See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/223157984

Anticipating dynamic loads in handling objects

Article · January 1998

CITATIONS READS

15 24

2 authors:

Alan Miles Wing John Randall Flanagan

University of Birmingham Queen's University
263 PUBLICATIONS 8,348 CITATIONS 136 PUBLICATIONS 9,396 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

School of Psychology, Birmingham University View project

All content following this page was uploaded by Alan Miles Wing on 10 August 2017.

The user has requested enhancement of the downloaded file.

 ANTICIPATING DYNAMIC LOADS IN HANDLING OBJECTS.

Alan M. Wing J. Randall Flanagan

Sensory Motor Neuroscience Centre, Department of Psychology,
The University of Birmingham, Queen's University,
Edgbaston, Birmingham, B15 2TT, UK. Kingston, Ontario, Canada.
[email protected] [email protected]

DSC-Vol 64, Proceedings of the ASME Dynamic systems and Control Division, ASME 1998, 139-143

SUMMARY
In this paper we review a set of studies showing that when people To study force coordination in lifting Johansson and Westling (1984)
pick up and move an object they continually adjust their grip force in asked subjects to use a precision grip to lift an instrumented
order to stabilize the object in the hand. These grip force adjustments manipulandum a small distance off a support while the surface
occur simultaneously with or slightly ahead of fluctuations in load slipperiness was systematically varied (in decreasing order of
forces and torques related to moving the object. They may therefore slipperiness - silk, suede, sandpaper). In all cases they noted that grip
be seen as anticipatory and we argue that a key line of research in force started to rise before load force. Load force then started to rise
manipulation should be to understand the integration of sensory and, provided subjects were familiar with the lifting conditions, the
motor information to build an internal model of the object and the ratio of grip force to load force decreased to a value somewhat above
effector system to support such anticipation. the minimum, given the surface, required to prevent slip (thus defining
a 'safety margin' against slip, Westling and Johansson, 1984). As load
LIFTING AND HOLDING force continued to rise the force ratio remained approximately
Consider picking up and holding an object with parallel sides, such as constant until the object was lifted from the surface. Although quite
a glass tumbler, using a precision grip with the thumb and one or more different levels of grip force were required for the different surfaces,
fingers pressing in on opposite sides. Frictional force, generated by Johansson and Westling found that the time taken to lift the
grip force normal to the surfaces, is used to overcome load forces manipulandum off the surface was relatively constant, provided
tangential to the surface which would otherwise result in the hand subjects knew the surface in advance. This timing constancy was
slipping over the surface of the glass. Up to the point at which the achieved by scaling the rate of increase of grip force from the outset
glass lifts off the surface the load force depends on the lift developed and constitutes strong evidence of anticipatory control.
by the arm. After lift-off the load force is generally determined by the MOVING
combined effects of gravity and any acceleration of the hand used in
moving the object. In order to stabilize the object in the hand and During static holding, grip force typically exceeds the minimum to
prevent slip a minimum level of grip force is required which is a prevent slip by a safety margin which can be as small as 10%. In
function of the tangential load and the frictional properties of the moving an object, peak accelerations can easily reach twice the
hand-object interface. acceleration due to gravity, which results in a significant fluctuation in
load force superimposed on the steady contribution of gravity. We
With a wealth of sensory information available from tactile receptors have studied modulation of grip force associated with load force
in the skin, it might be thought that stabilization of an object in the fluctuations due to arm movements when holding an object in
hand would involve feedback adjustment of grip force. As the arm precision grip (Flanagan et al, 1993; Flanagan and Wing, 1993, 1995).
muscles begin to develop forces to lift the hand and object off a In the case of vertical movements, the summation of inertial and
support surface or to accelerate the hand and object from rest, the gravitational forces results in load force functions that reach their
resulting shear or tangential load force at the finger tips might drive an maximum near the start of upward movements but near the end of
increase in grip force. However, the reflex pathways involved would downward movements. Because grip force functions for up and down
introduce a delay of around 80 ms and are too slow to guarantee grasp movements reflect this contrast, Flanagan and Wing were able to argue
stability. Instead, in familiar situations where loading is predictable, that grip force adjustments are anticipatory of loading and not simply
grip force is typically adjusted in phase with changes in load force. linked to movement per se (see also Wing, 1996).

1
Fig 1: Modulation of grip force when moving an inertial load held between the tips of the thumb and index finger. The load force pattern varies
depending on whether the object is moved up (left) or down) and this is reflected in the grip force pattern. Illustrative single trial data adapted from
Flanagan, Tresilian & Wing, (995).

