The effects of task constraints on the organization of interception ...

Jour nal of Sports Sciences, 1997, 15, 581- 586

The effects of task constraints on the organization of interception m ovem ents VA L EÂ R IE FAYT, 1 R E IN O U D J. B O OT SM A , 1* RO N A L D G. M A RT E N IU K , 2 C H R IST IN E L . M AC K E N Z IE 2 and M ICH E L L AU R E N T 1 1

U M R M ovement & Perception, U niversity of the M edite rranean and C NRS, M arseille, France and 2School of K inesiology, Faculty of Applied Sciences, Sim on Fraser U niversity, B ur na by, B C, Canada

In the light of the intensity coupling effects reported in the literature, subjects’ capacities for independently controlling the absolute velocity of their movement at the point of interception was evaluated by asking them to learn to propel orthogonally approaching balls, varying in their speed of approach, into target boxes placed at various distances from the interception point. As required for successful accom plishment of the task, m ovement velocity was found to be adapted to the distance to be covered, with the near target giving rise to lower velocities and the far target giving rise to higher velocities even when the conditions were presented in random order. Nevertheless, even though target distance accounted for a substantial proportion of the total variance, a small but signi® cant effect of ball approach speed on movem ent velocity was found, suggesting that intensity coupling is a persistent but modulable phenom enon in interception tasks. K eywords : interception, motor control, task constraints.

Introduction T he consistently high levels of perform ance demonstrated by exp ert racquet ball gam e players are a result not only of their capacity to respond in a consistent way, but also of their ability to vary the velocity of their m ovem ent so as to suddenly break the rhythm of the ongoing exchange. Thus, experts are capable of demonstrating both stability, indicated by low levels of movem ent variability over repeated executions, as well as ¯ exibility, evidenced by the adaptation of the m ovem ent pattern to the prevailing conditions. W hereas theories of motor control initially focused on the stability characteristics only (e.g. Keele, 1968; Tyldesley and W hiting, 1975), recent theoretical developm ents have suggested that stability, rather than re¯ ecting the execution of stored m ovem ent patterns, results from the consistent applications of laws of control linking information and movem ent (e.g. Peper * Address all correspondence to Reinoud J. Bootsma, UM R M ovem ent & Perception, Faculty of Sport Sciences, University of the M editerranean, 163 Avenue de Lum iny, CP 910, 13288 Marseille Cedex 9, France. 0264 - 0414/97

© 1997 E & FN Spon

et al., 1994; Sch Èoner, 1994). In the latter perspective, adaptation of the m ovem ent pattern results naturally from a perceived change in the situation. O ne aspect that has not received m uch attention in the interception literature concerns the role played by task constraints in the organization of the kinematic patterns produced. M ost experim ental evidence cited as demonstrating the consistency of m ovem ent patterns in interceptive actions (e.g. Hubbard and Seng, 1954; Tyldesley and W hiting, 1975; Franks et al., 1985; Bootsma and van W ieringen, 1988, 1990) in fact stem s from studies in w hich the task required near m axim al m ovem ent velocities at the m om ent of contact w ith the ball (Li, 1995a). In the few studies that have addressed the kinem atics of m ovem ent in interception tasks with lower levels of constraints, a consistent ® nding is the adaptation of the m ovem ent velocity to the speed of the approaching ball (De V ries, 1992; Li and Laurent, 1994, 1995; Fayt, 1995; Li, 1995a,b). T his linking of the velocity of m ovem ent execution to the speed of the stim ulus was denoted `intensity coupling’ by Li and Laurent (1995), who suggested that it re¯ ects a basic behaviour in interception tasks, because

Fayt et al.

