Exp Brain Res (1998) 118:286±292

 Springer-Verlag 1998

RESEARCH NOTE

Patrick Haggard ´ Alan Wing

Coordination of hand aperture with the spatial path of hand transport

Received: 22 January 1997 / Accepted: 16 September 1997

Abstract We have investigated the coordination of hand aperture with the spatial path of hand transport in prehensile movement by comparing straight prehensile movements with curved movements, in which subjects had to pass over a ªvia pointº marked on the worksurface before picking up an object in the target location. Spatial plots of hand aperture against hand transport showed that the preshaping of the hand to prepare an appropriate grasp was delayed in the curved movements relative to the straight movements, with most of the preshaping of the hand occurring after passing the via point, even when the via point occurred late in the course of the movement. The postponement of hand preshaping was apparently not due to subjects segmenting the movement into two completely separate portions preceding and following the via point, since some degree of hand opening often occurred before the via point. We suggest that the delay in hand opening in curved movements involves a scheduling process, which uses information about hand transport to set an appropriate hand aperture. Key words Prehension ´ Via point ´ Hand transport ´ Hand aperture ´ Scheduling ´ Human

Introduction Jeannerod (1981) suggested that prehensile movements involve two independent visuo-motor channels. The first channel uses information about the position of the target in egocentric space to move the hand towards the target. We shall use the term hand transport to refer to the operation of this channel. The second channel uses information about the object itself (e.g. its size, weight, compli-

)

P. Haggard ( ) Department of Psychology, University College London, Gower St, London WC1E 6BT, UK, e-mail: [email protected] A. Wing Medical Research Council Applied Psychology Unit, 15 Chaucer Road, Cambridge CB2 2EF, UK

ance, etc.) to preshape a grasping configuration of the hand that is appropriate to picking up the object. We shall use the term hand aperture to refer to this second channel, because the opening of the hand during the course of movement is a suitable behavioural measure of its operation. Jeannerod suggested that the hand transport and hand aperture channels were independent, in the sense that they dealt with different kinds of information, and the information used by one channel was not available to the other. Jeannerod proposed instead that the two channels merely shared synchronisation signals, ensuring that hand aperture begins to increase at the start of a movement and then closes to grasp the target object as hand transport reaches the location of the target object. Since Jeannerods original article, a number of results have suggested that the channel controlling hand aperture does have access to information about the progress of hand transport. These studies have typically used experimental manipulations to alter the hand transport component of prehensile movements and have looked for concomitant changes in hand aperture. Thus, Paulignan et al. (1991) observed that sudden changes in the apparent location of an object during the course of a prehensile movement produced rapid adjustments in both hand transport and hand aperture. Wing and colleagues (Wing et al. 1986; AthØnes and Wing 1988) showed that subjects opened the hand wider in conditions where they might be uncertain about the accuracy of hand transport, such as particularly fast movements, movements in the dark and movements that might be mechanically perturbed. We have also shown that the spatial relation between hand aperture and hand transport is highly regular and precisely controlled, both in normal movement (Haggard 1991), and in movements where the arm is mechanically perturbed as it approaches the object (Haggard and Wing 1991, 1995). In general, hand aperture does appear to be coordinated with hand transport, and the pattern of coordination depends both on cognitive factors such as the subjects degree of certainty about the accuracy of hand transport and lower-level factors responsible for maintaining a synergy between the two components.

