in the study (three men, six women, aged 20-45 years). Informed consent was obtained according to the Declaration of Helsinki. Subjects sat with their right arm abducted about 30°, their elbow joint flexed at an angle of about 100°, and their forearm extended anteriorly intermediate between pronation and supination. They used the tips of the thumb and index finger to grasp an instrumented manipulandum, which was connected to a servo-controlled, brushless force motor (Fig. 1A). The force servo, designed at the Department of Physiology, UmeH University (Johansson et al. 1992a), was fully programmable using a laboratory microcomputer and could produce loads in the proximal or distal direction tangential to the grip surfaces, i.e., toward or away from the palm (0-10 N in each direction; band width ca. 0-15 Hz; noise less than 0.05 N). The hand and the manipulandum were shielded from the subject's view by a curtain, and the motor was noiseless.
The two parallel grip surfaces of the manipulandum (30-mmdiameter discs, spaced 25 mm apart) were covered by fine-grain sandpaper (no. 320). This material was selected because it provides a high and stable friction against the skin, that is, the coefficient of friction is little influenced by the reduced sweating that occurs during digital anesthesia, and it is quite stable across subjects (cf. Johansson and Westling 1984b; Westling and Johansson 1984). Arrays of strain-gauge force transducers measured the load force and the grip force (perpendicular to the grip surfaces) at each of the two grip surfaces (0-120 Hz; noise less than 0.02 N). The error in force measurements due to different points of force attack was typically &3% and always less than +5%. The sum of the load force at the two grip surfaces was servo-regulated. The position of the manipulandum was recorded with a servopotentiometer (Helipot 6263; Beckman, Fullerton, Calif., USA) attached to the motor shaft (0.05 mm effective resolution at the grip surfaces). In separate experiments we used a CCD camera placed in the ceiling to record the position of reflective markers placed at: (I) the radial aspect of the hand close to the rotation axis of the metacarpophalangeal (MCP) joint of the index finger, (2) the skin over the styloid process of the radius, and (3) the manipulandum. We recorded the displacements of the markers (spatial resolution ca. 0.01 mm) occurring in a plane coinciding with the plane in which the load forces were delivered and in which all load-related movements principally took place. Support conditions for hand and forearm Three support conditions were used to grade the number of loadrelated movements in different segments of the hand and forearm: 1. In the hand-support condition the forearm was gently strapped and rested in a vacuum cast molded to the individual's forearm. In addition, the subject firmly grasped a vertically oriented handle (dowel 100 mm long, 25 mm diameter) using the three ulnar fingers, which left the thumb and index finger free to hold the manipulandum (Fig. 1A, top panel). Because the handle was anchored to the support frame of the motor, the reaction forces to the load forces were primarily directed via the hand through the handle support. On average the handle was subjected to 70% of the load force (data from 540 trials by three subjects collected in test series in which the handle was equipped with strain gauges). 2. In the forearm-support condition, the subject did not gain support from any handle (Fig. 1A, middle panel). Hence, the reaction forces to the load forces were presumed to be directed mainly to the vacuum cast via the wrist and forearm. 3. In the no-support condition (Fig. lA, bottom panel), the reaction forces were directed through the wrist, arm, and shoulder, since no hand or forearm support was used. Load trials The load-force trials consisted of a sustained force increase starting from zero load-force (load phase) and a period of maintained load @lateau phase) (Fig. 1B). A sustained force decrease returned the load to zero at the same force rate as during the load phase. Data were analyzed for the load and the plateau phases only. Fifteen trials delivered in the distal direction (away from the palm) and 15 in the proximal direction (toward the palm) comprised a test series. In each load direction we ran three trials of the following five combinations of force amplitude and force rate: (1) 4 N at 4 Nls; (2) 4 N at 8 Nls; (3) 2 N at 4 Nls; (4) 2 N at 8 Nls, and (5) 1 N at 32 Nls. The sequence of trials was unpredictable with regard to direction, force rate, and force amplitude. Furthermore, the duration of the plateau phases was randomized between 2 and 3 s and the intertrial delay between 2.5 and 4.5 s. For the 1N trials, termed "step load trials," the load-force rate was approximately 32 Nls. The sole purpose of incorporating the step load trials was to further diversify the presentation of the load with regard to load-force rates and amplitudes. The subjects were asked to restrain the manipulandum from moving when load forces were ap-
Fig. 1 A Schematic drawing of the apparatus and the three support conditions. B Two single trials superimposed (load amplitude 4 N, rate 4 Nls), one in the distal (solid curves) and one in the proximal (dashed curves) loading direction. Circles on vertical lines indicate points of measurements (illustrated for the trial in the distal direction): a, position of manipulandum and preload grip force measured at the onset of load phase; b, time for onset of gripforce response; c, peak grip force rate; and d, position of manipulandum and static grip force measured 0.2 s prior to end of load-force plateau. C indicates the "catch-up" response and T the "tracking" response
A
Force motor
B Grip force rate, Nls
C
Grip force, N
Load force, N.
