1Department of Integrative Medical Biology, Section for Physiology, Umeå University, SE-901 87 Umeå, Sweden; and 2Department of Psychology, Queen's University, Kingston, Ontario K7L 3N6, Canada
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Burstedt, Magnus K. O., J. Randall Flanagan, and Roland S. Johansson. Control of Grasp Stability in Humans Under Different Frictional Conditions During Multidigit Manipulation. J. Neurophysiol. 82: 2393-2405, 1999. Control of grasp stability under different frictional conditions has primarily been studied in manipulatory tasks involving two digits only. Recently we found that many of the principles for control of forces originally demonstrated for two-digit grasping also apply to various three-digit grasps. Here we examine the control of grasp stability in a multidigit task in which subjects used the tips of the thumb, index, and middle finger to lift an object. The grasp resembled those used when lifting a cylindrical object from above. The digits either all contacted the same surface material or one of the digits contacted a surface material that was more, or less, slippery than that contacted by the other two digits. The three-dimensional forces and torques applied by each digit and the contact positions were measured along with the position and orientation of the object. The distribution of forces among the digits strongly reflected constraints imposed by the geometric relationship between the object's center of mass and the contact surfaces. On top of this distribution, we observed changes in force coordination related to changes in the combination of surface materials. When all digits contacted the same surface material, the ratio between the normal force and tangential load (Fn:L ratio) was similar across digits and scaled to provide an adequate safety margin against slip. With different contact surfaces subjects adapted the Fn:L ratios at the individual digits to the local friction with only small influences by the friction at the other two digits. They accomplished this by scaling the normal forces similarly at all digits and changing the distribution of load among the digits. The surface combination did not, however, influence digit position, tangential torque, or object tilting systematically. The change in load distribution, rather, resulted from interplay between these factors, and the nature of this interplay varied between trials. That is, subjects achieved grasp stability with various combinations of fingertip actions and appeared to exploit the many degrees of freedom offered by the multidigit grasp. The results extend previous findings based on two-digit tasks to multidigit tasks by showing that subjects adjust fingertip forces at each digit to the local friction. Moreover, our findings suggest that subjects adapted the load distribution to the current frictional condition by regulating the normal forces to allow slips to occur early in the lift task, prior to object lift-off.
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In common manipulatory tasks the loads that
potentially destabilize the grasp include time-varying forces and
torques tangential to the grasped surfaces. These tangential loads
develop as a consequence of the subject's actions on the object.
Linear load forces counteract gravitational and inertial forces and
occur, for instance, whenever we lift an object from a support. Torque
loads develop when we tilt an object or accelerate the hand-held object
with the center of mass (CM) off the grip axis (e.g., the line joining
the tips of the thumb and the index finger in a precision grip task).
Several studies during the last 15 yr have examined the control of
grasp stability under tangential force loads (for recent reviews see Johansson 1996, 1998
; Wing
1996
) and more recently under loads that include tangential
torque (Goodwin et al. 1998
; Johansson et al.
1999
; Kinoshita et al. 1997
; Wing and
Lederman 1998
). A number of sensory-motor mechanisms involved
in the control of grasp stability have been identified. These
mechanisms serve to prevent linear and rotational slips and excessive
forces by automatically regulating forces normal to the grasped
surfaces to match the tangential load. Furthermore, to cope with
different frictional conditions between digits and objects, subjects
use tactile information about friction to adapt the ratio between
normal force and tangential load to the prevailing frictional
conditions of the grasp. Importantly, in two-fingered lifting tasks
performed unimanually, bimanually, or by two cooperating subjects,
subjects tune this ratio independently at each digit to the local
frictional condition at that digit (Burstedt et al.
1997b
; Edin et al. 1992
). When subjects lift objects with parallel vertical contact surfaces covered with different materials, they achieve different ratios at the two digits by applying
a lower tangential force at the more slippery side. This strategy
typically results in some object tilt (toward the more slippery side)
after the tangential lift forces overcome gravity and the object lifts
off. Similarly, when subjects restrain active objects they also adjust
the distribution of tangential forces such that the force coordination
at individual fingers is adapted to the local frictional conditions
(Birznieks et al. 1998
; Burstedt et al.