582 it was observed even when the m ovem ent characteristics required for successful com pletion of the task were constrained to be constant. W hile such coupling m ight be interpreted as re¯ ecting an in¯ uence of perceived speed of approach on the organization of m ovem ent, this need not be the case. In the continuous control m odels for interception proposed by, for instance, Bootsm a et al. (1997) and Peper et al. (1994), the dependence of m ovem ent velocity on the speed of approach of the ball emerges from the control law s postulated without the speed of approach entering as an input variable. Although exp erim entally observed intensity coupling effects can thus be accounted for by the aforem entioned type of m odel, the ver y existence of such effects would, at ® rst sight, seem to be incompatible with the attainm ent of success in tasks in which the velocity of movem ent at the m oment of contact needs to be separated from the speed of approach. To appreciate better the span of task constraints that can be included in the m odel, this study exam ined how subjects deal with a constraint on absolute movem ent velocity at the m om ent of contact in the context of var ying ball approach speeds. Basically, two options exist. Either the intensity coupling phenom ena reported are so strong that, no m atter w hat the demands im posed by the task, the attainm ent of success is subordinate to a coupling of m ovem ent velocity with the approach speed of the ball, or, perhaps as a function of practice, subjects are able to override such a coupling, effectively rendering their m ovem ent independent of the m ovem ent speed of the ball.

F igure 1

M ethod Subjects Eight right-handed students (4 females and 4 m ales aged 24- 32 years) from the School of K inesiology of Sim on Fraser University volunteered to participate in the experiment. All subjects were naive with respect to the speci® c aim of the experim ent and reported som e experience of either tennis or squash. Apparatus A linear alum inium trackway (3 m long, triangular in section with sides 7 cm long) was positioned on a standard laboratory table at a slope of 11¡ with respect to the horizontal plane, supported at its higher end (furthest from the subject) by an inclined wooden plane. An alum inium trolley, equipped w ith a platform m aintained parallel to the horizontal plane, could be released to m ove along the track. An alum inium tube (2.5 cm diam eter, 10 cm high) was ® xed to the platform . T he tube, which had a sm all m agnet attached to its upper end, carried a table tennis ball, kept in place by the m agnet during the approach by m eans of a sm all piece of m etal glued to the ball. T he trolley ran into a stop at the lower end of the track, 20 cm beyond the designated interception point. By changing the release position of the trolley (234, 156 and 104 cm from the interception point), three ball speeds (100, 90 and 80 cm s -1 at the interception point) were available (Fig. 1). A linear wooden trackway (1 m long, 15 cm high) was placed perpendicular to this track, directly in front

Schematic representation of the apparatus (overhead view).

583

Task constraints in interception of the interception point. A m anipulandum with a handle (4 cm diam eter, 12 cm high), which supported a horizontally positioned wooden lance (1 cm diam eter, 12 cm long), was used to hit the approaching ball. Subjects were seated to the left of the interception device (holding the manipulandum with the right hand), with their body oriented parallel to the ball trackway, allowing the interception movem ent to be carried out in the subject’s sagittal plane. Beyond the interception point, a box (46 cm w ide, 30 cm long, 10 cm high) was placed at a height of 50 cm (25 cm below the surface of the table top), which ser ved as a target. Two different target positions were used: the centre of the target box was placed at a horizontal distance of either 20 or 95 cm from the interception point. A three-cam era O PTOTRAK system (Northern D igital Inc., Waterloo, O ntario), controlled by an IBM PC, was used to record the data. T he O PTOTRAK system is able to measure three-dim ensional m ovem ents through the use of infra-red em itting diodes (IRED s). Two IRED s, one placed on the extrem ity of the striking device and the other placed on the trolley transporting the ball, were used to m onitor the displacem ents of both. D ata were recorded at a sam pling rate of 200 H z. For each trial, raw position data were ® rst converted to three-dim ensional data on the IB M -PC and then analysed on a UN IX-b ased com puter using the WAT SM ART (WATerloo Spatial M otion Analysis and Recording Technique) program . W here possible, m issing data points were reconstructed by m eans of interpolation. The data were organized so that trolley displacem ent was on the x-axis and perpendicular hand m ovem ent was on the y-axis. T he data were ® ltered using a second-order dual-pass Butterworth ® lter with a cut-off frequency of 8 H z. Procedure Subjects were seated facing the ball trackway. T hey were asked to hold the handle with their right hand and place the m anipulandum on the start position (indicated by a m ark), 35 cm from the interception point. After a `ready’ signal from the experim enter, the trolley was released and the subjects were required to hit the m oving ball into the target box. N o instructions were given about the kinematic aspects of the m ovem ent to be perform ed, other than that one sm ooth striking m ovem ent was to be made. Subjects were random ly assigned to one of the two exp erim ental groups (four subjects per group). Both groups received training under both experim ental target location conditions, w ith one group starting w ith the near target and the other group starting with the far target.