287

Other studies have shown that the hand transport and hand aperture components of the movement may be modified as a result of a prior plan to adjust to changes in the circumstances of a movement. Wallace et al. (1990) recently investigated pre-planned changes in the hand aperture component of prehensile movements. They asked subjects to make reaching and grasping movements beginning with the hand in a normal closed posture or with the hand aperture already at 75% its maximum. They observed a ªtemporal constraintº in the relation between the hand aperture and hand transport components, such that the relative time to the closing velocity of hand aperture was similar despite the different initial postures of the hand. We speculate that the temporal regularity in their study arose because the initial hand aperture did not affect the hand transport component (cf. Haggard 1991). Since other studies have shown hand aperture to be influenced by hand transport parameters, we decided to investigate whether the hand aperture component of prehensile movements was adjusted for pre-planned changes in the way the hand transport component approached the target object. We reasoned that the spatial path by which the hand moved towards the object should have no direct effect on the hand aperture channel, since the intrinsic attributes of the object are unaffected by the path of approach. Therefore, any variations in hand aperture with changes in the spatial path of the hand must be an indirect effect, due to the transfer of information between the two visuomotor channels. The pattern of changes in hand aperture might indicate how the two channels are coordinated. An obvious way to alter the spatial path of hand transport involves asking subjects to pass over an intermediate target, or ªvia pointº, marked on the worksurface, as they moved towards the target object. The location of the via point alters the curvature of the hand transport component of the movement. Many studies of curved pointing and drawing movements have been reported (Flash and Hogan 1985; Viviani and Cenzato 1985; Viviani et al. 1987; Wann et al. 1988). All agree on the robust finding of a neative correlation between the instantaneous speed of the hand and the curvature of the spatial path. Viviani and colleagues have argued that changes in the value of a velocity gain factor in this relation (rather than the actual speed), can be used to psychologically segment such movements into a series of component paths. However, it is unclear whether similar segmentation rules apply to more complex actions such as reaching and grasping. We were particularly interested, therefore, to see whether hand aperture showed a segmentation effect as a result of hand transport curvature: such an effect would provide evidence for coordination of hand aperture with hand transport.

Materials and methods This experiment used via points, or intermediate targets, painted on the worksurface, to make subjects produce prehensile movements with varying degrees of curvature. Five via points were positioned in a series stretching along the start-target axis and 10 cm to the right of it, as shown in Fig. 1. An additional via point, marked in Fig. 1 as

Fig. 1 Schematic arrangement of workspace via point 0, was located on the start-target axis: a movement passing over this via point could be perfectly straight. Four subjects (three women, one man) participated in the experiment. They were aged between 23 and 35 and had no history of neurological disorders. One subject (A.J.T.) was left-handed, but nevertheless performed the task with her right hand. Subjects made movements passing over the via points to pick up a cylinder placed in the target location. Each trial began with the lateral edge of the subjects right hand resting on the worksurface over the starting point. Subjects were instructed to begin each trial with the pads on thumb and forefinger touching, and to reach out and pick up the dowel using a pincer grip after a ªgoº instruction from the experimenter, passing over the via point en route. No specific instructions were given about preshaping the hand. Subjects were instructed to make all their movements as smooth as possible. They were asked to try to make the via point the apex of the hands movement. They were particularly instructed not to stop their movement or touch the worksurface as they passed over the via point. Having grasped the cylinder, the subject lifted it about an inch off the worksurface, replaced it in its original location and returned to rest their hand on the start position, in readiness for the next trial. The experimenter verified the position of the dowel after each trial. Each subject practiced making some curved movements a few times before the start of the experiment. Subjects had no difficulty in passing over the via point with reasonable accuracy and in picking up the cylinder. However, the accuracy of passing over the via point was not quantified at the time of the experiment. Each subject made ten movements over each via point, with all ten movements occurring as a single experimental block. Subjects were told before each block which via point to use in the block, and a small marker was placed on the relevant via point for that block to remind subjects which via point was designated. The order of the blocks was randomised, with each subject receiving a different random sequence. There were thus six blocks, each corresponding to a different via point condition. Additional conditions, in which subjects made movements passing over the same via points and then knoecked over a cylinder at the target location, as opposed to grasping it, are not reported here. Movements were recorded using a Watsmart optoelectronic tracking system (Northern Digital, Waterloo, Ontario, Canada; see Haggard and Wing 1990). The start-to-target axis of the movements was approximately aligned with the Watsmart y-axis. Infra-red emit-

288

Fig. 2 Spatial paths of typical movements by subject D.F.M. in the straight movement condition and in each of the five curved movement conditions. Notice how the apex of the curved movements comes at a different stage in the movement, depending on the condition. Each point represents a sample taken at 100 Hz ting diodes (IREDs) were attached to the target dowel, to the medial surface of the distal interphalangeal joint of the index finger and thumb and to the wrist, using micropore surgigal tape. The IREDs were sampled at 100 Hz. The data were stored on a computer for subsequent analysis.