1997a
). Thus whenever the mechanical constraints imposed by the
task and the object permit, subjects appear to adjust the force
coordination at the individual digits for grasp stability.
The control of grasp stability has until recently been studied
primarily in tasks involving two digits only, typically the thumb and
index finger. Yet, most motor skills we associate with dexterous
manipulation involve more than two digits. Although a multidigit grasp
is inherently more stable than a two-digit grasp, it presents the
sensorimotor systems with an added challenge. That is, because the
orientations of the force vectors applied by the separate digits are
less constrained in multidigit grasps, the motor controller has to
choose from a number of possible solutions, i.e., grasp stability can
be achieved with many different combinations of fingertip forces
(Flanagan et al. 1999). The search for rules employed by
the nervous system for force sharing among digits under various
experimental conditions has been a topic of several previous
investigations (Amis 1987
; Imrhan and
Sundararajan 1992
; Kinoshita et al. 1995
,
1996
; Latash et al. 1998
; Li et
al. 1998
; Radhakrishnan and Nagaravindra 1993
;
Radwin et al. 1992
). However, these studies analyzed
neither the coordination of normal and tangential forces applied by
individual digits nor the distribution of fingertip force among the
digits in the context of control of grasp stability or specified task
constraints. To our knowledge, our prior study (Flanagan et al.
1999
) is the only previous study in which mechanisms supporting
grasp stability have been explicitly addressed during multidigit
manipulation. We examined in that study the control of fingertip forces
when subjects lifted an object with unimanual and bimanual three-digit
grasps that engaged the tips of the thumb and two fingers. The grasp
resembled those used when lifting a cylindrical object from above. We
found that many of the principles for coordination of fingertip forces
originally demonstrated for two-digit grasping (see Johansson
1996
) also apply to this multidigit manipulatory task. First,
the normal forces generated by each digit increased in parallel with
the vertical tangential force applied to lift the object (subjects tended to avoid horizontal tangential forces). The vertical lift forces
(and normal forces) were synchronized across the digits, and the
contribution by each digit reflected intrinsic object properties, i.e.,
geometrical relationship between the grasp sites and the center of mass
of the object. Second, the development of normal forces and load forces
before object lift-off reflected the object's weight (and mass
distribution) and was thus based on sensorimotor memory built up from
experience of object mass in previous lifts. Third, the sensorimotor
mechanisms engaged in the control of normal force appeared to take into
account the combined effect of linear (tangential force) and rotational
(tangential torque) and load components (see Goodwin et al.
1998
; Johansson et al. 1999
; Kinoshita et
al. 1997
; Wing and Lederman 1998
).
In the present study, we extended the analysis of three-digit manipulation by examining the control of grasp stability under various frictional conditions between the digits and the object. First, we wanted to investigate whether the frictional conditions influenced the magnitudes of the fingertip forces and the distribution of forces among the digits. Second, we wanted to know the extent to which the coordination of fingertip forces at the individual digits is tuned to the local frictional conditions at the separate grasp sites; that is, if the ratio between the normal force and the tangential load is adapted to the local frictional condition. Finally, in light of the fact that subjects should be able to achieve grasp stability with many combinations of fingertip actions in multidigit manipulation, we wanted to identify the ways in which they preferred to implement the grasp.
![]() |
METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Subjects and general procedures
Seven healthy subjects (3 men and 4 women between 19 and 45 yr of age) participated in this study after giving their informed consent. All subjects were naive concerning the specific purpose of the study. Subjects sat in an ordinary chair with their upper arms parallel to the trunk and the forearms extended anteriorly. An instrumented test object (Fig. 1A) was located on a small table in front of the subject. On instruction, the subject lifted the object to a height of ~6.5 cm by grasping the object from above using the tips of right thumb, index finger, and middle finger (Fig. 1A). No explicit instructions were given regarding the speed of lifting, forces to apply, or the orientation of the object in space, but the experimenter demonstrated the task. In addition, to familiarize the subject with the task, they lifted the object once with the contact surfaces covered with sandpaper before the experiment. Throughout the trials, subjects could see their hand and the test object. Before the experiments, subjects washed their hands with soap and water.
|
Test object
The instrumented test object (Fig. 1, A and
B) has been described in detail in an earlier report
(Flanagan et al. 1999). Briefly, each digit
contacted a separate vertically oriented contact disk (30 mm diam).