After a short fam iliarization period (® ve trials), subjects practised the task in one target location condition for a total of 90 trials (divided into six blocks of 15 trials) and then switched to the second target location for another 90 trials (six blocks of 15). T he ® rst block of 15 trials (containing 5 trials under each of the three velocity conditions, presented in random order) for each target location was used to evaluate performance before practice under that speci® c condition, while the last block of 15 trials served to evaluate the im m ediate effects of practice. Practice sessions were separated by 2 m in rest periods. The experim ent ended w ith a ® nal test, adm inistered 5 m in after the last practice session, consisting of 30 trials, where all conditions were presented in a com pletely random ized order, thus m ixing both target positions (2) and ball speeds (3). Design and analysis The data of interest concerned m ovem ent outcom e and m ovem ent kinem atics. The number of balls hit into the target box was used as the m easure of perform ance outcom e, whereas movem ent velocity at the interception point was the relevant m easure for the movem ent kinematics. M eans over the ® ve trials of each velocity condition were analysed for the practice sessions and the ® nal test separately. Pre- and post-practice perform ance was analysed using a 2 (groups) ´ 2 (target position: near vs far) ´ 2 (test: pre- vs post-) ´ 3 (ball speed: slow, m edium , fast) design, with repeated m easures on the last three factors. Perform ance on the ® nal test was analysed using a 2 (groups) ´ 2 (target position: near vs far) ´ 3 (ball speed: slow, m edium , fast) design, with repeated m easures on the last two factors.

Analysis of results The in¯ uence of the experim ental factors on the dependent variables (perform ance and m ovem ent velocity at the interception point) was tested by m eans of analysis of variance (AN OVA) on the m eans of the ® ve trials realized for each experim ental condition. W here appropriate, post-hoc tests were perform ed on Fratios that reached the level of signi® cance adopted ( a = 0.05). All signi® cant effects are reported; for each effect, the intensity of the effect (IE), which expresses the percentage of the total sum of squares accounted for, is also presented. Outcome performance The analysis of variance on outcom e perform ance (calculated as the percentage of balls hit into the target

Fayt et al.

584

F igure 2 M eans of the outcome performance measure as a function of target location and ball speed conditions for (a) the pre- and post-practice sessions and (b) the ® nal test session.

box) demonstrated signi® cant m ain effects for target position (F 1,6 = 5.76, P < 0.05, IE 4.5% ; m eans 58.3 and 46.7% for the near and far targets, respectively), ball speed (F 2,12 = 4.06, P < 0.05, IE 5.6%; m eans 61.2, 50.6 and 45.6% for the slow, m edium and fast speeds, respectively) and test (F 1,6 = 6.98, P < 0.05, IE 9.1%; m eans 44.2 and 60.8% for the pre- and posttest, respectively). Thus, as shown in Fig. 2a, outcom e perform ance increased w ith practice, although approach speed and target location continued to affect perform ance, with subjects demonstrating lower levels of perform ance when they had to reach the far target and w hen the ball approached faster. T he absence of any signi® cant effects for group suggests that transfer from one target location to the other did not occur, but this needs to be quali® ed in the light of the relatively sm all num ber of subjects in each group. W hen, after practice, subjects were tested with all exp erim ental conditions random ly interspersed, m ain effects were found for target position (F 1,6 = 7.29, P < 0.05, IE 12.5% ; m eans 84.2 and 70.0% for the near and far targets, respectively) and ball speed (F 2,12 = 10.59, P < 0.01, IE 21.5%; m eans 87.5, 78.7 and 65.0% for the slow, m edium and fast speeds, respectively). As shown in F ig. 2b, the results obtained on the ® nal test corroborate the results discussed above, with performance being affected by both target location and

