by stimulating a digit of one hand seem not to be associ- ated with spread ... with the right index and middle fingers and in the bimanual grasp the subjects used ...
Exp Brain Res (1999) 127:43–53
© Springer-Verlag 1999
R E S E A R C H A RT I C L E
Yukari Ohki · Roland S. Johansson
Sensorimotor interactions between pairs of fingers in bimanual and unimanual manipulative tasks
Received: 25 August 1998 / Accepted: 22 February 1999
Abstract The tuning of fingertip forces to the physical properties of objects during manipulation may be controlled partly by digit-specific mechanisms using local afferent information and partly by controllers that support interdigital coordination and use sensory information from more than one digit. In the present study we addressed digital interactions when humans used the tips of two fingers to restrain a pair of horizontally oriented plates from moving when subjected to tangential force loads in the distal direction. Subjects used the right and left index fingers in a bimanual grasp, and the right index and middle fingers in an unimanual grasp. The plates were loaded at unpredictable times by identical force profiles consisting of a force increase of up to a 3-N force plateau. The plates were concurrently loaded in 85% of the trials and each plate was loaded separately in 7% of the trials. For each plate, we measured its movement and the normal and tangential forces applied by the finger to restrain it. When a finger was loaded, the subject automatically responded by a normal force increase to a level that remained fairly constant during the subsequent load plateau. The initial part of this finger grasp response was affected by simultaneous loading of its partner finger; the magnitude of the response was boosted with a bimanual grasp, whereas the onset latency tended to be shorter with a unimanual grasp. Responses also occurred at a non-loaded finger during both bimanual and unimanual grasps, but these responses were weaker than those evoked when the same finger was loaded. In the bimanual grasp, they were largely characterized by a brief force pulse whose onset was delayed by some 15 ms compared with the response onset of the loaded finger, i.e., there was no sustained response. In the unimanual grasp, the onset of the response coincided in time with Y. Ohki (✉), R. S. Johansson Department of Physiology, Umeå University, S-90187 Umeå, Sweden Present address: Y. Ohki Kyorin University School of Medicine, Department of Physiology, 6–20–2 Shinkawa, Mitaka-shi, Tokyo 181-8611, Japan, Fax: +81-422-44-1816
that of the accompanying (loaded) finger, and the dynamic response was stronger and prolonged, with more than one force rate peak. There was also a significant static response present. We conclude that during unimanual as well as bimanual reactive restrain tasks there are interactions between digits engaged in terms of neural control that facilitate the response of a digit when an accompanying digit is simultaneously loaded. However, digit-specific afferent inputs are necessary for eliciting the full-size reactive grasp responses required to successfully restrain the manipulandum. Keywords Grip force · Grasp stability · Fingertip force · Human hand
Introduction When we hold and restrain objects subjected to unpredictably imposed external forces, we maintain grasp stability by automatically modulating the grip forces normal to the contact surfaces with the changes in tangential load forces (Cole and Abbs 1988; Johansson and Westling 1988; Johansson et al. 1992b, 1992c; Jones and Hunter 1992). These reactive grasp responses are initiated and parametrically controlled by sensory information from the digits (Häger-Ross and Johansson 1996; Johansson et al. 1992a; Macefield et al. 1996a). However, it is not known whether these reactive responses are driven by digit-specific sensorimotor mechanisms that rely on local sensory information or by sensorimotor mechanisms that impinge on all digits engaged. There is evidence from both unimanual and bimanual tasks that people use sensory information compiled across the engaged digits to partition fingertip forces to match the local friction at individual digit-object interfaces (Birznieks et al. 1998; Burstedt et al. 1997a). Furthermore, for fingers belonging to one hand, interactions between reactive responses are expected to occur between digits, because reflex responses evoked by electric and mechanical stimulation of one digit are not necessarily digit-spe-
44
