Effects of visual and auditory guidance on bimanual coordination complexity

Human Movement Science 54 (2017) 13–23

Contents lists available at ScienceDirect

Human Movement Science journal homepage: www.elsevier.com/locate/humov

Full Length Article

Effects of visual and auditory guidance on bimanual coordination complexity Daniela V. Vaz a,b,⇑, Bruce A. Kay b, Michael T. Turvey b a b

Federal University of Minas Gerais, Av. Pres. Antônio Carlos, 6627 – Pampulha, Belo Horizonte, MG 31270-901, Brazil University of Connecticut, Department of Psychology, 406 Babbidge Road, Unit 1020, Storrs, CT 06269-1020, United States

a r t i c l e

i n f o

Article history: Received 28 November 2016 Revised 16 February 2017 Accepted 24 February 2017

Keywords: Guidance Lissajous Bimanual coordination Complexity

a b s t r a c t Perceptual guidance of movement with simple visual or temporal information can facilitate performance of difficult coordination patterns. Guidance may override coordination constraints that usually limit stability of bimanual coordination to only in-phase and antiphase. Movement dynamics, however, might not have the same characteristics with and without perceptual guidance. Do visual and auditory guidance produce qualitatively different dynamical organization of movement? An anti-phase wrist flexion and extension coordination task was performed under no specific perceptual guidance, under temporal guidance with a metronome, and under visual guidance with a Lissajous plot. For the time series of amplitudes, periods and relative phases, temporal correlations were measured with Detrended Fluctuation Analysis and complexity levels were measured with multiscale entropy. Temporal correlations of amplitudes and relative phases deviated from the typical 1/f variation towards more random variation under visual guidance. The same was observed for the series of periods under temporal guidance. Complexity levels for all time series were lower in visual guidance, but higher for periods under temporal guidance. Perceptual simplification of the task’s goal may produce enhancement of performance, but it is accompanied by changes in the details of movement organization that may be relevant to explain dependence and poor retention after practice under guidance. Ó 2017 Elsevier B.V. All rights reserved.

1. Introduction The existence of spontaneous movement tendencies and movements that are very difficult to perform suggests constrains to coordination. For example, oscillating both hands at the same frequency, in mirrored or opposite directions (in phase and antiphase), comes naturally and is considerably easier than oscillating the hands at different frequencies or other phase relations. Motor constraints, such muscular anatomical symmetry and activation pathways have been considered essential to explain these coordination tendencies (Carson, Riek, Smethurst, Párraga, & Byblow, 2000; Kelso, 1984; Salter, Wishart, Lee, & Simon, 2004; Temprado, Swinnen, Carson, Tourment, & Laurent, 2003). Perceptual factors, however, also play an indispensable role (Mechsner, Kerzel, Knoblich, & Prinz, 2001; Mechsner & Knoblich, 2004; Riek & Woolley, 2005). Current theory acknowledges that a variety of constraints interact to shape coordination (Kelso, Fink, DeLaplain, & Carson, 2001; Shea, Buchanan, & Kennedy, 2016; Swinnen & Wenderoth, 2004). These constraints can be manipulated to influence the stability and accuracy of coordination patterns (Park, Collins, & Turvey, 2001; Riek & Woolley, 2005; Swinnen & Wenderoth, 2004; ⇑ Corresponding author at: Federal University of Minas Gerais, Av. Pres. Antônio Carlos, 6627 – Pampulha, Belo Horizonte, MG 31270-901, Brazil. E-mail addresses: [email protected] (D.V. Vaz), [email protected] (B.A. Kay), [email protected] (M.T. Turvey). http://dx.doi.org/10.1016/j.humov.2017.02.010 0167-9457/Ó 2017 Elsevier B.V. All rights reserved.