ball approach speed, even after 180 trials of practice on this relatively simple task. M ovement velocity at the interception point Figure 3a shows the m ean m ovem ent velocities at the interception point for the pre- and post-practice tests. Analysis of variance revealed signi® cant main effects for target position (F 1,6 = 63.93, P < 0.001, IE 45.1% ; m eans 1049 and 1604 m m s -1 for the near and far targets, respectively), ball speed (F 2,12 = 5.17, P < 0.05, IE 0.6%; m eans 1282, 1336 and 1362 m m s -1 for the slow, m edium and fast velocities, respectively) and test (F 1,6 = 13.04, P < 0.05, IE 1.5% ; m eans 1276 and 1378 m m s -1 for the pre- and posttest, respectively). M oreover, signi® cant ® rst-order interactions were found between group and ball speed (F 2,12 = 4.37, P < 0.05, IE 0.5%) and between target position and test (F 1,6 = 6.34, P < 0.05, IE 1.3% ). Exam ination of the interaction between group and ball speed revealed that the subjects who practised ® rst on the near target were in¯ uenced by ball speed m ore than the subjects who practised ® rst on the far target. Analysis of the interaction between target position and test showed that practice affected m ovem ent velocity at the interception point m ore for the far target location than for the near target location. As can be seen in Fig. 3a, in line with the constraints im posed by the task, movem ent velocity increased with

Task constraints in interception

585

F igure 3 Means of m ovement velocity at the interception point as a function of target location and ball speed conditions for (a) the pre- and post-practice sessions and (b) the ® nal test session.

increasing distance to the target. However, m ovem ent velocity was also affected by the speed of the approaching ball, although this effect accounted for less than 1% of the total variance. Finally, m ovement velocity at the interception point was found to change with practice, with subjects increasing their m ovem ent velocity between the pre- and post-tests for the far target location. In the ® nal test session, signi® cant m ain effects were found for target position (F 1,6 = 61.94, P < 0.001, IE 48.0% ; m eans 1644 and 1117 mm s -1 for the far and near targets, respectively) and ball speed (F 2,12 = 5.64, P < 0.05, IE 1.3% ; m eans 1324, 1397 and 1420 m m s -1 for the slow, m edium and fast speeds, respectively). As is clear from Fig. 3b, these results demonstrate that, even after a few hundred practice trials, the ball approach speed continues to in¯ uence m ovem ent velocity, although the im portance of this factor (1.3%) is m uch less than that of the target location factor (48.0%), which represented the prim ary task constraint.

D iscussion To clarify the role played by task constraints in the organization of interception m ovements, so as to enrich the existing database for the further developm ent of m odels of trajector y form ation in interception tasks, we

evaluated the effects of absolute m ovement velocity requirem ents at the m om ent of contact in the context of varying ball approach speeds. T he latter m anipulation has been show n to give rise to so-called `intensity coupling phenom ena’ when the task did not require (near) m axim al m ovem ent velocities at the interception point (De Vries, 1992; Li and Laurent, 1994, 1995; Fayt, 1995; Li, 1995a,b) and would thus seem to be at odds with a precise control of absolute m ovem ent velocity. In asking subjects to hit orthogonally approaching balls into target boxes placed at various distances from the interception point, success was dependent solely on the adjustm ent of m ovem ent velocity at the tim e of contact w ith the ball to the distance to be covered by the ball after contact. O verall, the results obtained demonstrate that, from the ver y start, subjects were quite pro® cient on this task, even though their perform ance still im proved as a function of practice. As expected, perform ance was better for nearer targets (allowing a larger m argin of error on m ovem ent velocity) and for slower approach speeds (allowing a larger tem poral w indow in w hich contact can be m ade). Movem ent velocity was clearly adapted to the distance to be covered by the ball after contact, with the near target giving rise to lower movem ent velocities and the far target giving rise to higher m ovem ent velocities, even when the conditions were presented in random order. M oreover, because of an initial tendency to