cific (Caccia et al. 1973; Deuschl et al. 1995; Garnett and Stephens 1980). In contrast, reflex responses evoked by stimulating a digit of one hand seem not to be associated with spread of activity to digits of the other hand in healthy adults, but may occur in certain pathological conditions (Capaday et al. 1991; Farmer et al. 1990; Mayston et al. 1997). However, in both unimanual and bimanual manipulative tasks, it may be crucial that sensory triggered responses occur synchronously at the engaged digits to prevent an object from spinning, falling, or moving sideways. In general terms, during manipulative actions there is an obvious need for neural sensorimotor mechanisms that support interdigital coordination. For example, to produce a force output that reflects the overall manipulative intent, choice of grasp configurations and object geometry, temporal, intensive, and spatial coordination of digital actions clearly must rely on such control mechanisms. In this study we examined digit-specific and digit-interdependent features of the sensorimotor control of reactive force responses that support grasp stability. Two fingers were used to restrain a manipulandum that had two parallel, horizontally oriented contact surfaces that could be subjected to tangential loads at unpredictable times (see Burstedt et al. 1997a). A natural counterpart to this task is to place the index and middle fingers on a book that lies on a desk and to restrain the book from moving while someone else tries to drag it away. Because the two plates were concurrently loaded in the vast majority of the trials and therefore resembled a solid object, subjects could benefit from exploiting a tight sensorimotor coupling of the digits in terms of generating the reactive responses. To assess the presence of such a coupling we had a few randomly interspersed “catch” trials in which only one of the digits was loaded. By allowing the subjects to perform the task both unimanually (using right index and middle fingers), and bimanually (using right and left index fingers), we could determine whether the possible interdigital coupling was expressed in a similar fashion despite obvious differences in the anatomical substrates implementing the control. We will demonstrate that there are interactions between digits in terms of neural control, but digit-specific afferent inputs are necessary for eliciting purposeful grasp responses.
Materials and methods Subjects and general procedures With the approval of the local ethical committee, experiments were performed on seven healthy, right-handed human volunteers (two women and five men) aged between 23 and 53 years. The subjects gave their informed consent and were seated in a chair with their upper arms parallel to their trunk and their forearms extended anteriorly in the horizontal plane. The forearms up to the palms were supported by a tabletop and the hands were in intermediate pronation (palms down). The subjects used the tips of two fingers positioned side by side to restrain an instrumented manipulandum with two horizontally oriented, flat grasp plates (Fig. 1A). Subjects used each of two different grasp configurations: In the unimanual grasp, the subjects restrained the manipulandum
with the right index and middle fingers and in the bimanual grasp the subjects used the left and right index fingers. Each grasp plate could be loaded independently in the distal direction by force servomechanisms. The fingers were slightly flexed and the plane of the grasp plates approximately intersected the centers of the metacarpophalangeal joints. With such a posture, movements of the manipulandum caused minimal, passive normal force changes. The subjects were blindfolded during the experiments and the apparatus was quiet, i.e., it provided no sound cues. Apparatus The manipulandum is shown schematically in Fig. 1A. It consisted of two horizontally oriented, flat circular grasp plates (30 mm diameter; center-to-center distance, 32 mm) covered with suede. Each plate was connected to a separate servo-regulated torque motor (via a 10-cm-long stiff beam) that could generate loading forces at the grasp plate in the distal direction (0–10 N, 0–15 Hz bandwidth, noise less than 0.05 N; Johansson et al. 1992c). A laboratory computer controlled the force motors. Strain gauges at each plate transduced the load force tangential to the grasp plate and the normal force produced by the finger perpendicular to the grasp plate. The angular position of each plate was transduced to a resolution of 0.05 mm. The two grasp plates were servo-regulated to constant position (stiffness 1.2 N/mm) when the fingers were not touching the manipulandum. Load trials Load trials, configured as shown in Fig. 1B,C, were delivered to either one of the fingers or simultaneously to both fingers. During the first 20 ms of the load phase, the distal load increased abruptly by 0.8 N, starting from a 0.2-N baseline load. After this “load step”, the load continued to increase at a constant rate of 4 N/s for 0.5 s to a hold phase (3 N constant force), which was maintained for 1.0 s. The grasp plate was then rapidly unloaded. The 0.2-N baseline load was automatically applied when the grasp plate was contacted and ensured that the subjects maintained contact with the grasp plate between trials. The high initial load force rate (0.8N load step) served to trigger a distinct normal force response, and the following load ramp increase during the load phase guaranteed a substantial response amplitude (Johansson et al. 1992b). A test series included three types of trials with regard to loading conditions: (1) separate loading of digit 1 (right index finger and right middle finger in the