Edited by: Michael J. Richardson, University of Cincinnati, USA
Reviewed by: Lynden K. Miles, University of Aberdeen, UK; Kerry Marsh, University of Connecticut, USA
*Correspondence: Sabrina Golonka
This article was submitted to Cognitive Science, a section of the journal Frontiers in Psychology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Although it is well established that rhythmically coordinating with a social partner can increase cooperation, it is as yet unclear when and why intentional coordination has such effects. We distinguish three dimensions along which explanations might vary. First, pro-social effects might require in-phase synchrony or simply coordination. Second, the effects of rhythmic movements on cooperation might be direct or mediated by an intervening variable. Third, the pro-social effects might occur in proportion to the quality of the coordination, or occur once some threshold amount of coordination has occurred. We report an experiment and two follow-ups which sought to identify which classes of models are required to account for the positive effects of coordinated rhythmic movement on cooperation. Across the studies, we found evidence (1) that coordination, and not just synchrony, can have pro-social consequences (so long as the social nature of the task is perceived), (2) that the effects of intentional coordination are direct, not mediated, and (3) that the degree of the coordination did not predict the degree of cooperation. The fact of inter-personal coordination (moving together in time and in a social context) is all that's required for pro-social effects. We suggest that future research should use the kind of carefully controllable experimental task used here to continue to develop explanations for when and why coordination affects pro-social behaviors.
It is well-established that moving in time with other people can increase cooperation between co-actors (Anshel and Kipper,
In this paper we consider some classes of model that could characterize how coordination impacts cooperation. These models vary along three dimensions: (1) whether increased cooperation depends on in-phase
Movements are
If coordination, generally, and not just in-phase synchrony, has positive consequences on cooperation, then the effects should be obtained following coordination at any relative phase. We currently lack evidence to support this idea because the majority of tasks used to test the pro-social effects of coordination rely exclusively on in-phase coordination (and, to our knowledge, our experiment is the first work to address the effects of anti-phase coordination on cooperation, specifically). Those that have employed anti-phase conditions have found mixed evidence concerning whether anything besides in-phase synchrony impacts social variables (e.g., Miles et al.,
The effect of coordination on pro-social variables is indirect if coordination must impact an intervening variable (e.g., group cohesion) or coincide with a causally relevant variable (e.g., social context) in order to affect cooperation. If this is the case, then coordination only has positive consequences for pro-social variables by virtue of its effect on something like group cohesion or by providing the opportunity to engage in a certain type of social context. In contrast, the effect of coordination on pro-social variables could be direct. If the relationship is direct then coordination would not need to impact an intervening variable or coincide with another causally relevant variable to influence cooperation.
The literature, to date, is conflicted concerning directness. We first consider evidence for a mediating variable between coordination and cooperation. Research has focused exclusively on two potential mediators—group cohesion and self-other-overlap. Group cohesion is the feeling of being on the same team and being emotionally connected with other group members. Wiltermuth and Heath (
Others have investigated self-other-overlap as a potential mediator of the relationship between coordination and cooperation; again, evidence for the mediated model is inconclusive. Lumsden et al. (
Another way the effect of coordination on pro-social variables could be thought of as in/direct depends on whether a coordination task, in and of itself, (i.e., absent a particular social context), is sufficient to impact coordination. If it is direct in this way then coordinating movements with, say, a metronome or a computer display rather than a co-actor, would be sufficient to lead to social consequences. If it is indirect in this way, then coordination must be accompanied by some kind of social context to impact pro-sociality; i.e., effect would not be due to coordination “
In sum, the evidence from previous research is inconclusive about whether coordination must impact an intervening variable in order to have positive consequences on cooperation. Evidence is stronger for the idea that coordination must coincide with a social context in order to affect cooperation. The studies reported below provide the strongest evidence to date for D+ vs. D− models by testing a variety of potential mediators (i.e., group cohesion, self-other overlap, trust, self-rated success at coordination, self-rated task difficulty, task difficulty, and mood) within subjects at both pre- and post-coordination. In line with the substantial existing evidence that social context is important, all of the studies below involve pairs of participants completing an intentional coordination task together; however Followup 1 manipulates whether the information participants use to coordinate is social or non-social.