Each contact disk was exchangeable, and the surface facing the digits
was covered either by rayon or fine grain sandpaper (no. 320). The
distance between the center of each disk and the center of mass (CM) of
the object was 30 mm in the horizontal plane. The mass of the object
was 0.4 kg, and its center was located ~2 cm below the contact
surfaces when the object was level. Each contact disk was mounted on a
six-axis force-torque sensor (Nano F/T transducers, ATI Industrial
Automation, Garner, NC) that measured the normal force
(Fz) perpendicular to the disk and two
orthogonal forces tangential to the contact disks (Fig.
1C). Tangential forces were measured in the vertical
(Fy) and horizontal
(Fx) directions when the object was level,
i.e., Fy directly measured the vertical lift
force unless the object was tilted. The sensor likewise measured torques about these three force axes at the center of the contact surface (Fig. 1C). An electromagnetic position-angle
sensor recorded the linear position and angular orientation of the
object in three dimensions independent of the subject ("world"
coordinates; Fig. 1A; 3SPACE, FASTRAK, Polhemus,
Colchester, VT). The angular orientation of the object was represented
in Euler angles (azimuth, elevation, and roll). These were all zero
when the object was located on the table before lift-off. Notably, any
tilting of the object out of the horizontal plane was gauged by changes
in the elevation and roll angles. Subjects primarily moved the object
along the y-axis in the present lifting task.
Lifting trials
The frictional condition of the grasp was varied experimentally by letting the subjects lift the object under five different surface combinations: 1) all disks covered with sandpaper, 2) rayon at the index finger only, 3) rayon at the middle finger only, 4) rayon at both the index and middle fingers, and 5) all three disks covered with rayon. Each subject completed eight consecutive lifting trials with each of the five combinations of surface materials.
Three different auditory cues paced the subject through each
lifting trail (Fig. 1D). The first auditory cue notified the subject to grasp the object, lift it and hold it steady in air (hold phase). Four seconds after the first cue a second
auditory cue prompted the subject to perform a fiddling
procedure. By this procedure, we obtained estimates of the
coefficient of static friction for each digit on a trial-by-trial basis
as previously described (Flanagan et al. 1999). During
the fiddling procedure, the subject slid the tip of each digit, in any
order, across the contact surface while holding the object in air. For
each digit, the subjects typically chose to generate the slip by
simultaneously decreasing the normal force slightly and increasing the
vertical force (Fig. 1D). The decrease in vertical force at
a given digit was associated with object tilting and changes in
tangential torque and vertical force at the other digits (see
fluctuations in elevation and roll angles and in tangential torques in
Fig. 1D). Although all subjects generated three intentional
slips in all trials, one for each digit, only once did a subject drop
the object. A new period of stable holding commenced after the fiddling
procedure. A third auditory cue that appeared ~3 s later prompted the
subject to replace the object on the tabletop and release it.
Data analysis and statistical procedures
A flexible data acquisition and analysis system (SC/ZOOM,
Department of Physiology, Umeå University) was used to sample signals from the force-torque sensors (400 samples/s; 12-bit resolution) and
the position-angle sensor (120 samples/s; 14-bit resolution). The
sampling program transferred the origin of the forces and torques from
the surface of the transducers to the contact surfaces (Fig.
1C). The force tangential to the contact surface
(Ft) was computed as the vector sum of
the two tangential force components, and the force normal to the
contact surface (Fn) was defined
simply as Fz. The position of a
digit at its contact surface (Px and Py) was defined as the equivalent
point of normal force pressure calculated from the torques about the
x- and y-axes of the contact disks
(Tx and
Ty) and
Fz as described in Kinoshita et
al. (1997)
.
If the positions of the digits deviated from the z-axis
intersecting the centers of the contact surfaces, the measured torque (Tzo) could have differed from the
true tangential torques at the fingertips. That is, off-axis torques
caused by forces tangential to the contact surface could have
contributed to the measured torque. To determine the true tangential
torque, we subtracted the off-axis torques as follows:
Tz = Tzo Fy · Px + Fx · Py. We then defined the true torque
about the normal force vector as Tn =
Tz. The arrows representing torques
in Fig. 1C indicate positive directions of torque
measurements at each contact disk.