In the previous study the load force was inertial and so directly internal model of the dynamics of the motor apparatus and external
related to acceleration and deceleration. The effect of load force load.
functions related to other parameters of movement on grip force was In holding an object with a precision grip, the grasp axis joining the
examined by Flanagan and Wing (1997). Subjects grasped a points at which the two digits make contact with the object may not
manipulandum that was mounted on a linear motor. The motor could be accurately aligned with its centre of mass. In such cases
be servo-controlled to produce load force functions that, in different gravitational and/or inertial forces will result in torques at the digits
blocks, were primarily inertial, viscous (load proportional to velocity) and stabilization of the object is a matter of preventing it rotating, as
or elastic (load proportional to position). It was observed that the well as translating, within the hand. Wing and Lederman (In Press)
form of the hand trajectory changed with load for some subjects but have shown that subjects anticipate, and attempt to compensate for,
not for others. However, under all load conditions and in all subjects, such load torques when lifting or moving an object with the hand by
grip force was modulated in parallel with, and thus anticipated, increasing grip force in proportion to the distance between the grasp
fluctuations in load force (see Fig. 2). The adaptation of grip force to axis and the centre of mass. Again an internal model of the effector
changes in load force was taken as evidence of the operation of an system and external load would provide abasis for such prediction

Fig 2: When force at the hand (HF) is related to acceleration, velocity of position, grip force (GF) modulation is adapted to different load force (LF)f
unctions . Illustrative single trial data for one subject (Flanagan & Wing, 1997).
ADAPTING TO CHANGES IN LOAD In Johansson and Westling's (1984) experiment on lifting, the
slipperiness of the contact surfaces of the manipulandum was
Anticipatory stabilization of hand-held objects is evident when systematically varied. After a block of trials with one surface the
subjects are familiar with the conditions and/or vision provides cues experimenters changed the surface for the next block of trials. In this
sufficient to retrieve information remembered from previous situation the grip force initially rose at a rate appropriate to the
experience with the conditions. However, an important question is surface used in the previous block of trials but was then adjusted.
how that information is acquired and this leads to consideration of When the change was to a more slippery surface, an inadequate level
adaptive changes in performance when task conditions change. of grip force resulted in a small slip which led to a rapid corrective

140
increase in grip force. On the next trial the grip force rate would rise one subject over the first 8 trials (a-h) immediately after changing from
appropriately from the outset. Subsequently, Johansson and Westling the inertial to the viscous load. It will be observed that the load force
(1988a) showed similar adaptation to unexpected changes in object function evolves from a multi-peaked profile to a single smooth
weight. Findings such as these led to the suggestion that force function which is very similar to the superimposed profiles from the
coordination in lifting is based on sensory motor memories which last 10 trials (i). This suggests the subject was gradually becoming
represent physical properties of objects and appropriate magnitude familiar with the changed dynamics of the system.
parameters of the motor commands (see Johansson, 1996). These It is particularly interesting to note that despite the major changes in
memories are presumably updated whenever inappropriate motor load force function, the match between grip force and load force is
commands based on erroneous information result in unintended good from the second or third trial. This tendency was observed in
mechanical events such as slip. several subjects and the lower part of Fig. 3 summarizes this in terms
Flanagan and Wing's (1997) analysis of grip force adaptation to of correlations (averaged across the group) between grip force and load
different load force functions focused on stable performance. force functions (j) and between the load force function on a given trial
However, short-term changes in hand trajectory were seen and the average of the last 10 trials (k). The latter takes longer to settle
immediately after the change in load force condition (Flanagan and at a steady value than does the correlation between grip force and load
Wing, 1996) and certain interesting points emerge from consideration force.
of such changes. Figure 3 shows grip and load force and their ratio for

Fig 3: Adaptation to viscous load: (a-h) Load force functions converge on form evident in last 10 trials (i); illustrative single trial data from one
subjet. Group average correlations (j-k) between (j) LF on the first 20 trials and the average of trials 20-30 (k) GF and LF suggest GF and LF
function develop similarity more rapidly than the rate at which LF converges on its final form (Flanagan & Wing, 1996)

141
Fig 4: Forward model for learning arm movement control and predicting the grip force function necessary for object stabilization (Flanagan & Wing,
1996).