14

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

Temprado, Salesse, & Summers, 2007; Temprado et al., 2003) with a dramatic impact on performance (Mechsner et al., 2001). For example, perceptual constraints can be manipulated with feedback. Concurrent feedback of interlimb coordination by means of an angle-angle plot (i.e., a Lissajous figure) translates motion into a simple visual pattern and makes normally unstable bimanual coordination patterns achievable after little training (Kovacs, Buchanan, & Shea, 2010). Use of these plots can improve error detection and correction while also reducing attentional demands necessary to ‘tune-in’ the desired coordination pattern (Kennedy, Wang, Panzer, & Shea, 2016). Several phase relations other than the spontaneously stable inphase and antiphase can be consistently produced with reduced variability and error (Kennedy, Wang, Panzer, & Shea, 2016; Kovacs, Buchanan, & Shea, 2009a,b). The evidence suggests that simplification of perceptual goals can overrule basic coordination tendencies such that days of practice can be sidestepped. If the primary challenge in motor learning is the overcoming of basic coordination tendencies (Temprado & Swinnen, 2005; Zanone & Kelso, 1992), adequate task manipulations could become a formidable tool to aid learning in fields such as rehabilitation, sports and music. However, being able to execute movements under appropriately manipulated perceptual conditions does not imply that learning to execute the task in usual conditions will be facilitated (Kovacs & Shea, 2011). Movement coordination under guidance of simplified feedback and under usual perceptual circumstances involve different neural pathways (Debaere, Wenderoth, Sunaert, Van Hecke, & Swinnen, 2003) and might have different dynamical requirements. These differences may lie behind the poor retention observed after practice with guidance (Kovacs & Shea, 2011; Maslovat, Brunke, Chua, & Franks, 2009; Ronsse et al., 2011; Salmoni, Schmidt, & Walter, 1984). Do visual and auditory guidance produce synergies with qualitatively different dynamical organization? The structure of variability of a movement time series is informative about the dynamical processes of coordination that underlie performance (Riley & Turvey, 2002). Specifically, the nature of the coupling between degrees of freedom of a system can be assessed with measures of complexity. Although complexity has proved to be a slippery concept, recent explanatory proposals recognize that behavioral complexity derives from the activity of numerous substructures and subfunctions distributed over multiple spatiotemporal scales (Ihlen & Vereijken, 2010; Turvey, 2007). The tying together these numerous time scales in nested loops allows for the stable and adaptable control which is characteristic of coordinated movement (Turvey, 2007; West & Griffin, 1998). Multiscale entropy measures can capture such structural richness. Multiscale entropy measures are based on calculation of entropy values for several time scales of a time series. Contrary to usual single scale entropy measures, multiscale measures consistently yield higher complexity values for simulated long-range correlated stochastic series compared to uncorrelated, unstructured random stochastic series (Costa, Goldberger, & Peng, 2002). Multiscale entropy should be able to capture differences in structural richness of motor synergies assembled under distinct perceptual constraints. What changes should be expected under guidance of simplified perceptual patterns? Studies comparing independent and guided movement suggest hypotheses. In self-paced movement the time series of relative phase in 0° and 180° bimanual coordination (Torre, Delignières, & Lemoine, 2007) show an inverse relationship between power and frequency (1/f or pink noise). Pink noise, a kind of correlated, structurally rich noise, is indicative of a long-range memory process: a typical dependence in the series, for example a positive trend between successive values, appears nested with statistically similar trends expressed at larger scales (Diniz et al., 2011). Pink noise has been interpreted as a signature of interdependent interactions among the numerous components of a system, self-organization (the spontaneous organization that coordinates system behavior in the absence of a central controller) and emergence (the appearance of features that are not implicit in the parts of the system) (Turvey, 2007; Van Orden, Holden, & Turvey, 2003). Under metronome pacing, the typical 1/f observed in independent, self-controlled performance disappears. In its place, anti-persistent temporal correlations are observed (Chen, Ding, & Kelso, 1997; Torre & Delignières, 2008). These findings have generated interesting theoretical speculations: 1/f structure is likely to appear when the coordination is weakly constrained by external requirements. In contrast, enhanced sources of external constraint would reduce voluntary control (Kloos & Van Orden, 2010; Van Orden, Kloos, & Wallot, 2011; Van Orden et al., 2003). Given the dramatic effects of Lissajous guidance on performance, it appears to function as more global constraint on performance than metronomes. If reliance on a Lissajous plot can in fact substitute for mechanisms of voluntary control and interfere with the intrinsic pink noise dynamics of purposeful coordination, it should produce time series with fluctuations that change away from 1/f towards ‘‘whiter” variation. The washing away of temporal correlations would indicate less integration of components of the underlying synergy, with lessening of structural richness and acquisition of degrees of freedom that begin to vary with greater independence (Kiefer, Riley, Shockley, Villard, & Van Orden, 2009). These changes should be evidenced as reduced levels of complexity. The aforementioned bimanual coordination task performed under three different perceptual conditions was used to test these expectations. Specifically, an anti-phase wrist flexion and extension coordination task was performed under no specific perceptual guidance, under temporal guidance with a metronome, and under visual guidance with a Lissajous plot. The temporal correlations and the complexity of time series produced under each condition were compared. Results might be relevant to explaining why little retention is observed after practice with some kinds of perceptual guidance.

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

15

2. Methods 2.1. Participants Thirty right-handed undergraduate students aged between 18 and 30 years (12 male and 18 female) volunteered to participate in partial fulfillment of a course requirement. None had neuromuscular conditions affecting perception-action or previous experience with the task. The experimental protocol was approved by the Institutional Review Board of the University of Connecticut. All participants provided informed consent. 2.2. Procedure and design Fig. 1 depicts the device used in the experiment. It consisted of two vertical handles attached to optical encoders (AutomationDirect TRD-SH2500-VD, with a resolution of 2500 counts per revolution). A screen prohibited vision of the hands and wrists. A computer monitor was used to produce a Lissajous plot for visual guidance and a speaker was used as a metronome for temporal guidance. Hand position data was recorded at 200 Hz. Participants sat in front of the device and were instructed to grasp its handles and move them rhythmically, both hands at the same pace (1:1 frequency), and both going leftward and then rightward in the horizontal plane (right wrist flexion corresponding to left wrist extension and vice versa). They were told to choose a movement speed and movement amplitude that felt comfortable, and to keep them as consistent as possible. All participants performed two trials of 1:1 anti-phase coordination between left and right hands in each of three perceptual conditions: no augmented perceptual guidance (NG), temporal guidance with a metronome (TG) and visual guidance with a Lissajous plot (VG), in this order. Therefore, NG trials were performed first, followed by TG trials, and VG trials last. NG was performed first in order to establish the preferred pace of movement for each participant. The pace served as a reference for TG and VG, as described below. All participants first performed the NG trials at a freely chosen pace and amplitude. The individual participant’s preferred frequency dictated the length of time needed to complete a minimum of 64 cycles necessary for analysis (see Section 2.3). Participants took approximately 5 to 9 min to complete their trials and the shortest series had 87 cycles. The mean frequency used in the two NG trials dictated the metronome pace for the TG condition. In the two TG trials, participants were told to synchronize hand cycles with the metronome sound cycles. A continuous sound with tone oscillation instead of discrete beats was used to avoid induction of ‘‘anchoring,” a local stabilization of the cycle near the discrete pacing signal (Byblow, Carson, & Goodman, 1994; Torre, Balasubramaniam, & Delignières, 2010). Finally, participants performed two VG trials. The Lissajous plot used in VG was a diagonal negatively inclined line on the computer monitor representing perfect anti-phase. The length of the line corresponded to a maximal amplitude of 60° (30° of wrist flexion and 30° of extension). The Lissajous plot established a reference for performance (Fig. 1). A dot represented the conjugate movement of the right (plotted on the horizontal axis) and left hands (plotted on the vertical axis). Participants were instructed to move the handles of the device in wrist flexion and extension so that the dot movement would correspond to the general shape and size of the line. Before the actual data collection of VG trials, participants were given five minutes of practice time to learn how to use the Lissajous guidance (Kovacs et al., 2009a, 2010). VG trials were performed less than a minute after practice.

Fig. 1. Experimental setup.