Whether the effect of intentional coordination on cooperation is direct or indirect, there are two main types of relationship we might observe between these variables. The first possibility is that individual measures of coordination success predict individual levels of cooperation. That is, changes in cooperation occur in proportion to changes in coordination success. The second possibility is that there is a threshold relationship between coordination and cooperation. In this case, coordination would positively influence cooperation as long as some minimum threshold of coordination success was achieved.
Previous research paints a mixed picture in terms of what to expect on this dimension. The only work focusing on cooperation to take actual measures of coordination found that coordination did not predict cooperation (Kirschner and Ilari,
Researchers have used a variety of tasks to investigate the effect of intentional coordination on pro-sociality (e.g., waving cups and singing: Wiltermuth and Heath,
CRM tasks are essentially perception-action tasks, and have typically been studied as such in the experimental literature (e.g., Kelso,
Bingham et al. (Bingham,
The studies below are based on the task used by Bingham and colleagues to develop an explanatory model of CRM. This CRM task is particularly well-suited to the job of discriminating S+ and S− models, as it's possible to run the task with any target relative phase, and of discriminating P+ and P− models, as it allows us to compute precise and sensitive measures of coordination that can be used to determine how much actual coordination predicts post-task measures, if at all. We can then combine data from this task and other measures to discriminate D+ and D− models as well.
This choice of task is also ideal for constructing an appropriate control task, which has proven a major challenge in the literature. A good control task must be comparable to the CRM task, involving co-actors making comparable movements (though ones that are not rhythmically coordinated with their co-actors). However, control tasks in the six papers looking at how CRM affects cooperation varied considerably in how closely they match the experimental task (see Table
Anshel and Kipper, |
Group singing | Listing to music/watching a documentary |
Wiltermuth and Heath, |
Synchronized walking | Walking normally |
Wiltermuth and Heath, |
Synchronous cup waving and singing in time to Canadian anthem | Static cup holding and silently reading lyrics while listening to Canadian national anthem |
Kirschner and Tomasello, |
A game involving synchronously singing and walking in time to music | A game involving walking and vocalizing non-synchronously with no music |
Reddish et al., |
Synchronous movements in time to a metronome | Watching a video of other people performing the task |
Reddish et al., |
Synchronized foot tapping | Asynchronous foot tapping |
Kirschner and Ilari, |
Synchronized drumming | Solitary drumming |
Reddish et al., |
Synchronized foot tapping | Completing a jigsaw puzzle |
Our CRM task is amenable to a straight forward, well-matched control task whereby participants are instructed to move their joysticks at different frequencies while performing different movements. This control condition is minimally different from coordinated conditions (both involve rhythmically moving a joystick at a specified frequency), while breaking the coordination between partners.
The goal of the studies that follow is to begin homing in on the class of model that best captures the relationship between intentional coordination and cooperation. This work will place specific, empirically-driven constraints on future work concerning the mechanism by which coordination influences cooperation. Experiment 1 was designed to discriminate between S+ and S− models (in-phase synchrony or coordination), between D+ and D− models (direct or mediated), and between P+ and P− models (group or individual level effect). Based on the results of this experiment we conducted two follow-ups. The first further explores the S+/S− distinction by investigating the consequences of coordinating via social and non-social information. The second probes the necessary features of a coordination task by testing two control tasks.
Experiment 1 tested whether in-phase synchrony is necessary to the effect of coordination on cooperation or whether the effect obtains with other coordinations as well (S+ or S−). Since our task allows a kinematic record of each participant's movements, we also tested whether cooperation varies in proportion to coordination, allowing us to discriminate between P+ and P− models. Finally, we measured several potential mediators suggested from previous research, which provides some evidence for D+ vs. D− models.
Sixty-six undergraduate students at Leeds Beckett University volunteered to participate (19 males and 47 females
The study employed an experimental design with one between-subjects factor: Movement Phase. This had three levels: in-phase (0°), anti-phase (180°), or no coordination (control).
In both experimental conditions, pairs of participants, sitting side by side moved one joystick each (Logitech Pro joysticks with force feedback disabled) horizontally at 0.75 Hz using a point light display (PLD) to monitor their and their partner's movements. The PLD consisted of two white feedback dots displayed on a black background by a single laptop screen positioned approximately 1 m in front of them. The dots were 40 × 40 pixels, and separated by a visual angle of 0.14°, one above the other, positioned in the center of the screen (Wilson et al.,
For the control task, participants made uncoordinated movements at different frequencies. One participant always moved their joystick at 0.6 Hz and the other always moved at 0.9 Hz (0.75 ± 0.15 Hz). Participants alternated moving their joysticks vertically and in clockwise circles, so that partners never performed the same movement during a trial. Participants switched movements every trial (e.g., person 1 moved vertically on one trial, in circles on the next etc.; person 2 in the pair did the opposite).
