A Response to the Commentary by Bootsma et al. - PsycNET

Journal of Experimental Psychology: Human Perception and Performance 2010, Vol. 36, No. 4, 1064 –1066

© 2010 American Psychological Association 0096-1523/10/$12.00 DOI: 10.1037/a0020059

Similar Findings, Different Interpretations: A Response to the Commentary by Bootsma et al. (2010) A. J. (Knoek) van Soest and Peter J. Beek

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

VU University Amsterdam

In this paper, we respond to the commentary by R. J. Bootsma, L. Fernandez, A. H. P. Morice, and G. Montagne (2010) concerning our original study on the role of vision during the execution of fast interceptive actions (A. J. van Soest, L. J. R. Casius, W. de Kok, M. Krijger, M. Meeder, and P. J. Beek, 2010), that was inspired by the seminal study of R. J. Bootsma and P. C. W. van Wieringen (1990). Most importantly, we reiterate that the control strategy used in the simulation study (preprogrammed muscle stimulation, triggered at an appropriate time) was adopted on the sole ground that it was the simplest control strategy that allowed us to investigate the role of the visco-elastic properties of muscles. Regarding the visuomotor delay of our participants, we note that the assumption that the visuomotor delay can be reliably identified as the time from the occurrence of a minimum in the coefficient of variation of the relative rate of dilation to the instant of ball contact, is not generally accepted; lacking firm data on the visuomotor delay of our participants, any arguments on the relation between movement time and visuomotor delay are not well grounded. All in all, we believe that our original study added several new– but by no means final–insights to the understanding of the control of fast interceptive actions. Keywords: motor control, visuomotor coupling, perception-action coupling, interceptive actions, hitting

becoming available after drive initiation. When comparing our data with those reported in Bootsma and van Wieringen (1990), we found no qualitative differences except one: A minimum in the coefficient of variation of the relative rate of dilation during movement execution was observed in none of our participants, whereas such a minimum was observed in two of the five participants in the Bootsma and van Wieringen study. These results indicate that, among the combined pool of expert participants, this phenomenon is rather uncommon. In the simulation study, we opted for the simplest possible control strategy that allowed us to zoom in on the role of the viscoelastic properties of muscles in this type of task: In the simulation, we assumed that the neural input to muscles is fully preprogrammed, and further assumed that this muscle stimulation “program” is triggered at the appropriate instant in time. The simulation results indicated that some of the results of Bootsma and van Wieringen (1990) may in fact have arisen because of viscoelastic properties of muscles, including (a) kinematic convergence or “funneling,” (b) a timing accuracy at impact that is superior to the timing accuracy at drive initiation, and (c) negative correlations between tau at the instant of drive initiation and bat velocity as well as bat acceleration. Yet, not all of the key experimental findings were reproduced in the simulation study. We did not find a minimum in the coefficient of variation of the relative rate of dilation during movement execution, and the correlations between tau at the instant of drive initiation and both mean bat acceleration and mean bat acceleration during the second half of the drive, with the mean bat acceleration during the first half of the drive partialed out, were markedly lower than observed experimentally. Given the experimental and simulation results combined, we concluded that the control of the forehand drive in table tennis is neither simply open-loop nor based on a continuous, “real-time” visuomotor coupling. Instead, we argued that interceptive actions with a movement time less than 200 ms are most likely under

The control of fast interceptive actions is far from fully understood. In an influential correlative study of the attacking tabletennis forehand drive, Bootsma and van Wieringen (1990) concluded that continuous coupling between visual information and end-effector movement occurred until shortly before ball contact in two of their five expert performers. Inspired by this original research, we sought to deepen our understanding of the control of fast interceptive actions by conducting a similar experiment aimed at investigating (a) whether withdrawal of visual information during movement execution affects behavior, as would be expected if execution is under continuous visual guidance, and (b) to what extent the findings of Bootsma and van Wieringen may have been due to the viscoelastic properties of the musculoskeletal system that inevitably filter the behavior in any perceptual–motor task. In our experiment, we subjected eight female participants, all table-tennis players at the highest national level, to a manipulation that consisted of unexpected withdrawal of visual information shortly after drive initiation. The main results were that (a) none of the dependent variables differed significantly between the spectacles-open and spectacles-closed conditions, and (b) none of the dependent variables differed significantly between the condition where the participants were wearing the spectacles and the natural condition. These findings led us to conclude that there were no indications that movement execution of our expert performers was affected by visual information