16

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

All 30 participants performed the 3 perceptual conditions as described above. However, for 15 participants (assigned by order of arrival), the metronome was on for the last half of practice time before VG. The frequency of the metronome was the same as used in the previous TG trials. It served to remind participants of the frequency they should try to replicate while using the Lissajous in the upcoming VG trials. This group was termed the Frequency Reminder (FR) group. The other 15 participants did not receive frequency reminders or requirements and were termed the No Reminder (NR) group. Contrast between FR and NR provided an assessment of the influence of the frequency requirement on VG. A summary of the experimental design is available in Fig. 2.

Fig. 2. Schema of experimental design. FR: Frequency Reminder group, NR: No Reminder group.

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

17

2.3. Data analysis The experiment was run with a custom-made Simulink model that received data from the encoders. Displacement data were imported into Matlab for analysis. A time derivative of each displacement series was generated after the displacement series were filtered with a low-pass Butterworth filter with a cut-off frequency of 5 Hz. The derivative series were then filtered with a cut-off frequency of 10 Hz. A peak-picking algorithm was used to define four kinds of events in a cycle: maximum wrist flexion (corresponding to peaks in the displacement series), maximum wrist extension (valleys in the displacement series), the inflexion point between maximum flexion and maximum extension (valleys in the derivative speed series), and the inflexion point between peak extension and peak flexion (peaks in the derivative speed series). The use of four events per cycle allowed for longer event series to be used in the analyses of temporal correlations and complexity. The events were used to generate series of cycle amplitudes, cycle periods, and discrete relative phase (Torre et al., 2007). The number of data points N of all series ranged from 348 to 2677, all above the minimum 256 necessary for estimating the structure of noise in a time series (Delignières et al., 2006). The three event series were analyzed with Detrended Fluctuation Analysis and Multiscale Complexity. Detrended Fluctuation Analysis was used to assess the strength of temporal correlations of the dominant hand series of amplitudes, periods, and discrete relative phases. Detrended Fluctuation Analysis assesses the strength of temporal correlations by characterizing the relationship between the mean magnitude of fluctuations in the series and the length of the intervals over which these fluctuations are determined (Peng, Havlin, Stanley, & Goldberger, 1995; Peng et al., 1993). In the present study, Detrended Fluctuation Analysis with linear detrending was performed on interval sizes varying from 16 to a maximum size shorter than or equal to ¼ of the series length, with even data spacing for improved accuracy with short time series (Almurad & Delignières, 2016). A customized Matlab routine (based on Ihlen, 2012) was used to calculate Detrended Fluctuation Analysis’s coefficient ⍺. To ascertain that the log-log relation was linear (a requisite for a meaningful interpretation of ⍺) the regression coefficient of the log fluctuation size on log interval length had to be greater than 0.95. Additionally, a regression with a quadratic term should not be significantly better than a linear regression according to a chi-square test on deviance statistics. Detrended Fluctuation Analysis results that did not pass both these criteria were not used in statistical analyses (only 5.5% of all estimates). Complexity was characterized with Multiscale Entropy (or its multivariate version, when appropriate) after Multivariate Empirical Mode Decomposition (Ahmed, Li, Cao, & Mandic, 2011; Ahmed & Mandic, 2011; Ahmed, Rehman, Looney, Rutkowski, & Mandic, 2012). Empirical Mode Decomposition is a data driven method specifically developed for decomposing nonlinear, non-stationary signals into their intrinsic frequency components (Huang et al., 1998; Rilling, Flandrin, & Gonçalvés, 2003). The successive components obtained with Empirical Mode Decomposition are the intrinsic mode functions. Different intrinsic mode functions capture the properties of the original signal on different time scales. Each intrinsic mode function is narrowband and monocomponent, with the characteristic frequency decreasing with the intrinsic mode function number. Thus, the first extracted intrinsic mode function is the highest frequency component in a signal containing plenty of detail. Even if the original signal is non-stationary, the intrinsic mode functions are much better conditioned and are typically quasi-stationary (Ahmed et al., 2012). Multivariate Empirical Mode Decomposition is capable of simultaneous decomposing two (Rilling, Flandrin, Gonçalvés, & Lilly, 2007), three or more time series (Rehman & Mandic, 2010) into intrinsic mode functions. The advantage of using Multivariate Empirical Mode Decomposition is its mode alignment property. The intrinsic mode functions generated for each series of the multivariate series set are the same in number and belong to the same frequency band, making their comparison meaningful (Ahmed et al., 2012). Multivariate Empirical Mode Decomposition was the first step in the procedures used to calculate Multiscale Entropy. For each participant, time series were used to compose multivariate sets that were decomposed with Multivariate Empirical Mode Decomposition. Series of relative phases measured four times a cycle were used to assemble multivariate sets with 6 equal length series (two trials in each of three conditions: NG, TG and VG). Series of right and left hand amplitudes measured four times a cycle were used to assemble multivariate sets with 12 equal length series (series of each hand for the two trials in the three conditions). Series of periods were treated likewise. Multivariate Empirical Mode Decomposition outputted one collection of intrinsic mode functions (corresponding to time scales) for each of the series composing the multivariate input set. Eight or nine time scales, which corresponded to the maximum number of intrinsic mode functions definable for the shortest series, were used for all series in subsequent steps. The second step in calculating Multiscale Entropy involved extracting the intrinsic mode functions for each original series, and using them to generate multiple intrinsic data scales. The present study followed procedures delineated by Hu and Liang (2012) and defined coarse-to-fine scales by the cumulative sums of sequential intrinsic mode functions. To define coarse-tofine scales, the sum of all intrinsic mode functions, all but the last one, all but the last two, all but the last three, and so on, were taken. This way the first scale contains the full original signal and the last contains the intrinsic mode function of higher frequency content (Hu & Liang, 2012). The third and final step was to calculate Sample Entropy or its multivariate version for each data scale. The entropy of a time series is a measure of its average uncertainty, or conversely, of the regularity or orderliness of a time series. Sample Entropy measures the negative logarithm of the conditional probability that two sequences that are similar for m points remain similar at the next point, within a tolerance r (Richman & Moorman, 2000). The multivariate version of Sample Entropy (Ahmed & Mandic, 2011; Ahmed et al., 2012) is based on multivariate embedded reconstruction with composite delay vectors, with a typical time lag t = 1. It enables entropy calculations for multivariate time series data, taking into