Kinoshita et al. (1997) have recently shown that the
minimum normal force, or slip force
(Fs), required by a human fingertip to
prevent any slip, linear or rotational, in the face of a tangential force (Ft) and tangential torque
(Tn) can be estimated by the following
equation
![]() |
(1) |
We defined the phases of the lifting trial as in Flanagan et al.
(1999). Accordingly, we defined the preload
phase as the period between the moment the leading digit contacted the
object and the onset of the load phase. The latter began
when the first time derivative of the total vertical force generated by
the three digits last exceeded 0.5 Ns
1 before reaching
its maximum value, i.e., when the vertical force began to increase
steadily. The time at which the total vertical force reached the mean
total vertical force employed during the hold phase was the end of the
load phase and closely matched the time of lift-off. Force, torque,
position, and angle measurements determined for the hold phase were
computed as averages of the values recorded during the last 0.5 s
before the delivery of the auditory cue that prompted the subjects to
perform the fiddling procedure. As a measure of the safety margin
against slips during the hold phase, we used the relative safety margin
defined as SM = (Fn
Fs)/Fn.
We used linear regression and correlation analysis to examine relations
among variables and repeated measures ANOVA to assess effects of
surface combination and digit (unless otherwise indicated in the text).
We considered a P value of 0.05 statistically
significant. Values reported in the text for data pooled across trials
refer to means ± SD based on one mean value obtained for each
digit and each series of lifting trials (unless otherwise indicated in
the text). We ignored the first trial in each of the series in the
analysis because initial adjustments to new frictional conditions were
not the focus of the study (cf. Birznieks et al. 1998;
Edin et al. 1992
). Furthermore, we focused on the hold
phase before the fiddling procedure. In this phase the grasp conditions represented those initially chosen by the subjects, whereas the second
hold phase occurred after the fiddling procedure during which the grasp
was reorganized. Likewise, we did not consider the replacement and
release of the object in this study.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We first provided a general description of the subjects' behavior during the three-digit lifting task based on data pooled across the various combinations of materials at the contact surfaces. We then analyzed the influence of changes in frictional conditions at the digit-object interfaces. We focused on how subjects control the fingertip forces to obtain grasp stability. We specifically asked to which extent, and how, the ratio between the normal force and the overall load (Fn:L ratio) at each fingertip was adapted to the local frictional condition assessed as the minimum ratio required to prevent slip, i.e., the slip ratio.
General performance
The subjects' behavior corresponded to that described for the
"standard" grip in our previous study of the three-digit lifting task (Flanagan et al. 1999). Figure 1D shows
a single trial that illustrates the task and its phases. In this trial,
one digit (middle finger) contacted the slippery material rayon, and
the cooperating digits (thumb and index finger) contacted sandpaper.
PRELOAD AND LOAD PHASES. After contacting the object, all subjects exhibited a preload phase during which they applied normal forces (Fn) before they reliably applied vertical "lifting" forces (Fy). During this phase, the digits were also subjected to some tangential load (Fig. 1D) reflecting small vertical forces, forces tangential to the contact surfaces in the horizontal direction (Fx), and/or small tangential torques (Tn).
During the subsequent load phase, the normal force increased in parallel with the vertical force and the load at each contact surface (Figs. 2, A-D). This type of "parallel" coordination has previously been demonstrated for grasps involving two digits (Johansson and Westling 1984a
|
HOLD PHASE.
During this phase, as well as in the preceding load and lift phases,
the thumb applied stronger vertical and normal forces than the
cooperating fingers, which applied about equal amounts of force (Fig.
1D; also see Figs. 5). This overall distribution of force
was expected based on the geometric relationship between the contact
surfaces of the object and its mass distribution (for details see
Flanagan et al. 1999). Indeed, these object properties would fully determine the distribution of normal and vertical forces
among the digits if subjects applied the forces at the centers of the
contact surfaces, applied no horizontal force or torque tangential to
the disks, and held the object level. Thus deviations from this nominal
force distribution could depend on the positions of the digits,
application of tangential horizontal forces and torques, and on object tilt.