586 undershoot the far target, practice resulted in a slight increase in m ovement velocity in this condition. N evertheless, even though target distance accounted for a substantial proportion of the total variance (i.e. 48% ), a sm all (i.e. < 2%) but signi® cant effect of ball approach speed was found even after practice, reinforcing the idea that intensity coupling is a non-trivial phenom enon in interception tasks. Interestingly, the m agnitude of the intensity coupling effect that was observed after practice appeared to be a function of the task requirements encountered, w ith subjects having practised in the near (i.e. less constraining) target condition ® rst showing a greater dependence on ball approach speed. T he modulated, but nevertheless apparently ubiquitous, effects of ball approach speed on m ovem ent velocity m ay provide an explanation for the dif® culties reported in precisely gauging a drop shot as well as novices’ tendency to respond to a fast ball with a fast m ovem ent.

References Bootsm a, R.J. and van Wieringen, P.C.W. (1988). Visual control of an attacking forehand drive. In Com plex M ovem ent B ehaviour: `The’ M otor- Action Controversy (edited by O.G. M eijer and K. Roth), pp. 189- 199. Amsterdam: NorthHolland. Bootsm a, R.J. and van W ieringen, P.C.W. (1990). Timing an attacking forehand drive in table tennis. Jour nal of E xperim ental Psychology: H um an Perception and Perfor mance, 16 , 21- 29. Bootsm a, R.J., Fayt, V., Zaal, F.T.J. and Laurent, M . (1997). On the inform ation-based regulation of movement: Things Wann (1996) might want to consider. Jour na l of E xperim ental Psychology: H uman Perception and Perform ance , 23 , 1282- 1289. De Vries, M .M. (1992). The timing of an interceptive grasping m ovement: Functional interaction between perception

Fayt et al. and action. Unpublished manuscript, Simon Fraser University, Vancouver, Canada. Fayt, V. (1995). Visual tim ing of a striking action under varying approach conditions. In Studies in Perception and Action III (edited by B.G. Bardy, R.J. Bootsm a and Y. Guiard), pp. 183- 186. Hillsdale, NJ: Lawrence Erlbaum Associates. Franks, I.M ., Weicker, D. and Robertson, D.G.E. (1985). The kinematics, movement phasing and timing of a skilled action in response to varying conditions of uncertainty. Hum an M ovement Science, 4 , 91- 105. Hubbard, A.W. and Seng, C.N. (1954). Visual movements of batters. Research Q uarterly, 25 , 42- 57. Keele, S.W. (1968). Movem ent control on skilled motor perform ance. Psychological B ulletin, 70 , 387- 403. Li, F.-X. (1995a). Anticipation-coincidence at low level of constraint: Sources of inform ation and intensity coupling. Unpublished doctoral dissertation, University of the Mediterranean, Marseilles. Li, F.-X. (1995b). Intensity coupling in peripheral vision. In Studies in Perception and Action III (edited by B.G. Bardy, R.J. Bootsma and Y. Guiard), pp. 195- 198. Hillsdale, NJ: Lawrence Erlbaum Associates. Li, F.-X. and Laurent, M. (1994). Effect of practice on intensity coupling and economy of avoidance skill. Jour nal of Hum an M ovem ent Studies, 27 , 189- 200. Li, F.-X. and Laurent, M . (1995). Intensity coupling in interceptive tasks. In Studies in Perception and A ction III (edited by B.G. Bardy, R.J. Bootsma and Y. Guiard), pp. 191- 194. Hillsdale, NJ: Lawrence Erlbaum Associates. Peper, C.E., Bootsma, R.J., Mestre, D.R. and Bakker, F.C. (1994). Catching balls: How to get the hand to the right place at the right time. Jour nal of E xperim ental Psychology: Hum an Perception and Perfor m ance, 20 , 591- 612. Sch Èoner, G. (1994). Dynamic theory of action- perception patterns: The time-before-contact paradigm . H um an M ovem ent Science, 13 , 415- 439. Tyldesley, D.A. and Whiting, H.T.A. (1975). Operational timing. Jour nal of H um an M ovem ent Studies, 1 , 172- 177.