18

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

account both within and cross-series dependencies. As is usual in the literature, m was set to 2 and r was set to 0.15  the standard deviation of the standardized time series. Standardization—transformation to z-scores—was used prior to the Multivariate Sample Entropy calculation to allow for a similar threshold criterion across all time series. Previous studies indicated good statistical reproducibility for the univariate Sample Entropy with these parameter values (Costa et al., 2007; Richman & Moorman, 2000). In sum, the three steps produced Multiscale Sample Entropy enhanced by Multivariate Empirical Mode Decomposition for relative phase in each trial of each condition, and Multiscale Multivariate Sample Entropy enhanced by Multivariate Empirical Mode Decomposition for the bivariate series period and amplitude data of right and left hands in each trial and each condition. Performance was also characterized with traditional and circular statistical measures. Accuracy was quantified by taking the circular mean of relative phase over time within each trial. Stability was quantified by taking the circular standard deviation of relative phase over time within each trial. Movement pace was quantified by taking the mean of half cycle periods (in seconds) over time within each trial. All measures were averaged between the first and second NG, first and second TG, and first and second VG trials for each participant for improved reliability of estimates (Torre & Delignières, 2008; Torre et al., 2007). 2.4. Statistical comparisons Mixed 2 (group)  3 (perceptual conditions) ANOVAs were used do compare performance and temporal correlations. For complexity measures, Mixed 2 (group)  3 (perceptual conditions)  8 or 9 (time scales) ANOVAs were used. The HuynhFeldt correction for violation of sphericity was used when appropriate. When significant interaction effects or significant main effects of perceptual conditions were found, differences were further investigated with pre-planned contrasts. The significance level was set to 0.05. 3. Results 3.1. Task performance No significant effects were found on accuracy (mean relative phase over time in each trial). Stability was, however, significantly different between conditions, F(1.169, 32.727) = 37.75, p < 0.001), being lower for VG compared to TG, F(1, 28) = 32.10, p < 0.001, and to NG, F(1, 28) = 49.77, p < 0.001. Pace did not differ across conditions for the FR group but did differ across conditions for the NR group, F(1.090, 14.045) = 10.18, p = 0.006, given that participants were free to chose the pace for VG performance. Movement pace was significantly lower for VG compared to TG, F(1, 14) = 10.36, p = 0.007, and for VG compared to NG, F(1, 14) = 10.00, p = 0.007. Accuracy, stability and pace values averaged across participants are shown in Table 1.

Table 1 Performance results. Accuracy: mean of discrete relative phase over time, Stability: standard deviation of discrete relative phase over time, Pace: mean halfcycle period over time. Performance measure

Group

Condition

Mean

SD

Accuracy

FR

NG TG VG NG TG VG

178.70 178.10 176.95 175.60 176.98 179.66

4.42 5.90 7.92 5.99 8.27 4.80

NG TG VG NG TG VG

19.50 20.51 37.08 18.72 19.61 31.45

7.60 6.39 17.53 3.70 4.11 10.02

NG TG VG NG TG VG

0.39 0.38 0.43 0.43 0.42 0.76

0.07 0.07 0.18 0.08 0.07 0.40

NR

Stability

FR

NR

Pace

FR

NR

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

19

3.2. Temporal correlations As illustrated in Fig. 3, perceptual conditions significantly affected the temporal correlations (Detrended Fluctuation Analysis coefficient ⍺) of all three kinds of movement time series: relative phase, F(1.643, 46.00) = 36.61, p < 0.001, amplitudes, F (1.373, 37.083) = 52.96, p = 0.001, and periods, F(1.536, 41.461) = 35.38, p < 0.001. In general, VG decreased temporal correlations of the series of relative phases and amplitudes, while TG decreased temporal correlations of the series of periods. Repeated contrasts showed that temporal correlations for relative phase were significantly lower for VG compared to TG, F(1, 28) = 50.83, p < 0.001, and for VG compared to NG, F(1, 28) = 30.06, p < 0.001. Likewise, for the series of amplitudes, the value of the Detrended Fluctuation Analysis coefficient ⍺ was lower for VG compared to TG, F(1, 27) = 62.45, p < 0.001, and for VG compared to NG, F(1, 27) = 53.19, p < 0.001. An interaction effect and post hoc tests indicated that the effect of visual guidance on the temporal correlations of amplitudes was more intense when participants executed the movement at their preferred pace (Group B), t(27) = 3.04, p = 0.005. Perceptual guidance conditions had a distinct effect on periods. Repeated contrasts showed that their temporal correlation values were significantly lower for TG compared to VG, F(1, 27) = 42.03, p < 0.001, and for TG compared to NG, F(1, 27) = 38.38, p < 0.001 while they did not differ significantly between NG and VG (p > 0.328). 3.3. Complexity Perceptual conditions had significant main effects on multiscale entropy for the series of relative phases, F(2, 56) = 57.57, p < 0.001, amplitudes, F(1.463, 40.96) = 31.80, p < 0.001, and periods, F(1.180, 33.048) = 32.97, p < 0.001. Differently from temporal correlations, however, visual guidance produced lower complexity for all three kinds of movement series, i.e., not only phases and amplitudes but also periods (Figs. 4–6). Interactions between perceptual conditions and groups were not significant. For relative phase (Fig. 3), pre-planned contrasts indicated that averaged over groups and time scales, VG produced significantly lower complexity values compared to TG, F(1, 28) = 85.76, p < 0.001, and to NG, F(1, 28) = 78.36, p < 0.001. For amplitudes (Fig. 4), VG produced significantly lower complexity values compared to TG, F(1, 28) = 36.19, p < 0.001, and to NG, F(1, 28) = 34.91, p < 0.001. For periods (Fig. 5), VG also produced significantly lower complexity values compared to both TG, F(1, 28) = 38.38, p < 0.001) and NG, F(1, 28) = 29.92, p < 0.001, with TG exhibiting higher complexity than NG, F(1, 28) = 5.24, p = 0.03. 4. Discussion This study investigated differences in the dynamics of a coordination task under different perceptual guidance conditions. The expectation was that visual guidance of a Lissajous plot would change the typical 1/f fluctuations seen in voluntary selfcontrolled coordination towards ‘‘whiter” variation. These changes were expected to correspond to a reduction of complexity, that is, a decrease in the structural richness of underlying synergies. To our knowledge, this is the first study investigating the effects of visual guidance by Lissajous on temporal correlations and complexity of a bimanual coordination task. Previous studies have investigated the effects of temporal guidance by metronomes on the on temporal correlations of bimanual coordination tasks, using Detrended fluctuation analysis. Our results are consistent with previous findings. Under self-paced conditions, ⍺ coefficients for the series of periods have been reported to be around 0.86 (±0.18) (Torre et al., 2010) and range from 0.73 to 0.79 for the series of relative phases (Torre et al., 2007). In this study, both the series of periods and

Fig. 3. Effects of perceptual conditions on temporal correlations (Detrended Fluctuation Analysis coefficient ⍺) averaged across groups FR and NR.