Effects of changing surface structures
Previous work on precision grip control has shown that subjects adjust the balance between normal forces and tangential loads to the frictional conditions of the grasp in a manner that supports grasp stability (for references see INTRODUCTION). In our multidigit task, the friction between the digits and the object did indeed influence the coordination of normal forces and tangential loads from the load phase and onwards. During the load phase the surface combination influenced the slope of the relationship between normal force and load at the individual digits, whereas the parallel change in normal force and load remained regardless of surface combination (Fig. 2, A-D). For all five surface combinations, the median correlation between normal force and load were between 0.97 and 0.98 (data pooled across subjects and digits). Consistent with this result, the surface combination influenced the coordination of normal forces and load at the level of individual digits during the lift and hold phases. To further examine this coordination, we focused on subjects' behavior during the hold phase.
RATIO BETWEEN NORMAL FORCE AND LOAD (FN:L) DURING THE HOLD PHASE. As a measure of coordination of normal force and load we calculated the ratio between normal force and load (Fn:L ratio) at each digit and compared this "employed" ratio to the minimum Fn:L ratio required to prevent slips, linear or rotational. This minimum ratio, termed the slip ratio, represents the inverse of the coefficient of linear friction at the digit-object interface (see METHODS). On average, subjects adapted the Fn:L ratio at each digit to the local friction (Fig. 3A). The thumb and the index and middle fingers employed significantly higher ratios when they contacted the slippery rayon than when they contacted sandpaper [F(1,6) = 31.43; P < 0.01, F(1,6) = 7.28; P < 0.05 and, F(1,6) = 13.92; P < 0.01, respectively]. The slip ratios at the cooperating digits had small effects on the employed ratio in comparison to the effect of the local slip ratio (Fig. 3A). Subjects applied significantly higher ratios at the thumb when it contacted sandpaper and both cooperating fingers contacted rayon than when all digits contacted sandpaper [F(1,6) = 6.38; P < 0.05]. The index finger applied a slight higher ratio when it contacted sandpaper and the middle finger contacted the slippery material rayon [F(1,6) = 27.93; P < 0.01] and the middle finger a slightly higher ratio when all digits contacted rayon than when one of the cooperating digits or both contacted sandpaper [F(1,6) = 9.19; P < 0.05]. Figure 3, C and D, shows the regulation of employed ratios to local friction for individual subjects (and digits) by plotting the relationship between the slip ratio and the employed Fn:L ratio. The two subjects chosen for illustration showed large frictional differences between the two types of surface materials and therefore a correspondingly strong adjustment of the ratios.
|
|
FRICTIONAL EFFECTS ON NORMAL FORCES AND DIGIT LOADS. In this section we first describe and provide statistical analysis based on all subjects behavior and then describe the results for a representative subject in some detail. The combination of surface materials at the contact disks influenced the magnitudes of the normal forces [F(4,24) = 19.29; P < 0.001; main effect]. For all three digits, the normal force tended to increase as the number of digits contacting rayon increased (Fig. 4A). Subjects applied more normal force when one of the digits, index or middle finger, contacted rayon than when all digits contacted sandpaper [F(1,6) = 6.28; P < 0.05], but they applied even more normal force when both fingers contacted rayon [F(1,6) = 6.94; P < 0.05]. Likewise, subjects applied more normal force when all digits contacted rayon then when the thumb contacted sandpaper [F(1,6) = 25.03; P < 0.01). Thus subjects scaled the normal forces to the frictional condition by essentially changing the forces in parallel. The common scaling of normal forces to changes in the frictional conditions is demonstrated by the fact that the proportional distribution of the normal force is relatively constant across the surface conditions (Fig. 4B). There was, however, a small but significant interaction between digit and surface combination [F(8,48) = 8.76; P < 0.001], indicating that the surface combination did influence the distribution of normal force to some extent.