Position, mm
abc plied, whilst not using excessive grip forces. If the manipulandum escaped from the grasp, it was returned to the initial position, regrasped by the subject, and the test series was resumed by repeating the current trial. Five subjects performed two test series with the hand support and one test series for the remaining support conditions during normal digital sensibility and during local anesthesia of the index finger and thumb. Digital anesthesia was achieved by blocking the digital nerves by injections of bupivacain at the midshaft of the proximal phalanges (ca. 5 mgldigit). The tests were not run until a complete clinical anesthesia of the digits to touch, pinprick, and squeezing was obtained. Four subjects with normal digital sensibility participated in the experiments in which movements of different hand and forearm segments were recorded using reflective markers. Each subject received 36 load trials (ramp rate 4 Nls and plateau phase amplitude 2 N), of which 18 were delivered in the distal and 18 in the proximal direction and presented in a random order. Data collection and analysis All signals were digitized (12 bits), stored, and analyzed using the SCIZOOM microcomputer-based data acquisition and analysis system (Department of Physiology, UmeH University). The grip and load forces at the two surfaces were sampled at 400 Hz, and the position of the manipulandum and of the reflective markers at 100 Hz. The first time derivatives of the forces (force rates) were obtained using a symmetrical k5-point numerical differentiation of the sampled force signals. The grip and load forces recorded at each of the grip surfaces were presented as the mean grip and load force at the two grip surfaces. For each load trial the following measurements were derived for statistical analysis (see Fig. 1B). The preload grip force was the grip force at the onset of the load phase, and the static grip force was measured 0.2 s before the termination of the load-force plateau. The overall restrain performance was the displacement of the manipulandum that occurred between these two points in time. For each trial a grip-force modulation index was defined as the difference between the static and the preload grip force expressed as the fraction of the static grip force (cf. Johansson et al. 1992~). Thus, this index not only reflects the size of the grip responses but also is influenced by the compensatory increase in intertrial grip force that may occur during impaired digital sensibility. The peak grip force rate was the maximum rate of grip force increase within a 2-s period starting at the onset of the load phase. Since this maximum almost always occurs during the catch-up response, peak
T
d
grip force rate provides an early measure of the scaling of the triggered grip force response to the load-force rate (Johansson et al. 1992b) and to the friction in the object-digit interface (see Cole and Johansson 1993; Hager-Ross et al. 1996). The grip response latency, measured from the force rate signals, was the time interval from the onset of the load phase to the start of the grip-force increase. A grip response was considered present when the peak grip force rate exceeded 5 Nls during the same 2-s interval indicated above, i.e., it was the minimum response rate that could be detected reliably in single-trial records. A subject's responsiveness was defined as the percentage of trials in which grip responses were observed. Thus, the responsiveness not only provides an important measure of response reduction during digital anesthesia (Johansson et al. 1992c) but also indicates the fraction of trials for which latency measurements have been made. To obtain a measure of the movements of the hand-arm system taking place prior to the subject's grip response, the initial displacement of the manipulandum was measured during the first 150 ms following the onset of the load increase. For trials with the lowest load force rate (4 Nls), the restrain response triggered by the load had not yet begun to influence the trajectory of the position signal (cf. pre-response stiffness, in Johansson et al. 1992a). For such trials during the same 150-ms period, the digital yield was gauged as the change in distance between the manipulandum and the MCP-joint reflective marker, the hand yield as the change in distance between the MCP-joint marker and the wrist marker, and the arm yield was the displacement of the wrist marker. These yield measures were all computed as the net positional changes of the markers in the coordinate of the loading direction, i.e., essentially all load-related movements occurred in this direction (see Results). Statistical methods
A 2 x 3 ~ 2 ~ 2 ~MANOVA 2~5 (Software STATISTICA, StatSoft 1994) was used to evaluate the influence of digital sensibility (normal and impaired), support condition (hand-, forearm-, and nosupport), load-force rate (4 Nls and 8 Nls), load-force amplitude (2 N and 4 N), load direction (distal and proximal), and subject (1-5) on preload grip force, static grip force, overall restrain performance, grip-force modulation index, peak grip force rate, grip response latency, and initial displacements of the manipulandum as defined above. However, because the analysis of yield components was carried out in separate experiments with only one type of load trial and during normal sensibility only, a separate 3 x 2 ~ 4 MANOVA was run to evaluate the influence of support condition,
load direction, and subject (1-4) on the digital yield, hand yield, arm yield, and initial displacement of manipulandum in these experiments. For each MANOVA the level of probability chosen as statistically significant was P