20

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

Fig. 4. Effects of perceptual conditions on complexity (Sample Entropy) of relative phase as a function of perceptual conditions and scales, averaged across groups FR and NR.

Fig. 5. Effects of perceptual conditions on complexity (Multivariate Sample Entropy) of amplitudes as a function of perceptual conditions and scales, averaged across groups FR and NR.

Fig. 6. Effects of perceptual conditions on complexity (Multivariate Sample Entropy) of periods as a function of perceptual conditions and scales, averaged across groups FR and NR.

phases produced ⍺ coefficients around 0.7 when performed with no guidance. This is consistent with 1/f structure. More importantly, under metronome guidance, the ⍺ coefficient of the series of periods decreased to values around 0.4, which indicates anti-persistent fluctuation, as has previously been reported (Torre et al., 2010).

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

21

Before considering results of visual guidance on temporal correlation and complexity measures, it is important to point out that overall, performance as characterized by mean phase was similar between all groups and conditions. Contrary to expectations (Kovacs et al., 2009b) however, the stabilizing effect of Lissajous guidance was not reproduced in this study, as variability of phase was greater under visual guidance. Handedness affects bimanual coordination, leading to a phase shift with a small lead of the dominant hand (Treffner & Turvey, 1996). Individuals are usually unaware of this shift. The present study might have used a finer resolution than previous studies, allowing better discrimination of small deviations, as Lissajous plots improve error detection and correction mechanisms (Kennedy et al., 2016). Individuals apparently produced repeated attempts of phase corrections under visual guidance and this resulted in increased phase variability (Schmidt & Bjork, 1992). Also, in this study, participants performed the 1:1 coordination task around 0,8 Hz. In contrast, Kovacs et al. (2009b) instructed participants to perform task above 1 Hz. Slower movements are more sensitive to perturbations, which can result in higher variability (Leinen et al., 2016). The absence of a stabilizing effect of VG, however, did not preclude its effects on underlying coordination dynamics. The expectation that a reduction in complexity as measured by Multiscale Entropy would correspond to weaker temporal correlations as measured by Detrended Fluctuation Analysis was only partially confirmed. Visual guidance produced a reduction in degree of temporal correlations of phases and amplitudes, and at the same time it reduced complexity not only for phases and amplitudes but also for periods. Also, temporal guidance reduced the temporal correlation of periods compared to no guidance, but complexity of periods was higher under temporal guidance than under no guidance. Temporal correlations and complexity thus appear to vary somewhat independently. These results suggest the following. First, Detrended Fluctuation Analysis and Multiscale Entropy measures appear to capture different aspects of coordination dynamics. Related dissimilarities have been indicated before (Duarte & Sternard, 2008). Results of Detrended Fluctuation Analysis appeared to be specific to the nature of guidance (temporal guidance affected the series of periods and visual guidance affected the series of amplitudes and phases) while complexity measures capture broader effects on underlying coordination. Second, and relatedly, these results indicate that attentional and intentional processes related to external guiding may not induce a kind of ‘‘oversimplification” of the system, as suggested by Hausdorff et al. (1996). Evidence suggests that the 1/f structure of series of periods is washed away under external pacing. However, long-range correlations in the series of asynchronies to the pacing signal suggest that the intrinsic complexity of the system is still at work, only expressed differently in overt performance (Delignières & Torre, 2009). Possibly, Multiscale Entropy captures, in the multivariate series of periods, this hidden aspect of complexity, while Detrended Fluctuation Analysis captures the relation between voluntary control and external constraints. Visual guidance, in any case, appears to have broader effects on coordination dynamics than temporal guidance. Lower entropy was observed in all time scales for the series of phases, amplitudes and periods. This broader effect is possibly related to the fact that a Lissajous plot provides low dimensional spatiotemporal information about the two limbs, conveying information for relative phase, a collective variable. The ability to perceptually resolve phase and phase variability play an important role in the coordination of movement patterns (Bingham, 2004). Also, the early phase of coordination learning is characterized by discovery of the relevant collective variable (Amazeen, 1996; Mitra, Amazeen, & Turvey, 1998). The results suggest, however, that the perceptual simplification of the task’s collective variable, and the attendant enhancement of performance, is accompanied by changes in the organization of movement details that may render task performance less dexterous. Visual guidance with phase-specific information introduces qualitative differences in the nature of the dynamical organization of movement, changing its complexity and temporal correlations. Results of the present study can also have implications for learning. Future studies could investigate whether these changes changes in complexity and temporal correlations of coordination related to poor retention observed after practice with Lissajous plots. Traditionally, poor learning after practice with frequent feedback has been explained with reference to the guidance hypothesis (Salmoni et al., 1984), according to which too frequent feedback prevents learners from focusing on their own movements. Such focus would be a precondition for the development of an effective movement representation, which would be, in turn, a precondition for independent production of movement. Frequent feedback is not always detrimental, however. Several studies demonstrate the superior learning effects of focusing attention on environmental effects of movement rather than on movement itself, for a large variety of tasks (reviewed in Wulf, 2013). With an external rather then internal focus of attention, more frequent feedback has been beneficial rather than detrimental (Wulf, Chiviacowsky, Schiller, & Ávila, 2010; Wulf, McConnel, Gärtner, & Schwarz, 2002). Lissajous effects, nevertheless, contradict this evidence: the aforementioned kind of augmented information does establish an external focus of attention but impairs learning if used frequently (Kovacs & Shea, 2011). Why? Maybe the answer lies in the reactive rather than productive movement dynamics furnished by this kind of external constraint. The conflict may be resolved in the investigation of structural richness of motor synergies (complexity) and voluntary control (temporal correlations) under distinct perceptual constraints. 5. Conclusions This study investigated temporal correlations and complexity of a coordination task performed under no specific perceptual guidance, under temporal guidance with a metronome, and under visual guidance with a Lissajous plot. Temporal correlations of amplitudes and relative phases deviated from the typical 1/f variation towards more random variation under visual guidance. The same was observed for the series of periods under temporal guidance. Complexity levels for all time