Results Our analyses concentrated on the effects of curvature on the hand transport and hand aperture components of the movement. The hand transport component was taken as the position of the thumb IRED, and the hand aperture component was taken as the horizontal planar distance between the thumb and finger IREDs. Hand transport and hand aperture signals were smoothed using a Fourier domain filter with a half-cosine transition band extending from 8 to 21.3 Hz (Teulings and Maarse 1984). Twelve trials were discarded out of the total of 240 recorded, because of unreliable reconstruction of signals from one or more IREDs. Hand transport Figure 2 shows the spatial paths of the IRED mounted on the thumb in a typical trial in each condition by subject D.F.M. Each datapoint corresponds to a sample at 100 Hz. Inspection of many such trials showed that subjects passed quite accurately over the via point and were able to make smooth, fluent movements to pick up the target dowel successfully in all conditions.

Fig. 3 A Tangential velocity profiles of the thumb in typical straight movements for each subject. Note the single-peaked bell shape of the profiles during the reaching phase of the movement. B Tangential velocity profiles from typical trials in each of the curved movement conditions

The speed of the typical movements shown in Fig. 2 can be inferred from the spacing of adjacent samples. Figure 3 shows explicitly the tangential velocities of typical straight trials (Fig. 3A) and typical curved trials (Fig. 3B) for each subject. The straight trials show an essentially bell-shaped, single-peaked tangential velocity profile during the reaching portion of the movement, while the curved trials show a bimodal pattern, with a local minimum corresponding to the time of passing the via point. These patterns are similar to those observed by Flash (1987) in pointing movements. The tangential speeds in the two conditions were quite similar throughout a large portion of the duration of the movement. The five curved movements shown in Fig. 3B reached their effective via point (the sample at which the lateral deviation of the hand from the start-target axis was maximal) in 580±700 ms, a quite narrow range given

289 Fig. 4 A Spatial plots of hand aperture against hand transport for the same typical straight movements shown in Fig. 3A. Note the smooth, continuous opening of the hand as hand transport progresses. B Spatial plots of hand aperture against hand transport for the same typical curved movements shown in Fig. 3B. A filled circle marks the apex of the curve (i.e. the effective via point) for each movement. Note that opening of the hand is delayed relative to the movements shown in A and is sometimes postponed until after the via point

the spread of the via points throughout the entire movement. This finding extends to prehensile movement the isochrony principle developed by Viviani et al. (1987) for freehand drawing movements. This principle states that points of high curvature naturally divide a trajectory into sections of approximately equal duration, with the duration being largely independent of the amplitude of each section owing to the tendency for larger-amplitude sections to exhibit higher movement velocities. Hand aperture Inspection of hand aperture in the straight movement condition (via point 0) showed that the hand began to open to preshape a suitable grasp as soon as movement towards the target began. Hand aperture then increased smoothly to a single clear peak a few centimetres short of the target and then closed, much more slowly, to grasp the dowel between thumb and forefinger. A similar pattern has previously been observed by many researchers (Jeannerod 1981; Wing et al. 1986; Marteniuk et al. 1987). We chose to represent this pattern as a spatial plot of hand aperture against the forward progress of hand transport. Figure 4A shows spatial plots for the same straight movements seen in Fig. 3A. The abscissa in Fig. 4A uses the coordinates of the Watsmart system, relative to an arbitrary origin. Notice that subjects begin and end the movement at slightly different values on the abscissa, because of individual differences in the initial and final posture of the hand. This spatial pattern shown in Fig. 4A was altered in curved movements. In curved movements, preshaping often began later than in straight movements. Figure 4B shows spatial plots for the same movements as seen in Fig. 3B. The effective via point at the apex of the thumbs curve is marked by a solid dot in each case. In some cases (.e.g. trace LAJ V2 in Fig. 4B) almost no hand opening occurred until after passing the via point. In most trials, the initial opening of the hand was delayed (e.g. trace AJ V3) or the slope of the spatial plot was more gradual than for that subjects straight movements (e.g. traces KAH V4 and KAH V5). In general, the opening of the