A. J. (Knoek) van Soest and Peter J. Beek, Research Institute MOVE and Faculty of Human Movement Sciences, VU University Amsterdam, Amsterdam, the Netherlands. Correspondence concerning this article should be addressed to A. J. (Knoek) van Soest, Faculty of Human Movement Sciences, VU University Amsterdam, van der Boechorststraat 9, 1081 BT Amsterdam, the Netherlands. E-mail: [email protected] 1064


OBSERVATIONS

model-based predictive control of the neural input to muscles—a control scheme in which visual information (possibly combined with proprioceptive information) is used to predict the near future, in which neural input to muscles is determined on the basis of this prediction, and in which the actual motion follows from the confluence of these control signals and the viscoelastic properties of muscles. We proposed this form of control to account for the fact that no qualitative differences were found in the dependent variables of interest between the spectacles-open and the spectaclesclosed conditions, whereas significant correlations between tau at the instant of drive initiation and mean bat acceleration during the second half of the drive, with the mean bat acceleration during the first half of the drive partialed out, were found in the spectaclesopen and the spectacles-closed conditions. In their commentary, Bootsma, Fernandez, Morice, and Montagne (2010) repeatedly suggest that our study was aimed at finding support for the control scheme used in the simulation study, and they further suggest that model-based predictive control was introduced “out of the blue” to salvage this control scheme when it turned out to be unsuccessful in predicting the data, even for the small number of strokes that served as the basis for the simulation study. As is evident from our article, this suggestion is incorrect: We explicitly stated that our experiment had “. . . the aim of determining whether the hitting action and the evidence for compensatory variability and continuous visuomotor control as reported by Bootsma and van Wieringen would be influenced by the removal of vision during its execution” (van Soest et al., 2010, p. 1041). Furthermore, to avoid any misunderstanding of the goal of our simulation study, we purposely stated “that we are not suggesting that fast interceptive actions are in fact controlled according to the deliberately simple control scheme investigated in our simulation study. . .” (p. 1052). The reason for embracing the control scheme used in the simulation study was that it was the simplest scheme that allowed us to investigate the stabilizing role of the viscoelastic properties of muscle, which was the sole purpose of the simulation study. It is unfortunate that, in their commentary, Bootsma et al. appear to have misinterpreted the status of the control scheme employed in the simulation study because this misinterpretation led them to criticize several aspects of our study on inappropriate grounds. For instance, they criticize the fact that no attempt was made to reproduce all drives in all kinematic detail as an important shortcoming of the simulation study. Admittedly, this should have been done if the model had in fact been intended to provide a full account of the data and thus had to be tested in all regards. However, as it was solely intended to demonstrate the stabilizing role of the viscoelastic properties of muscle, it was sufficient to simulate the drives of a single participant. Likewise, the Bootsma et al. criticism that vision should have been blocked at least one visuomotor delay before movement initiation to test the control scheme used in the simulation study is based on their misinterpretation of the purpose of this study. In a similar vein, Bootsma et al. (2010) argue that optimization of the time at which the muscle stimulation pattern is triggered is a “ghost in the machine” (p. 1061) and that, instead, all simulated movements should have been triggered at a single value of tau. In our view, however, the conclusions regarding the stabilizing effect of the viscoelastic properties of muscle drawn from the simulation study are unlikely to depend critically on the information that is used to trigger the muscle stimulation pattern. We therefore simply