22

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

series were lower in visual guidance, but higher for periods under temporal guidance. Perceptual simplification of the task’s goal may produce enhancement of performance, but it is accompanied by changes in the details of movement organization that may be relevant to explain dependence and poor retention after practice under guidance. Funding This work was supported by the Brazilian Council of Scientific and Technological Development [Grant 200676/2009-1], and by the University of Connecticut [Outstanding Scholar and Doctoral Dissertation Fellowships] awarded to the first author. References Ahmed, M. U., Li, L., Cao, J., & Mandic, D. P. (2011). Multivariate multiscale entropy for brain consciousness analysis. In Engineering in Medicine and Biology Society, EMBC, 2011 Annual international conference of the IEEE (pp. 810–813). IEEE. doi: http://dx.doi.org/10.1109/IEMBS.2011.6090185. Ahmed, M. U., & Mandic, D. P. (2011). Multivariate multiscale entropy: A tool for complexity analysis of multichannel data. Physical Review E, 84(6), 061918. http://dx.doi.org/10.1103/PhysRevE.84.061918. Ahmed, M. U., Rehman, N., Looney, D., Rutkowski, T. M., & Mandic, D. P. (2012). Dynamical complexity of human responses: a multivariate data-adaptive framework. Bulletin of the Polish Academy of Sciences. Technical Sciences, 60 (3), 433–445. http://dx.doi.org/10.2478/v10175-012-0055-0. Almurad, Z. M. H., & Delignières, D. (2016). Evenly spacing in detrended fluctuation analysis. Physica A, 451, 63–69. http://dx.doi.org/10.3389/ fphys.2012.00141. Amazeen, P. G. (1996). Learning a bimanual rhythmic coordination: Schemas as dynamics [Doctoral dissertations]. Paper AAI9717500. . Bingham, G. P. (2004). A perceptually driven dynamical model of bimanual rhythmic movement (and phase perception). Ecological Psychology, 16(1), 45–53. http://dx.doi.org/10.1207/s15326969eco1601_6. Byblow, W. D., Carson, R. G., & Goodman, D. (1994). Expressions of asymmetries and anchoring in bimanual coordination. Human Movement Science, 13(1), 3–28. http://dx.doi.org/10.1016/0167-9457(94)90027-2. Carson, R. G., Riek, S., Smethurst, C. J., Párraga, J. F. L., & Byblow, W. D. (2000). Neuromuscular-skeletal constraints upon the dynamics of unimanual and bimanual coordination. Experimental Brain Research, 131(2), 196–214. http://dx.doi.org/10.1007/s002219900272. Chen, Y., Ding, M., & Kelso, J. S. (1997). Long memory processes (1/fa Type) in human coordination. Physical Review Letters, 79(22), 4501–4504. http://dx.doi. org/10.1103/PhysRevLett. 79.4501. Costa, M., Goldberger, A. L., & Peng, C. K. (2002). Multiscale entropy analysis of complex physiologic time series. Physical Review Letters, 89(6), 068102.1–068102.4. 10.1103/PhysRevLett. 89.068102. Costa, M., Priplata, A. A., Lipsitz, L. A., Wu, Z., Huang, N. E., Goldberger, A. L., & Peng, C. K. (2007). Noise and poise: Enhancement of postural complexity in the elderly with a stochastic-resonance–based therapy. Europhysics Letters, 77(6). http://dx.doi.org/10.1209/0295-5075/77/68008. 68008.p1–68008.p5. Debaere, F., Wenderoth, N., Sunaert, S., Van Hecke, P., & Swinnen, S. P. (2003). Internal vs external generation of movements: Differential neural pathways involved in bimanual coordination performed in the presence or absence of augmented visual feedback. Neuroimage, 19(3), 764–776. http://dx.doi.org/ 10.1016/S1053-8119(03)00148-4. Delignieres, D., Ramdani, S., Lemoine, L., Torre, K., Fortes, M., & Ninot, G. (2006). Fractal analyses for ‘short’ time series: A re-assessment of classical methods. Journal of Mathematical Psychology, 50(6), 525–544. http://dx.doi.org/10.1016/j.jmp.2006.07.004. Delignieres, D., & Torre, K. (2009). Fractal dynamics of human gait: A reassessment of the 1996 data of Hausdorff et al.. Journal of Applied Physiology, 