Participants in all conditions first saw two 15 s demonstrations of dots moving at the desired phase and frequency. In the experimental conditions both dots moved at 0.75 Hz (at either 0 or 180° relative to each other). In the control condition one dot moved at 0.6 Hz and the other at 0.9 Hz. After each demo participants had 30 s practice time to acquaint themselves with the required movements. Following this brief initial practice, participants completed six 60 s trials. Each trial was preceded by a four second version of the demonstration pacing them to the required phase and frequency of movements. This experiment was run on a MacBook Pro with a custom Matlab toolbox programmed by the second author and incorporating the Psychtoolbox (Brainard,
Self/other overlap was measured using the Inclusion of the Other in Self (IOS) scale (Aron et al.,
Five questions were used to measure mood, trust and cohesion (see Data Sheet
This included both a Public Goods Game (PGG) and an investment game (see Data Sheet
After Round 5 of the PGG, participants played an investment game (adapted from Berg et al.,
This study was conducted in pairs. Sessions lasted approximately 25 min. Participants completed the IOS and the cohesion scale (pre-test measures of potential mediators, and mood item) followed by the movement task. Participants then rated their perceived success at the coordination task as well as task difficulty and enjoyment using four-point Likert scales. Next, participants completed a second copy of the IOS and cohesion scale (post-test measures of potential mediators, and mood item). Finally, participants took part in the Economic (public goods and investment) Game.
We checked whether mood, task difficulty, task enjoyment, and perceived success differed between in-phase, anti-phase, and control tasks. The distribution of scores on each of these variables was found non-normal from Shapiro-Wilkes tests (SW tests of normality used throughout) (
All movement trials except for the first two practice rounds were analyzed. A low-pass Butterworth filter with a cut-off frequency of 10 Hz filtered each dot's position time series. A 60 Hz time series of the relative phase between the two dots was computed as the difference between the arctangent of each dot's velocity over position at each sample.
Mean vector length (MVL) is the circular equivalent of the standard deviation (Batschelet,
The distribution of MVL scores of those who moved in-, anti-phase and those who did not coordinate all differed significantly from normality (
Next we examined whether participants in the in- and anti-phase conditions were more cooperative post movement task than those in the control condition. A univariate ANOVA found a significant effect of phase on the mean public account donation [
Next we conducted a simple linear regression with each pair's MVL scores and each pair's average public goods donation to determine if the degree of coordination success predicts the degree of cooperation. A pair's coordination score did not significantly predict their average cooperation score [
Trust was measured using the first part of the investment game (choosing what to invest with the other player: investing nothing, a quarter, half, or all). The distributions of those who moved in-phase, anti-phase, and those who did not coordinate all deviated significantly from normality (
As a further check that coordination had no effect on trust, we compared self-reported measures of trust across the coordination conditions. Change scores for the self-reported trust measure were first calculated by subtracting each person's “before” score from their “after” score. The distributions for those who moved in-phase, anti-phase, and those who did not coordinate all deviated significantly from normality (
Reciprocity was measured using the option chosen in the second part of the investment game (choosing to return nothing, return only the original investment, return the original investment plus half of the bonus, or, return the original investment plus all of the bonus). Reciprocity scores for those who moved in-, anti-phase and those who did not coordinate all deviated significantly from normality (
Change in group cohesion was measured as the sum of the difference between the three cohesion change questions (how similar/close/connected they felt to each other). A univariate ANOVA with phase (in-, anti-phase, no coordination) showed no significant effect of phase on group cohesion [
Change in self-other overlap was measured as the difference in self-other overlap before and after engaging in the coordination task (post-coordination—pre-manipulation). The distribution of overlap change scores for those who moved at in-, anti-phase, and those who did not coordinate all deviated significantly from normality (
Analysis previously reported also confirmed that self-report measures of trust, mood, task difficulty, task enjoyment and perceived success did not differ between movement conditions.
