Energetic Costs of Incidental Visual Coupling during Treadmill Running


Medicine & Science in Sports & Exercise:
doi: 10.1249/MSS.0b013e31817057a8

Purpose: To determine the role of visual-spatial information in stabilizing movement during treadmill locomotion.

Methods: Physiological, biomechanical, and psychological indices of coordination stability were recorded while participants were visually coupled with a whole-body image of themselves during treadmill locomotion. Ten participants ran on a treadmill under three visual conditions: two dynamic images (symmetrical, mirror image; asymmetrical, reversed mirror image) and one static. Performance was examined at two speeds.

Results: Participants ran more economically (mL·kg·min−1) when they were visually coupled with a symmetrical rather than with an asymmetrical or static image. An asymmetrical coupling resulted in increased variability in footfall position at the faster speed, in comparison to the symmetrical and static conditions. However, at slower speeds, footfall variability and frequency were higher under both dynamic visual conditions in comparison to the static control. Changes in metabolic economy (mL·kg·min−1) were only partially mediated by movement kinematics.

Conclusion: Visual information influences treadmill locomotion and associated measures of stability even when there is no intention to coordinate with external stimuli.

Author Information

1School of Social Sciences and Law, The University of Teesside, Middlesbrough, UNITED KINGDOM; 2School of Human Kinetics, University of British Columbia, Vancouver, CANADA; and 3Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, Liverpool, UNITED KINGDOM

Address for correspondence: Daniel L. Eaves, School of Social Sciences and Law, The University of Teesside, Middlesbrough, TS1 3BA, United Kingdom; E-mail: d.eaves@tees.ac.uk.

Submitted for publication November 2007.

Accepted for publication February 2008.

Article Outline

Many everyday movements involve coordination between the limbs. When two limbs are moving simultaneously, there is a tendency to synchronize the movements of each in both time and space (15). Researchers examining bimanual coordination have raised questions about the principles underlying the production of these movement patterns. It seems that perceptual-spatial constraints play a dominant role in the emergence of coordinated movements both between limbs (16) and people (e.g., (20)). In this experiment, we examined whether visual information serves to encourage coordination stability in a whole-body task, where participants are visually coupled with a dynamic or static image of themselves when running on a treadmill. Of interest was the potential for various "effort"-related variables (i.e., physiological, biomechanical, and psychological) to provide insight into the processes that underlie coordination.

Certain spatial-temporal patterns of movement are generated more readily than others during concurrent arm or finger movements in the horizontal plane (for reviews, see Kelso (10) and Swinnen and Wenderworth (25)). These movements have been termed in-phase and antiphase. Typically, they reflect bimanual limb motion performed in symmetry (i.e., 0° relative phase) or in asymmetry (180° relative phase), respectively. The relative stability of these movements has been well documented since Haken et al. (9) formulated a mathematical model to describe (and predict) empirical phenomenon associated with these movement patterns. Antiphase becomes destabilized at higher frequencies, resulting in a phase transition to the more stable in-phase mode. The reasons for this behavior have been debated, with interpretations placing varying degrees of emphasis on neural, muscular, and perceptual contributions (3). In- and antiphase movements could be described in terms of homologous and nonhomologous muscles, respectively. Consequently, a motoric or neuromuscular view of coordination dominated explanations for the relative stability of these movement patterns. Neural crosstalk or neural leakage at higher movement frequencies was one explanation for the transition from antiphase to in-phase (25). However, perceptual-spatial interpretations of this phenomenon have also been considered, with recent work suggesting that visual information plays a dominant role in determining coordination stability.

Original evidence supporting a dominant role for visual perception in bringing about certain coordination preferences came from research into interpersonal coordination (19-21,26). Rhythmic oscillations of the legs and arms of two seated people facing each other seemed to be coordinated according to the same dynamic principles as those previously observed in single-person coordination (20), even if there was no attempt to intentionally couple (19). Schmidt et al. (20) predicted that both the "formation" and "tuning" (8) of coordinated states is an inherently perceptual process. However, there is evidence that the strength of coupling in the in-phase mode is generally greater during intrapersonal than interpersonal coordination, which would necessitate consideration of neuromuscular factors (26).

A dominant role for perceptual information in the strength of the coupling between two effectors has also been proposed by Mechsner et al. (14-16). In their initial studies, Mechsner et al. (16) used three experimental paradigms: a finger oscillation task, a bimanual finger tapping task, and a bimanual circle drawing task, where the goal of each task was visually defined. Performance was most stable when movements were visually in-phase (i.e., spatially symmetrical), regardless of the effect or orientation and position and hence neuromuscular contributions. They concluded that the external properties of the task dominated interlimb stability, arguing for the dominance of perceptual over motoric constraints. Although this view has come under criticism (e.g., (18,28,29)), the role of visual-perceptual information in influencing both the type and strength of various coordination modes is largely unquestioned.