1065

assumed that the muscle stimulation “program” used in the simulation study was triggered at an optimally appropriate instant in time. (As an aside, triggering all simulated movements at a single value of tau would have annihilated all correlations between tau and the end-effector kinematics.) Bootsma et al. (2010) also criticize the experimental part of our study by arguing that blocking vision altogether shortly after movement initiation as was done in our experimental study may have led participants to adjust their strategy. According to Bootsma et al., such a change in strategy may well result in behavior that is virtually indistinguishable from the behavior under normal conditions; consequently, Bootsma et al. contend, our experiment is mute on the role of vision during execution of the motion. In response, we would like to emphasize the fact that we analyzed all main dependent variables of Bootsma and van Wieringen (1990), including the two variables that they considered to be telltales of continuous visuomotor coupling, yet we found no differences in any of these variables. If we would accept the line of argument of Bootsma et al., then this would imply that not only our results in the spectacles-open condition but also the original results of Bootsma and van Wieringen should not be interpreted as evidence for continuous guidance as both sets of findings would be then equally amenable to multiple interpretations. Furthermore, we compared behavior under natural circumstances (conditions NSB and NSA) to behavior when wearing the open LCD spectacles (condition SO) and again found no differences that might suggest a change in strategy. The claim that, despite these findings, a change in strategy that has no behavioral correlates may have occurred in the spectacles-closed condition is in our view impossible to falsify and therefore unproductive for the present discussion. Another important argument in the commentary Bootsma et al. (2010) is that the participants in our experimental study were not as proficient as those in the Bootsma and van Wieringen (1990) study. In response, we reiterate that our participants all played at the highest national level, on which ground we consider it entirely reasonable to label them as “experts.” At the same time, we agree that Bootsma et al. convincingly demonstrated that our participants were not as consistent as those of Bootsma and van Wieringen; apparently, the skill level of the participants in the Bootsma and van Wieringen study stands out even among a group of expert performers. Thus, although we cannot rule out the possibility that continuous visual guidance occurs in players who excel even among a group of expert players, the combined results of the Bootsma and van Wieringen study and our study do indicate that continuous visual guidance is not a widespread behavioral characteristic among expert table-tennis players (any findings in other sports like volleyball notwithstanding). Along similar lines, Bootsma et al. (2010) suggest that it is not surprising that the behavior of our participants was not affected by blocking vision shortly after drive initiation because their visuomotor delay was longer than their movement time. The suggestion that movement time was shorter than the visuomotor delay is not based on actual measures of the visuomotor delay but rather on the absence of a minimum in the coefficient of variation of relative rate of dilation in any of our participants. Thus, this suggestion hinges on the assumption that visuomotor delay can be reliably identified as the time from the occurrence of such a minimum in the coefficient of variation of relative rate of dilation to the instant of ball contact. However, this assumption is not generally accepted (e.g., Wann, 1996). Although Bootsma et al. acknowledge this in their commentary (“. . . one may


1066

OBSERVATIONS

question whether the moment of occurrence of the minimum in variability in the evolving first-order temporal relation between player and ball can be taken to sign the ultimate moment of control being possible” [p. 1059]), they nonetheless consider the aforementioned assumption to be plausible. Given this lack of consensus, it is unfortunate that objective data on the visuomotor delay are neither available for our participants nor for the participants in the Bootsma and van Wieringen (1990) study. Yet, considering that the visuomotor delay is generally assumed to lie between 100 ms and 200 ms, and that movement time ranged from 92 ms to 195 ms in the combined pool of expert participants in our study and that of Bootsma and van Wieringen, it seems unlikely that only two of 12 expert participants in total had a visuomotor delay that was shorter than their movement time. We further wonder how, in the account of Bootsma et al., time to contact at movement initiation can be correlated to (prehit) bat kinematics if movement initiation falls within a single visuomotor delay from ball– bat contact; after all, information obtained within this period per definition cannot affect the movement. Given these uncertainties, it would be very interesting to independently (but in a task-specific manner!) estimate the individual visuomotor delay of the participants of any follow-up studies. All in all, we are well aware of the limitations of our study, but we do not share the opinion of Bootsma et al. (2010) that these limitations “severely limit the impact of their findings” (p. 1056). In our view, such limitations are difficult to avoid in any study motivated by the interest in the control of complex real-life perceptuomotor tasks that we share with Bootsma et al. Yet, although we fully agree that “appropriate experimentation and principled theorization” are required to further improve our understanding of the control of fast

interceptive actions, we do not agree with the suggestion that our study would principally fail to meet these requirements. We find it unfortunate that this suggestion was generated through a misportrayal of the objectives of our study and arguments referring to unmeasured and unknown quantities. With a manipulation of vision and a simulation of musculoskeletal dynamics, we believe our study added several new— but by no means final—insights to the understanding of the control of fast interceptive actions, thus taking the seminal study of Bootsma and van Wieringen (1990) a few steps further.

References Bootsma, R. J., Fernandez, L., Morice, A. H. P., & Montagne, G. (2010). Top-level players’ visual control of interceptive actions: Bootsma and van Wieringen (1990) 20 years later. Journal of Experimental Psychology: Human Perception and Performance, 36, 1056 –1063. Bootsma, R. J., & van Wieringen, P. C. W. (1990). Timing an attacking forehand drive in table tennis. Journal of Experimental Psychology: Human Perception and Performance, 16, 21–29. Van Soest, A. J., Casius, L. J. R., De Kok, W., Krijger, M., Meeder, M., & Beek, P. J. (2010). Are fast interceptive actions continuously guided by vision? Revisiting Bootsma and van Wieringen (1990). Journal of Experimental Psychology: Human Perception and Performance, 36, 1040 – 1055. Wann, J. P. (1996). Anticipating arrival: Is the tau margin a specious theory? Journal of Experimental Psychology: Human Perception and Performance, 22, 1031–1048.

Received November 5, 2008 Accepted April 27, 2010 !