Furthermore, regardless of surface combination, the forces in the horizontal plane of the object were primarily normal forces. The arctangents of the ratios of normal force to horizontal tangential force were, on average, 87.3 ± 3.9, 86.2 ± 5.7, and 90.9 ± 5.4° for the thumb, index finger, and middle finger, respectively; data pooled across surface conditions. These were not significantly different from 90° (P > 0.2 in all 3 cases). Thus in agreement with our previous observations, subjects tended to avoid producing tangential force horizontal to the contact surfaces and, as a result, the force vectors in the horizontal plane intersected near the center of the object (Flanagan et al. 1999
|
|
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
In the present thee-digit manipulatory task, subjects either
lifted an object with the same surface material at all three contact
disks, one slippery and one less slippery material, or with a different
material at one of the three contact surfaces. First, when the digits
contacted the same surface material, subjects adjusted the fingertip
forces at all digits such that the ratio between the normal force and
load (Fn:L ratio) was
similar across digits and adapted by an adequate safety margin to the
minimum ratio at which slips would have occurred. This control policy, which supports grasp stability, was originally demonstrated for precision grip tasks engaging two digits (Johansson and Westling 1984b; also see Cadoret and Smith 1996
;
Flanagan and Wing 1997b
; Flanagan et al.
1995
; Forssberg et al. 1995
; Kinoshita
and Francis 1996
; Smith et al. 1997
). Second,
with different materials at the three contact surfaces, subjects
adapted the Fn:L ratios at the separate digits to the local frictional conditions. Again, this
behavior has previously been demonstrated in two-digit manipulatory tasks carried out by digits belonging to one hand, two hands, or to two
subjects (Birznieks et al. 1998
; Burstedt et al.
1997a
,b
; Edin et al. 1992
). Thus the present
findings strongly suggest the control for grasp stability by
adjustments of the Fn:L
ratios at individual digits to the local frictional condition is a
general control policy that supports grasp stability in manipulation. As previously demonstrated for two digit grasps, subjects principally adjusted the Fn:L ratios by
collectively scaling the normal forces to the "average" slip ratio
across the grasp sites and by partitioning the load among the digits.
However, it must be stressed that adjustments of the
Fn:L ratios at all digits
engaged is overruled in many manipulatory tasks by mechanical
constraints imposed by the task, the object, and the grasp
configuration. For instance, when subjects rotate an object held
between the thumb and the index finger by pronation and supination
movements, the Fn:L ratio
is regulated to the slip ratio with an adequate safety margin at only
one of the digits, whereas the ratio of the opposing digit that will support the object from underneath may be very high (Johansson et al. 1999
).
In the present study and our previous study dealing with three-digit
manipulation (Flanagan et al. 1999), we observed that subjects change the normal forces in parallel with changes in tangential loads during the load phase. Likewise, for a number of
two-fingered manipulative tasks, in which the load was primarily tangential force, it has been demonstrated that the normal force is
controlled to increase and decrease in parallel with changes in
tangential load (for reviews see Johansson 1996
;
Wing 1996
). More recently, we have demonstrated that
this type of coordinative constraint also applies when the load is
composed of combinations of tangential force and tangential torque
(Goodwin et al. 1998
; Johansson et al.
1999
; Kinoshita et al. 1997
). Because this
coordinative constraint appears to be expressed in all types of
manipulatory tasks requiring grasp stability, it seems to represent a
general rule in dexterous manipulation. Given that the
Fn:L ratios at the relevant
digits are adjusted to local frictional conditions, this coordinative
constraint effectively supports grasp stability in skilled manipulation
by ensuring that the normal force is above the minimum required to
prevent slip at any tangential load.
Coordination of forces among digits
The distribution of normal and vertical forces among digits
strongly reflected constraints imposed by the task (i.e., to grip the
object at the contact disks and lift it vertically) and by certain
properties of the test object (the geometric relationship between the
object's center of mass and the contact surfaces) (Flanagan et
al. 1999). The adjustments of the
Fn:L ratios to the changes
in the frictional conditions took place on top of this distribution and
involved changes in the magnitude of normal forces as well as
redistributions of load among the digits.
Previous work on precision grip control suggests that the controller
attempts to reduce fingertip forces, but without compromising grasp
stability. First, subjects regulate the normal forces both to
tangential load (Flanagan and Wing 1993,
1997a
; Johansson and Westling 1988
;
Westling and Johansson 1984
) and to frictional aspects
of the grasp (Goodwin et al. 1998
; Jenmalm and
Johansson 1997
; Johansson and Westling 1984
) in
a manner that results in a reasonably small safety margin against slips
over a wide parameter space, i.e., they avoid excessive normal forces.