The results showed that participants who moved in-phase with one another were more cooperative than those who moved in an uncoordinated manner. None of the measured candidate mediators were related to cooperation, and cooperation was not predicted by the level of coordination between partners. The results of Experiment 1 lend support to S+, D+, and P− models of how intentional coordination affects cooperation.
MVL scores suggested participants coordinated equally well at both in- and anti-phase. Coordination in both of these experimental conditions was better than in the control condition. MVL scores did not significantly predict cooperation, which suggests that the social effects seen post-entrainment do not vary linearly at an individual level with coordination. This is consistent with Kirschner and Ilari (
MVL is a measure of coordination (i.e., the extent to which people are doing
Further analyses of the proportion-time-on-target scores revealed that those who were instructed to move in-phase were more successful than those that were instructed to move anti-phase (See Figure
Against predictions, changes in trust, group cohesion and self/other overlap did not differ between conditions, suggesting that these factors do not mediate CRM's effect on cooperation (supporting D+ models). The finding that increases in group cohesion do not mediate these effects supports the work of Dong et al. (
One reason for the inconsistencies in findings could be that the present study is the first to take “before and after” coordination measures of possible mediators. It may be the case that CRM does not actually foster changes in the given variables and that previous studies simply found group differences across these variables as opposed to actual increases in mediators as a result of CRM. Alternatively it could be that the measures used here are not sensitive enough to be used as a before and after measure. Completion of the pre-test measures may have restricted participant's answers to post-test measures, therefore leaving participants unable or unwilling to give more natural responses which may have otherwise led to us finding increases in potential mediators. For the cohesion measure we saw a mean change score of 2.27 with a standard deviation of 5.63. For the overlap measure we saw a mean change score of 0.45 with a standard deviation of 1.3. Considering we find considerable variation in individual change scores, we do not believe this interpretation alone can explain our findings.
This experiment did not provide conclusive evidence that cooperation was improved by coordination more generally. Significantly greater cooperation was only seen after in-phase coordination compared to control. Anti-phase coordination did not promote greater cooperation than after control, however cooperation levels following anti-phase coordination did not significantly differ from cooperation levels following in-phase coordination either. While this might initially lend some support to the S+ class of models (synchrony, rather than coordination being required). Findings lead us to further question whether in-phase synchrony is crucial? Anti-phase coordination is a stable form of coordination (Kelso,
The findings of Kokal et al. (
One potential limitation of our task was the use of simple PLDs to transmit movement information. These displays are informative about the dynamics of a person's action (Johansson,
The fact that relevant social information may be harder to detect during anti-phase coordination might explain why anti-phase coordination did not significantly differ from control. A follow up explores this possibility by having participants coordinate at both relative phases using direct visual information of each other's movements. This set up makes the social nature of the task more salient. If post task cooperation is higher following anti-phase coordination given this change, it would add further support for D− models, where an additional causally relevant factor (e.g., social context) is necessary for coordination to affect cooperation.
In this follow up, we used a modified version of the CRM task in which co-actors coordinated by looking at each other in a full-length mirror instead of using PLDs. Only the two experimental conditions (in- and anti-phase) were run in order to test whether increased social information would allow cooperation following the anti-phase condition to reach the level seen after in-phase coordination in Experiment 1. It was hypothesized that coordinating via a mirror would allow anti-phase CRM to affect cooperation similarly to in-phase CRM.
Forty-four psychology students at Leeds Beckett University volunteered to participate (8 males and 36 females,
The design was identical to the in-phase and anti-phase conditions from Experiment 1 except that participants watched each other using a 6 ft mirror placed horizontally 1m in front of them, below the laptop screen so that they could each view both of their upper bodies. These data were compared to the corresponding conditions from Experiment 1 to see whether enriched visual social information influenced cooperation. This follow up employed an experimental design with one between-subjects factor: Movement Phase, with two levels in- and anti-phase. This enabled us to analyse the coordination data using the superior proportion time-on-target measure. The remaining measures and procedure were identical to Experiment 1.