Movement stability is usually inferred from kinematic variability (10). However, there is evidence that attentional and metabolic measures can also provide insights into the processes underlying coordination preferences. For example, reduced reaction time (RT) scores in a dual-task paradigm are closely linked to reductions in movement stability (27,32). These findings have been replicated in tasks involving intentional coordination in both between- and within-person scenarios (e.g., (26)). Other research indicates that metabolic energy expenditure is a prime candidate for optimization in the establishment of movement patterns (1,22,24). Measures of V˙O2 have been shown to differentiate across relative phase patterns in a unique bicycle ergometer task (11,23) where high intercorrelations were observed between kinematic, metabolic, and attentional variables.

In summary, there are converging lines of evidence to suggest that influences external to the motor system play an important role in the stabilization and destabilization of preferred coordination. Moreover, these influences are observed at many levels of the motor system. According to Swinnen and Wenderoth (25), the more constraints acting in coalition with each other, the more stable and accurate the coordination pattern will be. However, the attentional (26,27,32) and metabolic (11,23) effects discussed in the above paragraphs have only been observed during intentional coordination. In the following study, we explore how these variables are mediated by perceptual constraints, as afforded during incidental coupling with dynamic images akin to in-phase and antiphase.

We predicted that visual coupling with symmetrical limb movements (i.e., a dynamic mirror image) during treadmill locomotion would afford a greater level of movement stability than an asymmetrical and/or static image. Furthermore, increased levels of stability afforded under a symmetrical coupling should be indexed through increased metabolic economy and reductions in attentional load, footfall frequency, and footfall variability. An increment in treadmill speed during the experimental runs was expected to heighten the energetic demands on the motor system and accentuate any effects of the vision conditions.

Back to Top | Article Outline



Participants were male volunteers (N = 10) whose mean age was 21.9 yr (SD = 4.02 yr, range = 16-28 yr). Ethical approval was obtained for the study from Liverpool John Moores University. Participants then provided (parental where appropriate) written informed consent before testing. Participants were recreational runners, regularly undertaking a minimum of one and a maximum of three 20- to 30-min runs per week. The participant group had a mean height of 1.76 m (SD = 0.26 m) and mean body mass of 74.6 kg (SD = 18.2 kg).

Back to Top | Article Outline

The treadmill used was a Woodway ELG55 running machine (Weil am Rhein, Germany). Expired air was collected using standard Douglas bag protocol. This procedure consisted of a mouthpiece connected to a circuit of Douglas bags via tubing. The mouthpiece was connected to a unidirectional valve to ensure that inspired air was acquired directly from atmospheric sources and that expired air was breathed directly into the Douglas bags. A nose clip was fitted to prevent any nasal expiration. Analysis of the expired air was carried out to obtain measures of oxygen (V˙O2) and carbon dioxide (V˙CO2) concentration using a Servomex 1440 gas analysis system (East Sussex, England). The Servomex 1440 was recalibrated after every trial with gases of known concentration. Measurement of expired air volume was then performed using a Harvard dry gas meter (Kent, England). HR was recorded using a chest and wrist mounted Polar Sports Tester telemetric HR monitor (Kempele, Finland).

A Panasonic NV-MS5 SVHS movie camera (Northampton, England) was used to film the participants' face-on while they ran. A Sharp XG-NV2E LCD projector (Wrexham, Wales) displayed images received from the movie camera as either normal or reversed live video footage (i.e., dynamic concurrent feedback) or alternatively as a static image on a screen situated 2.5 m in front of the treadmill. The projection screen measured 2.5 m (h) × 2 m (w) where images presented were scaled approximately to body size. To avoid any distractions, the camera and projector were positioned so that each participant's head was occluded from view on all runs. A custom-built reaction timer was connected to a handgrip-style switch and to four individual red LED bulbs. The LEDs were positioned behind the projection screen so that they shone brightly through the material and were moveable so that they could be aligned with the lower arms and legs of each participant's image on the screen. The LED were activated by the experimenter in a predetermined randomized sequence that was constant across participants and experimental conditions. The frequency of LED activation was constrained to no less than 20 s and no more than 90 s apart. If no reaction was initiated within 10 s of stimulus presentation, runners were reminded to maintain their focus on the screen.

All kinematic data were gathered using three infrared motion analysis cameras (Pro-Reflex; Qualisys, Gothenburg, Sweden) sampling at 240 Hz. The cameras were positioned at the side of the treadmill, on each participant's right side. A reflective marker was fixed to the right lateral malleolus (ankle), the distal head of their right fifth metatarsal (toe), and the distal head of their left first metatarsal (toe).