One interpretation of this finding is that the hand controller was able exhibited impaired coordination of grip force and load force in lifting
to predict the load force before learning the commands required to objects. Moreover they failed to scale their grip force rates to different
generate smooth movements. A possible scheme based on Jordan and loads. A possible interpretation therefore is that the performance
Rumelhart (1992) is shown in Figure 4 (Flanagan and Wing, 1996). As impairment reflects a flawed internal model due to the cerebellar
in their original exposition, muscle commands from the controller are damage.
entered into a forward model which predicts sensory consequences.
The discrepancy between observed and predicted sensory CONCLUSIONS
consequences can then be used to train the controller. However, the Anticipatory adjustments of grip force in lifting and moving objects
output of the forward model might also be used to determine the level suggest the operation of an internal model of the effector system and
of grip force required to provide a stable grasp. In this perspective, the object. An important function of sensory feedback during object
the importance of sensory feedback is not just in reacting to manipulation may therefore be to provide a basis for maintaining the
unpredictable load forces but also for updating and fine-tuning internal model under changing environmental conditions.
anticipatory control mechanisms in case the unpredictability reflects
REFERENCES
changes in the dynamics of the environment which, once learned, will
allow control to revert to an anticipatory basis. Flanagan JR, Tresilian JR & Wing AM (1993) Coupling of grip force
and load force during arm movements with grasped objects.
Given the evidence favouring the existence of an internal model it is
Neuroscience Letters, 152, 53-56.
natural to ask what are the underlying neural mechanisms? Miall et al
Flanagan JR, Tresilian JR & Wing AM (1995) Grip force adjustments
(1993) have suggested a cerebellar contribution to learning of hand
during rapid hand movements suggest that detailed movement
aiming movements in the form of two internal models. A forward
kinematics are predicted. Behavioral and Brain Sciences 18: 753-754
model of the motor apparatus predicts the sensory consequences of
Flanagan, J.R & Wing, A.M. (1993) Modulation of grip force with
motor commands and a second model predicts time delays in the
load force during point to point arm movements. Experimental Brain
control loop in order to delay the predicted sensory feedback so that
Research, 95, 131-143.
it can be directly compared with actual sensory feedback. The error
Flanagan, J.R. & Wing, A.M. The stability of precision grip forces
signal from this comparison may then be used to modify motor
during cyclic arm movements with a hand-held load. Experimental
commands and update the first model. Miall et al describe how this
Brain Research, 1995, 105, 455-464.
model predicts certain failures of hand aiming movements after
Flanagan, J.R. & Wing AM (1996) Internal Models of Dynamics in
cerebellar damage. If the internal model for arm movement also
Motor Learning and Control. Paper presented at the 26th Annual
provides the basis for anticipatory adjustment of grip force cerebellar
Meeting of the Society for Neuroscience, Washington, DC, Soc
damage might also be expected to affect grip force control. Some
Neurosci Abstr, 22, Part 2, p 897
support for this suggestion comes from a study by Muller and
Dichgans (1994). Patients with degenerative cerebellar conditions

142
Flanagan, J. R., & Wing, A.M. (1997). The role of internal models in
motion planning and control: evidence from grip force adjustments
during movements of hand-held loads. Journal of Neuroscience, 17,
1519-1528.
Johansson, R.S. (1996). Sensory control of dexterous manipulation in
humans. In A.M. Wing, P. Haggard & J.R. Flanagan (Eds) Hand and
Brain: The neurophysiology and psychology of hand movements. San
Diego: Academic.
Johansson, R. S. & Westling, G. (1984). Roles of glabrous skin
receptors and sensorimotor memory in automatic control of precision
grip when lifting rougher or more slippery objects. Experimental Brain
Research, 56, 550-564.
Johansson, R. S. & Westling, G. (1988a). Coordinated isometric
muscle commands adequately and erroneously programmed for the
weight during lifting with precision grip. Experimental Brain Research,
71, 59-71.
Johansson, R. S. & Westling, G. (1988b). Programmed and triggered
actions to rapid load changes during precision grip. Experimental Brain
Research, 71, 72-86.
Johansson, R. S. (1996) Sensory control of dexterous manipulation in
humans. In A.M. Wing, P. Haggard & J.R. Flanagan (Eds) Hand and
Brain: The neurophysiology and psychology of hand movements. San
Diego: Academic.
Jordan, M.I. & Rumelhart, D.E. (1992) Forward models: supervised
learning with a distal teacher. Cognitive Science, 16: 307-354.
Miall, R.C., Weir, D.J., Wolpert, D.M. & Stein, J.F. (1993) Is the
cerebellum a Smith predictor? Journal of Motor Behavior, 25, 203-
216.
Westling, G. & Johansson, R. S. (1984). Factors influencing the force
control during precision grip. Experimental Brain Research, 53, 277-
284.
Wing, A.M. (1996) Anticipatory control of grip force in rapid arm
movement. In A.M. Wing, P. Haggard & J.R. Flanagan (Eds) Hand
and Brain: The neurophysiology and psychology of hand movements.
San Diego: Academic.
Wing, A.M. & Lederman, S. Anticipating load torques produced by
voluntary movements. Journal of Experimental Psychology: Human
Perception and Performance (In Press).

143

View publication stats