We first examined mood, task difficulty, task enjoyment and perceived success measures for these two new conditions to see whether these varied across conditions, using a series of Kruskal-Wallis tests (all data distributions non-normal,
We investigated differences in coordination scores across conditions using proportion-time-on target as a measure of coordination. The distributions of those who coordinated using the PLD and mirror at both in- and anti-phase (
We then explored how rhythmically coordinating at different relative phases via differing Coordination Information affected cooperation using a 2 way ANOVA. There was no main effect of either Coordination Information or Movement Phase (
Next we conducted a simple linear regression with each pair's proportion-time-on target scores and each pair's average public goods donation, to determine if coordination success predicts cooperation. A pair's coordination score did not significantly predict a pair's average cooperation score [
Separate 2 Way ANOVA's were conducted for each of the potential mediators as reported in Experiment 1, no significant main effects of either Movement Phase or Coordination Information and no significant interactions were found in any of these analyses (all
Participants coordinating at anti-phase were more cooperative if they coordinated via direct visual information of their partner's movements rather than via PLDs. In fact, those coordinating at anti-phase using the mirror saw cooperation levels comparable to participants in the in-phase condition. There was no such increase in effect for those coordinating in-phase using direct visual info. This supports the claim of Kokal et al. (
Coordination scores (proportion-time-spent-on-target) again did not significantly predict cooperation scores (supporting a P− model). There is still no evidence that coordination success is driving CRM's effect on cooperation, replicating the result from Experiment 1 and supporting work by Kirschner and Ilari (
Greater cooperation can therefore follow either in- and anti-phase CRM compared with uncoordinated movements. However, analyses of coordination scores have shown that actual coordination does not seem to be driving this effect. The degree of coordination does not successfully predict the degree of cooperation. So what is it about the CRM task that is driving differences in cooperation? What are the critical differences between the coordinated and uncoordinated versions of this task?
In the CRM task people make the same (horizontal) movements at a shared frequency (0.75 Hz), while in the control task people make different movements (circular and vertical) at different frequencies (0.6 or 0.9 Hz). This means there are two potential differences between the CRM task and the control, type of movement and frequency of movement. Having participants perform different movements is essential to break coordination in the control task, since research shows people will end up falling into one of the two stable phases of coordination when performing the same kinds of movement unless they are trained to achieve out-of-phase coordination (Kelso,
When engaging in CRM in everyday life (e.g., when dancing), people often coordinate different movements to the same overall rhythm. What is more, Lakens (
Twenty-two undergraduate students at Leeds Beckett University volunteered to participate (4 males and 18 females,
Participants made different movements but at the same frequency (0.75 Hz). One participant moved the joystick vertically and the other in clockwise circles. Participants switched movements each trial. Otherwise the structure of the movement task was identical to the Control in Experiment 1. This condition (Coordinated) was then compared with the original in-phase (In-phase) and control condition (Control) from Experiment 1. With no defined target relative phase we analyzed coordination using MVL. The remaining measures and procedure were identical to those reported in Experiments 1.
We first examined mood, task difficulty, task enjoyment and perceived success measures to see whether these varied across conditions using a series of Kruskal-Wallis tests (All data's distributions not normal,
We then investigated whether coordination scores differed across conditions using an independent samples Kruskal-Wallis test (recall coordination data previously failed normality tests). There was a significant effect of Movement Type on coordination scores [
Next we examined the cooperation scores of those in the Coordinated compared with the original In-phase and Control conditions from Experiment 1. A univariate ANOVA was performed to see whether cooperation (mean public account donation) differed across the three movement conditions (In-phase, Coordinated and Control). There was a significant effect of Movement Type [
A univariate ANOVA and Kruskal Wallis test (recall previous normality scores) again confirmed that there were no significant differences in any of the candidate mediators between conditions (all
The results of this follow up show that similar levels of cooperation are seen after coordinating different movements to a common frequency as are seen after in-phase coordination, despite levels of actual coordination being significantly lower. MVL scores show that coordinating different movements to a common frequency produced significantly less tight coordination than coordinating at in-phase but significantly tighter coordination than in the original control. This was not the pattern observed in cooperation, however. The Coordinated and In-phase conditions produced comparable levels of cooperation, and both showed higher cooperation than the Control condition.
These results suggest that people do not need to perform the same type of movements for coordination to have cooperative social consequences and emphasize again that tightness of coordination is not directly linked to the magnitude of cooperation (P− model). The important factor appears to be that they coordinate to a common rhythm. Verbal reports from participants in this new condition also indicated that participants felt they were coordinating their actions. Multiple participants in this condition reported that they were trying to coordinate one full cycle of their movements to a full cycle of the other's movements (i.e., trying to complete one full up-down-up cycle on the time it took the other to complete a full circle).