106(4), 1272–1279. http://dx.doi.org/10.1152/japplphysiol.90757.2008. Diniz, A., Wijnants, M. L., Torre, K., Barreiros, J., Crato, N., Bosman, A. M., ... Delignières, D. (2011). Contemporary theories of 1/f noise in motor control. Human Movement Science, 30(5), 889–905. http://dx.doi.org/10.1016/j.humov.2010.07.006. Duarte, M., & Sternad, D. (2008). Complexity of human postural control in young and older adults during prolonged standing. Experimental Brain Research, 191(3), 265–276. http://dx.doi.org/10.1007/s00221-008-1521-7. Hausdorff, J. M., Purdon, P. L., Peng, C. K., Ladin, Z. V. I., Wei, J. Y., & Goldberger, A. L. (1996). Fractal dynamics of human gait: Stability of long-range correlations in stride interval fluctuations. Journal of Applied Physiology, 80(5), 1448–1457. Hu, M., & Liang, H. (2012). Adaptive multiscale entropy analysis of multivariate neural data. IEEE Transactions on Biomedical Engineering, 59(1), 12–15. http:// dx.doi.org/10.1109/TBME.2011.2162511. Huang, N. E., Shen, Z., Long, S. R., Wu, M. C., Shih, H. H., Zheng, Q., ... Liu, H. H. (1998). The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis. Proceedings of the Royal Society of London. Series A, 454(1971), 903–995. http://dx.doi.org/10.1098/ rspa.1998.0193. Ihlen, E. A. (2012). Introduction to multifractal detrended fluctuation analysis in Matlab. Frontiers in Physiology, 3(141), 1–18. http://dx.doi.org/10.3389/ fphys.2012.00141. Ihlen, E. A., & Vereijken, B. (2010). Interaction-dominant dynamics in human cognition: Beyond 1/f a fluctuation. Journal of Experimental Psychology: General, 139(3), 436–463. http://dx.doi.org/10.1037/a0019098. Kelso, J. A. (1984). Phase transitions and critical behavior in human bimanual coordination. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology, 246(6), R1000–R1004. Kelso, J. S., Fink, P. W., DeLaplain, C. R., & Carson, R. G. (2001). Haptic information stabilizes and destabilizes coordination dynamics. Proceedings of the Royal Society of London. Series B: Biological Sciences, 268(1472), 1207–1213. http://dx.doi.org/10.1098/rspb.2001.1620. Kennedy, D. M., Wang, C., Panzer, S., & Shea, C. H. (2016). Continuous scanning trials: Transitioning through the attractor landscape. Neuroscience Letters, 610, 66–72. http://dx.doi.org/10.1016/j.neulet.2015.10.073. Kiefer, A. W., Riley, M. A., Shockley, K., Villard, S., & Van Orden, G. C. (2009). Walking changes the dynamics of cognitive estimates of time intervals. Journal of Experimental Psychology: Human Perception and Performance, 35(5), 1532–1541. http://dx.doi.org/10.1037/a0013546. Kloos, H., & Van Orden, G. (2010). Voluntary behavior in cognitive and motor tasks. Mind and Matter, 8(1), 19–43. Kovacs, A. J., Buchanan, J. J., & Shea, C. H. (2009a). Bimanual 1: 1 with 90° continuous relative phase: Difficult or easy! Experimental Brain Research, 193(1), 129–136. http://dx.doi.org/10.1007/s00221-008-1676-2. Kovacs, A. J., Buchanan, J. J., & Shea, C. H. (2009b). Using scanning trials to assess intrinsic coordination dynamics. Neuroscience Letters, 455(3), 162–167. http://dx.doi.org/10.1016/j.neulet.2009.02.046. Kovacs, A. J., Buchanan, J. J., & Shea, C. H. (2010). Impossible is nothing: 5: 3 and 4: 3 multi-frequency bimanual coordination. Experimental Brain Research, 201(2), 249–259. http://dx.doi.org/10.1007/s00221-009-2031-y. Kovacs, A. J., & Shea, C. H. (2011). The learning of 90 continuous relative phase with and without Lissajous feedback: External and internally generated bimanual coordination. Acta Psychologica, 136(3), 311–320. http://dx.doi.org/10.1016/j.actpsy.2010.12.004. Leinen, P., Vieluf, S., Kennedy, D., Aschersleben, G., Shea, C. H., & Panzer, S. (2016). Life span changes: Performing a continuous 1: 2 bimanual coordination task. Human Movement Science, 46, 209–220. http://dx.doi.org/10.1016/j.humov.2016.01.004.