hand in preparation for grasping seems to be delayed. Although this delay appears to be associated with the curve in the spatial path of hand transport, some degree of hand opening can occur prior to the effective via point. This last result indicates that subjects did not psychologically decompose the curved prehensile movements into a strictly serial combination of a pointing movement to the via point, followed by a prehensile movement towards the target. Figure 4B also suggests a regression to the mean effect (Poulton 1974) on the curvature of hand transport: although the actual via points were spaced over a 25-cm range, the effective via points, or apices of the movements shown in Fig. 4B, fell within a narrower range around the middle of the movement. This finding is quite compatible with the subjects passing over the via point as instructed, merely showing that the actual marked via point was not always the site of the maximum deviation from the starttarget axis. We investigated the delay in hand opening statistically by calculating the displacement in the forward direction from the start of each movement at which the hand aperture exceeded its initial value in that movement by 1 cm. Large, positive values of this variable thus indicate a delay in hand opening until a late point in the spatial path of the movement. A repeated-measures ANOVA, using condition as a between-subjects factor with six levels corresponding to the six via points marked in Fig. 1, revealed a highly significant main effect of condition (F5,15 =6.09, P=0.028). Post hoc Ryan-Einot-Gabriel-Welch multiplerange testing (Einot and Gabriel 1975) disclosed two significantly different groups of conditions at the 5% level. The first group comprised via points 0, 1 and 2, and the second, via points 2Ð5. Table 1 shows the mean location of 1-cm hand opening for each subject and each condition. Since each subjects movements were made in a slightly different portion of the Watsmart workspace, we have shown the difference in thumb position at hand opening (in millimetres) between each of the curved conditions and the straight condition, thus removing the arbitrary origins of the Watsmart coordinate system. The data for each subject are shown in a separate column. Taking each subjects

290 Table 1 Difference in location of 1-cm hand opening (mm) by subject. Delayed opening of the hand in curved movements. Values give the difference in millimetres between the hand transport position at which the hand opened by 1 cm above its initial value, in each curved movement condition and in the straight movement condition (via point 0)

Table 2 Movement times (milliseconds) for each subject in each condition

Condition

D.F.M.

A.J.T.

K.A.H.

L.A.J.

1 2 3 4 5

18.1* 24.4* 18.7* 31.6* 19.2*

7.5 11.8 16.8* 17.7* 17.6*

24.9* 22.7* 24.9* 28.3* 12.6*

32.3* 31.3* 32.5* 41.6* 33.8*

0 1 2 3 4 5

vs vs vs vs vs

0 0 0 0 0

* Significant at 5% or lower for one-tailed Dunnetts t-tests, using straight movements as a control

straight movements as a control condition, we have shown one-tailed significance values of Dunnetts t-statistic for each comparison, using asterisks. Eighteen of the twenty comparisons shown in Table 1 were significant at the 5% level or above, supporting the hypothesis that hand opening occurs later in the course of curved movements than in straight movements. Table 1 shows, however, that the delay in hand opening is quite subtle. Although the via points are 50 mm apart, the delay in hand opening typically increases by just a few millimetres from one condition to the next. Second, the increase is not strictly monotonic: the delay in hand opening when passing over via point 5 is always less than that when passing over via point 4. Finally, there are substantial differences between subjects, both in the extend of the delay, and in the order of delays for different via points. In summary, the delay in hand opening is a robust consequence of curved movement, but is not proportionately related to the locus of curvature. Could the delay in hand opening be an artefact arising from some other difference between straight and curved movements, rather than a genuine effect? The most obvious candidate for such an artefact is the confound between spatial path and movement time: curved movements are likely to exhibit longer movement times than straight movements, because of the additional deviation required to pass over the via point. Were hand opening to begin at a fixed interval before the anticipated arrival at the final target (for example, using a time-to-contact strategy), the delayed opening observed here would result solely from the difference in movement time. Although previous studies have shown that hand aperture begins as soon as the hand moves forward in straight movements at very different speeds (Wing et al. 1986), we felt it important to exclude this possible explanation. We therefore calculated the movement time for each trial by inspection and interactive cursor measurement of the position of the wrist IRED. The mean movement times are shown in Table 2 for each subject and condition. As expected, a significant ANOVA effect of via-point location on movement time was observed (F5,15 =13.46, P