This, along with the other findings reported in this paper, suggests that it is not moving at some particular phase, or a given tightness in coupling which fosters cooperation. Rather, the crucial factor appears to be just intentionally moving in time with somebody in a clearly social context, regardless of whether the same movements are performed or whether there is a specific phase locking.
The experiment and follow ups detailed here showed that those who perform a simple CRM task are more cooperative post-task than those who perform a control task. We also showed that similar effects obtain following anti-phase coordination and after coordinating different movements to the same overall rhythm. We found no evidence that the degree of coordination predicts the degree of cooperation, and no evidence that increases in group cohesion or blurring of self/other overlap were mediating CRM's effects on cooperation. The effects on cooperation seem to mostly stem from simply moving in time in a social context.
The results of Experiment 1 initially supported S+ models, with no significant effect of anti-phase movement on cooperation. However, the point-light displays we used only provided information about the coordinated rhythmic movement, and may detract from the social context. Increasing the salience of the social context by using mirrors led to anti-phase movements affecting cooperation to the same extent as in-phase movements. In addition, different movements at the same frequency led to greater cooperation than different movements at a different frequency. The former are still coordinated in that they are matched in time (and participants reported working to coordinate this timing). Overall, these results suggest it is temporal coordination, and not just synchrony, which can lead to pro-social consequences and so future models should be of the S− class.
Across all three studies, we found no effects of any candidate mediating variable on cooperation. It's worth noting at this point that we only looked at interactions between pairs of coordinating co-actors, and different dynamics may be at play when groups of 3 or more engage in CRM. This may be especially relevant for the group cohesion findings, as group cohesion may not be an appropriate construct for two person groups. Petersen et al. (
Alternatively it may be the case that we failed to see changes in potential mediators due to a testing effect confound. It is possible that including pre as well as post-test measures of mediators may have restricted participants post-test responses. We do not however believe that this is a likely explanation, since in other work (Cross et al., Submitted) increases in group cohesion amongst larger groups have been found using these test-retest measures.
Still, results reported here showed greater cooperation amongst pairs who had performed coordinated movement than those who had performed uncoordinated movement, which was not mediated by any of the variables suggested by the literature.
We did observe an effect of social context, whereby having visual access to one's partner during the coordination task was necessary to obtain an effect of anti-phase coordination on cooperation. This pattern of results supports a D− model and is consistent with previous work showing that coordination does not have positive social consequences if the coordination task does not have a social component.
Again, we found no evidence that the quality of coordination between participants predicted the amount of cooperation they exhibited. In addition, there was no increase in coordination stability in anti-phase movements when co-actors coordinated via direct movement information, but cooperation did increase. Once people perceive that they are temporally coordinating in a social context, greater cooperation follows. This supports P- class models for future work.
The findings presented in this paper apply only to cases of intentional coordination. They may not necessarily generalize to instances of unintentional coordination. This remains an interesting point for future work to explore. A further limitation is that the results of Experiment 1 were analyzed in conjunction with both of the follow ups. These results are effectively exploratory and require independent replication.
The current studies demonstrated that people who engage in a simple CRM task are more cooperative post task than people who engage in a control task. By relying on a well-defined and well-understood CRM task (see Golonka and Wilson,
Participants were provided with an information sheet explaining the nature of the research prior to making an appointment to participate in the research. When potential participants arrived at their appointment, they were provided with a detailed consent form detailing their rights as participants. They were free to ask the researcher any questions. If they were happy to continue, they signed the consent form and the experimental session began. The research was approved by the Leeds Beckett University Psychology Ethics Committee
LC conducted the experiments and analyses reported in this paper as part of his Ph.D. under the supervision of SG (director of studies) and AW (second supervisor). All authors therefore contributed to the design and analysis of the studies and we all contributed equally to the writing.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at:
1Anti-phase is typically less stable than in-phase; this is one of the hallmarks of coordinated rhythmic movement. The lack of a difference here is a common issue with the MVL measure because it does not account for what relative phase people are actually performing. Anti-phase coordination can show an elevated MVL if people end up switching to in-phase coordination, and do that well (Wilson et al.,