D.V. Vaz et al. / Human Movement Science 54 (2017) 13–23

23

Maslovat, D., Brunke, K. M., Chua, R., & Franks, I. M. (2009). Feedback effects on learning a novel bimanual coordination pattern: Support for the guidance hypothesis. Journal of Motor Behavior, 41(1), 45–54. http://dx.doi.org/10.1080/00222895.2009.10125923. Mechsner, F., Kerzel, D., Knoblich, G., & Prinz, W. (2001). Perceptual basis of bimanual coordination. Nature, 414(6859), 69–73. http://dx.doi.org/10.1038/ 35102060. Mechsner, F., & Knoblich, G. (2004). Do muscles matter for coordinated action? Journal of Experimental Psychology: Human Perception and Performance, 30(3), 490–503. http://dx.doi.org/10.1037/0096-1523.30.3.490. Mitra, S., Amazeen, P. G., & Turvey, M. T. (1998). Intermediate motor learning as decreasing active (dynamical) degrees of freedom. Human Movement Science, 17(1), 17–65. http://dx.doi.org/10.1016/S0167-9457(97)00023-7. Park, H., Collins, D. R., & Turvey, M. T. (2001). Dissociation of muscular and spatial constraints on patterns of interlimb coordination. Journal of Experimental Psychology: Human Perception and Performance, 27(1), 32–47. http://dx.doi.org/10.1037/0096-1523.27.1.32. Peng, C. K., Havlin, S., Stanley, H. E., & Goldberger, A. L. (1995). Quantification of scaling exponents and crossover phenomena in nonstationary heartbeat time series. Chaos: An Interdisciplinary Journal of Nonlinear Science, 5(1), 82–87. http://dx.doi.org/10.1063/1.166141. Peng, C. K., Mietus, J., Hausdorff, J. M., Havlin, S., Stanley, H. E., & Goldberger, A. L. (1993). Long-range anti-correlations and non-Gaussian behavior of the heartbeat. Physical Review Letter, 70(9), 1343–1346. http://dx.doi.org/10.1103/PhysRevLett. 70.1343. Rehman, N., & Mandic, D. P. (2010). Multivariate empirical mode decomposition. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Science, 466(2117), 1291–1302. http://dx.doi.org/10.1098/rspa.2009.0502. Richman, J. S., & Moorman, J. R. (2000). Physiological time-series analysis using approximate entropy and sample entropy. American Journal of PhysiologyHeart and Circulatory Physiology, 278(6), H2039–H2049. Riek, S., & Woolley, D. (2005). Hierarchical organization of neuro-anatomical constraints in interlimb coordination. Human Movement Science, 24(5–6), 798–814. http://dx.doi.org/10.1016/j.humov.2005.10.002. Riley, M. A., & Turvey, M. T. (2002). Variability and determinism in motor behavior. Journal of Motor Behavior, 34(2), 99–125. http://dx.doi.org/10.1016/j. humov.2005.10.002. Rilling, G., Flandrin, P., & Gonçalvés, P. (2003). On empirical mode decomposition and its algorithms. In IEEE-EURASIP workshop on nonlinear signal and image processing NSIP (Vol. 3, pp. 8–11). Rilling, G., Flandrin, P., Gonçalvés, P., & Lilly, J. M. (2007). Bivariate empirical mode decomposition. Signal Processing Letters, IEEE, 14(12), 936–939. http://dx. doi.org/10.1109/LSP.2007.904710. Ronsse, R., Puttemans, V., Coxon, J. P., Goble, D. J., Wagemans, J., Wenderoth, N., & Swinnen, S. P. (2011). Motor learning with augmented feedback: modalitydependent behavioral and neural consequences. Cerebral Cortex, 21(6), 1283–1294. http://dx.doi.org/10.1093/cercor/bhq209. Salmoni, A. W., Schmidt, R. A., & Walter, C. B. (1984). Knowledge of results and motor learning: a review and critical reappraisal. Psychological Bulletin, 95(3), 355–386. http://dx.doi.org/10.1037/0033-2909.95.3.355. Salter, J. E., Wishart, L. R., Lee, T. D., & Simon, D. (2004). Perceptual and motor contributions to bimanual coordination. Neuroscience Letters, 363(2), 102–107. http://dx.doi.org/10.1016/j.neulet.2004.03.071. Schmidt, R. A., & Bjork, R. A. (1992). New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training. Psychological Science, 3(4), 207–217. Shea, C. H., Buchanan, J. J., & Kennedy, D. M. (2016). Perception and action influences on discrete and reciprocal bimanual coordination. Psychonomic Bulletin & Review, 23(2), 361–386. http://dx.doi.org/10.3758/s13423-015-0915-3. Swinnen, S. P., & Wenderoth, N. (2004). Two hands, one brain: Cognitive neuroscience of bimanual skill. Trends in Cognitive Sciences, 8(1), 18–25. http://dx. doi.org/10.1016/j.tics.2003.10.017. Temprado, J. J., Salesse, R., & Summers, J. J. (2007). Neuromuscular and spatial constraints on bimanual hand-held pendulum oscillations: Dissociation or combination? Human Movement Science, 26(2), 235–246. http://dx.doi.org/10.1016/j.humov.2007.01.012. Temprado, J. J., & Swinnen, S. P. (2005). Dynamics of learning and transfer of muscular and spatial relative phase in bimanual coordination: Evidence for abstract directional codes. Experimental Brain Research, 160(2), 180–188. http://dx.doi.org/10.1007/s00221-004-1998-7. Temprado, J. J., Swinnen, S. P., Carson, R. G., Tourment, A., & Laurent, M. (2003). Interaction of directional, neuromuscular and egocentric constraints on the stability of preferred bimanual coordination patterns. Human Movement Science, 22(3), 339–363. http://dx.doi.org/10.1016/S0167-9457(03)00049-6. Torre, K., Balasubramaniam, R., & Delignières, D. (2010). Oscillating in synchrony with a metronome: Serial dependence, limit cycle dynamics, and modeling. Motor Control, 14(3), 323–343. http://dx.doi.org/10.1123/mcj.14.3.323. Torre, K., & Delignières, D. (2008). Distinct ways of timing movements in bimanual coordination tasks: Contribution of serial correlation analysis and implications for modeling. Acta Psychologica, 129(2), 284–296. http://dx.doi.org/10.1016/j.actpsy.2008.08.003. Torre, K., Delignières, D., & Lemoine, L. (2007). 1/fb fluctuations in bimanual coordination: An additional challenge for modeling. Experimental Brain Research, 183(2), 225–234. http://dx.doi.org/10.1007/s00221-007-1035-8. Treffner, P. J., & Turvey, M. T. (1996). Symmetry, broken symmetry, and handedness in bimanual coordination dynamics. Experimental Brain Research, 107(3), 463–478. http://dx.doi.org/10.1007/BF00230426. Turvey, M. T. (2007). Action and perception at the level of synergies. Human Movement Science, 26(4), 657–697. http://dx.doi.org/10.1016/j. humov.2007.04.002. Van Orden, G. C., Holden, J. G., & Turvey, M. T. (2003). Self-organization of cognitive performance. Journal of Experimental Psychology: General, 132(3), 331–350. http://dx.doi.org/10.1037/0096-3445.132.3.331. Van Orden, G., Kloos, H., & Wallot, S. (2011). Living in the pink: Intentionality, wellness, and complexity. In C. Hooker (Ed.), Handbook of the philosophy of science. Philosophy of complex systems (10, pp. 639–682). Amsterdam, Netherlands: Elsevier. West, B. J., & Griffin, L. (1998). Allometric control of human gait. Fractals, 6(02), 101–108. http://dx.doi.org/10.1142/S0218348X98000122. Wulf, G. (2013). Attentional focus and motor learning: a review of 15 years. International Review of Sport and Exercise Psychology, 6(1), 77–104. http://dx.doi. org/10.1080/1750984X.2012.723728. Wulf, G., Chiviacowsky, S., Schiller, E., & Ávila, L. T. G. (2010). Frequent external focus feedback enhances motor learning. Frontiers in Psychology, 1, 190. http://dx.doi.org/10.3389/fpsyg.2010.00190. Wulf, G., McConnel, N., Gärtner, M., & Schwarz, A. (2002). Enhancing the learning of sport skills through external-focus feedback. Journal of Motor Behavior, 34(2), 171–182. http://dx.doi.org/10.1080/00222890209601939. Zanone, P. G., & Kelso, J. A. (1992). Evolution of behavioral attractors with learning: Nonequilibrium phase transitions. Journal of Experimental Psychology: Human Perception and Performance, 18(2), 403–421. http://dx.doi.org/10.1037/0096-1523.18.2.403.