bimanual and unimanual grasp configurations, respectively); (2) separate loading of digit 2 (left and right index finger in the bimanual and unimanual grasp configuration, respectively); and (3) concurrent loading of both fingers. Each test series was composed of 55 trials. Both grasp plates were concurrently loaded in 47 trials (85% of the trials). Eight trials were randomly interspersed, four in which digit 1 was loaded in isolation and four trials with loading of only digit 2. Those trials with single digit loading occurred during the last 37 trials of the test series. The construction of the series allowed consistent early experiences of the prevailing loading condition. Moreover, prior to the series, the subjects were informed that the digits would be concurrently loaded in nearly all trials. The interval between consecutive load trials was randomly selected from 1.5 to 2.5 s. The subjects were instructed to hold the grasp plates at the original positions during the test series. The manipulandum slipped in 9% of the test series; in these cases the test series was repeated from the beginning. Data collection and analysis Data were collected and analyzed with a laboratory computer system (SC/ZOOM; Department of Physiology, University of Umeå). The force signals were sampled at 400/s and the position signal at 100/s (12-bit resolution). Event markers related to onsets
45 Fig. 1A–C Schematic of the apparatus and examples of single-trial responses during the bimanual and the unimanual grasp. A Schematic of the manipulandum. It consisted of two horizontally oriented, flat circular grasp plates and each plate was connected to a separate servo-regulated torque motor via a stiff beam. Normal and tangential load forces were measured for both grasp plates. B, C Superimposed records from single trials in which the right index finger was loaded (solid lines) during the bimanual grasp in one subject and the unimanual grasp in another subject. The accompanying finger (dashed lines) was not loaded in these particular trials. Thick lines indicate one of the trials in each grasp configuration that is used to show measurements taken for statistical analyses (arrows and dots). Vertical dotted lines indicate the onset and the end of the load increase. For further details, see text
of the various phases of the load trials were sampled at ±0.1 ms time resolution. Force rates were obtained as a function of time by symmetrical numerical time differentiation within a time window corresponding to ±5 data samples. The following measurements were taken from single trials and from each digit (Fig. 1B,C): The pre-trial normal force was the normal force present at the onset of the load increase. The grasp response onset latency was the time interval from the onset of the load force increase to the onset of the reactive normal force increase as assessed from the force rate signals. We identified one or more peaks in the normal force rate profile (Fig. 1B,C). For the 1st peak and, when readily detectable, also for the 2nd peak, we measured its amplitude and time of occurrence relative to the onset of load force increase. A normal force response was considered present when the maximum normal force rate exceeded 5 N/s; responses of weaker strengths could not be reliably detected in single trial records. For trials in which a force response was not present according to this criterion, none of the above response parameters were measured; the response size measures pertaining to peak rates were approximated at zero in the statistical analyses for these trials. Likewise, for trials in which only one rate peak could be reliably discerned according to the 5-N/s criterion, the response size measures for the 2nd peak was approximated at zero. The dynamic normal force response was defined for all trials as the difference between the maximum normal force during the load phase and the pre-trial normal force. Similarly, the size of the static normal force response was
measured as the difference between the normal force at 50 ms before the end of the hold phase and the pre-trial normal force. Statistical methods Numerical values of normal forces and tangential forces were transferred to a statistical program (STATISTICA; Statsoft, Tulsa, Okla.). Unless otherwise stated, the statistical reports emanate from two-way repeated-measures analysis of variance (ANOVA). One type of ANOVA was performed to primarily assess the influence of grasp configuration by comparing responses by the right index finger during the two different grasps. The factors were Grasp (2 levels: bimanual or unimanual) and Loading condition (3 levels: finger concurrently loaded, loaded in isolation or not loaded). The other type of ANOVA was performed to analyze differential behaviors of two cooperating fingers in either grasp. The factors were Digit (2 levels: digit 1 or digit 2) and Loading condition (3 levels, as above). In addition, planned comparisons were performed to analyze specific effects as described in the Results section. The level of probability selected as statistically significant was P