Second, in two-fingered manipulatory tasks with different friction at
the two contact surfaces, when the tasks admits, subjects distribute
the load between digits in a manner that decreases the normal force
required to maintain grasp stability (Burstedt et al.
1997a
,b
; Edin et al. 1992
). Because this
behavior also applied to the subjects' performances in the present
multidigit manipulatory task, we propose that attempts to reduce
fingertip forces represent one general control role in dexterous
manipulation. As one alternative, subjects could have scaled the normal
forces to the friction at the most slippery contact while not changing
the distribution of load among the digits. However, this would require
greater normal forces also at the less slippery contact sites resulting
in inflated safety margins.
CONTROL OF NORMAL FORCES.
Subjects changed the magnitude of normal force with changes in the
frictional conditions of the grasp, whereas the distribution of normal
forces among the digits was modestly influenced. This behavior agrees
with that observed in previous studies of lifting tasks involving two
digits (Burstedt et al. 1997b; Edin et al. 1992
). In these two-digit tasks, however, the digits were bound to apply similar normal forces, whereas in the present three-digit task, the distribution of normal force among the digits was less constrained (Flanagan et al. 1999
). In fact, in our
three-digit task, subjects could have chosen quite different
distributions of normal force and still have been able to lift and hold
the object. This occurred during the fiddling procedure in which the normal force often reached quite low values at an individual digit without the subject losing the object. Yet, during the ordinary load
and hold phases, regardless of surface combination, subjects maintained
a force distribution that largely reflected the position of the digits
in relation to the center of mass of the object. Interestingly, we also
recently observed a parallel scaling of the normal forces to the
"average" friction at the engaged digits in a two-fingered
restraint task in which subjects were free to apply any force
combination (Birznieks et al. 1998
; Burstedt et al. 1997a
). Thus these findings suggest that a default control strategy in manipulation is to changes normal force in parallel at all
digits engaged.
DISTRIBUTION OF LOAD AMONG DIGITS.
In addition to a global scaling of the normal forces, an important
factor for the adjustment of
Fn:L ratios in response to changes in the frictional condition were changes in the distribution of
load across digits. Although the position of the digits on the contact
surface strongly influenced the distribution of vertical force among
the digits, subjects did not systematically vary the position of digits
to adapt the force coordination to changes in the frictional
conditions. Rather, the position of the digits appeared to be poorly
controlled in this respect. Likewise, the surface combination did not
systematically effect the tangential torques at the digit-object
interfaces. Because the load component accounted for by the tangential
torque was of similar magnitude regardless of surface combination (on
average ~20% of total load), we can safely conclude that subjects
did not rely solely on a "torque strategy" to adapt the
Fn:L ratios to local
frictional conditions. Changes in object orientation would also be a
helpful strategy for changing the distribution of load across digits, but the results indicate that object tilt could not fully explain the
observed changes in load distribution. Object tilt appears to be
secondary to redistributions of vertical forces among the digits in
response to local frictional changes (Burstedt et al. 1997b; Edin et al. 1992
). Thus the frictional
conditions of the grasp did not appear to systematically influence any
of the above factors (i.e., digit position, tangential torque, and
object tilt). However, the frictional conditions had a systematic
effect on the load distribution, and together these factors explained
nearly all of the variability in load observed across trials at the
level of individual digits. These results indicate that the changes in
load distribution resulted from interplay of these factors and that the
nature of this interplay varied between subjects and across trials
within subjects.
![]() |
ACKNOWLEDGMENTS |
---|
We thank A. Bäckström for technical support.
This study was supported by the Swedish Medical Research Council (project 08667), Department of Naval Research (Grant N00014-92-J-1919), the Göran Gustafsson Foundation for Research in Natural Sciences and Medicine, and the Canadian Natural Sciences and Engineering Research Council.
![]() |
FOOTNOTES |
---|
Address for reprint requests: M.K.O. Burstedt, Dept. of Integrative Medical Biology, Section for Physiology, Umeå University, SE-901 87 Umeå, Sweden.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 16 February 1999; accepted in final form 21 June 1999.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|