Back to Top | Article Outline

Before running under experimental conditions, each participant's height and body mass were recorded, as was room temperature and barometric pressure. Maximal oxygen uptake (V˙O2max) was then assessed using a graded exercise test to volitional exhaustion, as stipulated by Bird and Davison (2). All participants reached volitional exhaustion within 12 min. A series of regression analyses were performed on the V˙O2max data so that treadmill speed (km·h−1) could be standardized as a percentage of V˙O2 relative to each participant's V˙O2max score. Experimental data collection took place for a minimum of 3 d after the V˙O2max test, and the sessions were scheduled so that there was one rest day between each run. This design was used to maintain consistency in the physiological recovery periods between runs. Sessions were performed at approximately the same time of day (SD = 62 min).

During each session, participants were required to run at two different speeds. The first 15 min of each run was performed at a treadmill speed equivalent to 60% of each participant's V˙O2max (Slow), after which the speed was increased to 80% V˙O2max (Fast) for the remaining 5 min. Exercise at 60% V˙O2max is widely recognized as a comfortable intensity at which to run in terms of the physiological effects of fatigue (13). Exercise performed at 80% V˙O2max is generally above the body's anabolic threshold (13).

Participants ran on the treadmill while facing a video image of themselves projected on to the screen directly in front of them. Participants were instructed to pay full attention to the image throughout the experiment. They were informed that by using the handheld switch, they were to react as fast as possible to separate illuminations of individual LED stimuli that would appear intermittently on the screen in the lower arm and leg areas of the image. This manipulation served two purposes: 1) to give some indication of the attentional demands as assessed through RT and 2) to ensure that visual coupling with areas associated with phased limb movement was maintained throughout each run. Following these instructions, an HR monitor was fitted, and briefings were given on the scheduling and duration of oxygen samples to be taken.

Back to Top | Article Outline
Experimental Conditions

Participants performed each run on separate days under one of three experimental conditions, which differed only in the nature of the visual information available. There were two dynamic experimental conditions. The first required each participant to run while watching a live video feedback of themselves face-on, as would be the case when treadmill running in front of a standard mirror. The dynamic image provided concurrent feedback of limb movement that always appeared to be phased in symmetry (or 0° relative phase) with the participant's own limb movements during treadmill running. This condition was called "Symmetrical." Under the second condition, each participant ran while visually coupled with a mirror image of his body reversed in the sagittal plane. The implications of this condition were that the participants saw their body face-on while running (as in a normal mirror), but their actions were reversed so that when they moved their right hand, for example, it was their left hand that moved in the image. Thus, limb movements in the image appeared to be phased in asymmetry (or 180° relative phase) with each participant's limbs during running. This condition was called "Asymmetrical." In the control condition, each participant ran while facing a static image of his body. The static image was created by digitally editing the video footage taken before experimental testing. A frozen image of each participant was recorded for a 25-min period and was played back on the projection screen during the run. The order of conditions was counterbalanced across days. With six potential orders, two conditions (asymmetrical-symmetrical-control and control-symmetrical-asymmetrical) were not repeated across participants. In the interest of visual consistency, runners wore the same clothes for each session.

Back to Top | Article Outline
Data Analysis

Metabolic, kinematic, and psychological indices of performance "stability" were obtained as detailed below. All vision conditions were analyzed using preplanned orthogonal contrasts where: first, the static control condition was compared to the two dynamic conditions, and second, the two dynamic conditions were compared to each other. Speed (i.e., slow and fast) and time were repeated-measures factors in the majority of the analyses. Due to the differential durations of each speed condition and variations across measures in terms of the number of data points collected, the time variable and the number of levels of this variable differed across measures as detailed below. When data points were missing for participants (due to experimenter or equipment error), the adjusted n value is reported for that measure, otherwise n = 10. Significant interactions involving vision were further analyzed using Tukey post hoc procedures. Significance was set at P < 0.05. Partial eta squared (ηp2) values are reported as measures of effect size.

Back to Top | Article Outline
Metabolic Measures
Oxygen consumption.

Samples of oxygen consumption (mL·kg·min−1) were measured during performance at 60% and 80% V˙O2max for each of the three experimental runs. One-minute samples were taken during the 3rd, 6th, 9th, and 12th min of testing at the slow speed, yielding four data points. These means were compared in a three condition × four period mixed ANOVA. At the fast speed, 1-min samples were recorded during the 17th and the 19th min, yielding two means. To enable statistical comparisons across speed, only two means corresponding to the last two measures of the slow condition were compared to the two means at fast condition in a three-way mixed ANOVA.

Back to Top | Article Outline
Heart rate.

HR (beats·min−1) was recorded continuously. These data were averaged over 5-min periods, yielding four mean data points (three slow and one fast). Two analyses were conducted on these data: first comparing across the two speeds (last block at the slow speed) and then comparing across the three blocks in the slow speed only.

Back to Top | Article Outline
Rating of perceived exertion.

Participants indicated their perceived level of physical exertion using Borg's (6) scale for rating of perceived exertion (RPE) at 3-min intervals after the start of each run. Participants called out the numeric rating they perceived as most representative of their overall (central and peripheral) level of physical exertion. The scale ranged from 6 to 20 (where 6 = no exertion at all, and 20 = maximal exertion). The 3-min scheduling provided six data points across the run (four at the slow speed and two at the fast speed). For statistical analyses, comparisons were made across four means corresponding to the last two blocks of the slow condition and the two means of the fast condition. A second analysis was conducted, comparing across the four means for the slow condition only.

Back to Top | Article Outline
Kinematic Measures
SD of footfall variability.

Kinematic data were collected for 5-s capture periods at intervals of 2.5 min from the start of the run during performance at 60% and 80% V˙O2max. This analysis yielded six data points at the slow speed. Kinematics were captured every 1.5 min from the start of the shorter duration fast condition, yielding three data points. After data tracking, data were exported using a Qualisys Track Manager (QTM) macroprogram named PCReflex into a Microsoft-based application called Excelmr. These data were then filtered at 7 Hz using a dual-order Butterworth filter. Lateral and vertical displacement of the toe marker was calculated, as was the velocity of this marker in the vertical plane. Peak negative velocity was used to locate the point of toe strike within the gait cycle. Footfall variability was derived from the standard deviation of lateral displacement of the toe marker in the z axis between full gait cycles across each capture period. The SD of lateral displacement of the toe marker between full gait cycles across each capture period provided a measure of footfall variability. As in other analyses, variability was compared across the slow and fast speeds (on the basis of the mean variability of the final three time points in each condition). A further analysis compared variability at the slow speed only as a function of time. SD on the basis of the first, second, and third sample periods (at 2.5, 5, and 7.5 min, respectively) were compared to those from the fourth, fifth, and sixth samples (at 10, 12.5, and 15 min, respectively).

Back to Top | Article Outline
Footfall frequency.

A common point in the gait cycle, as determined by the lowest value in the gait cycle for the right toe's vertical (z) velocity was identified. Cycle frequency was then obtained by dividing the 5-s capture period by the number of completed step cycles within that period. Similar analyses to those carried out for footfall variability were conducted.

Back to Top | Article Outline
Psychological Measures
Reaction time.

Mean RT were calculated for approximately 5-min intervals resulting in three data points for the slow speed and one mean data point for the fast condition. Two analyses were performed on these data as detailed for the other measures above.

Back to Top | Article Outline
Rating scale for mental effort.

At 3-min intervals, participants indicated their perceived level of mental effort using the standardized rating scale for mental effort (RSME) (33). Participants were presented with a scale and were asked to call out the numeric rating they perceived as most representative of the mental workload they were experiencing (0-150, where 0 = absolutely no effort, and 150 = extreme effort). Measurements were scheduled to take place at 3-min intervals from the beginning of each trial, resulting in four mean values for the slow condition and two for the fast condition. The means of the first and last samples at the slow speed were compared to the two means at the fast speed in a three-way ANOVA with repeated measures for speed and time block. A second analysis was performed on the slow condition only comparing the four means.

Back to Top | Article Outline


Metabolic-Dependent Variables
Oxygen consumption.

Mean V˙O2 data across the slow and fast conditions are presented in Figure 1. Preplanned contrasts comparing the dynamic conditions to the static condition, F(1,9) = 7.75, P < 0.05, ηp2 = 0.46, and the asymmetrical to the symmetrical condition, F(1,9) = 5.78, P < 0.05, ηp2 = 0.39, were both statistically significant. The dynamic conditions were generally more economical than the static condition, but this was primarily due to the lower V˙O2 consumption for the symmetrical versus the asymmetrical condition (mean = 37.46, SD = 6.49 vs mean = 39.02, SD = 6.88 mL·kg·min−1, respectively). Main effects for both speed, F(1,9) = 170.91, P < 0.001, ηp2 = 0.95, and time, F(1,9) = 36.88, P < 0.001, ηp2 = 0.80, were observed. The slow condition was more economical than the fast, and the earlier period showed lower V˙O2 consumption than the later one. The only interaction effect was for visual condition and speed, F(1,9) = 5.34, P < 0.05, ηp2 = 0.37, when comparing the static condition to the two dynamic conditions. The difference between the conditions was increased under faster speeds. Although similar trends were observed for visual condition at the slow speed, no significant differences were observed.

Back to Top | Article Outline
Heart rate.

The mean HR data (n = 8) showed that the lowest values were observed under the symmetrical visual coupling condition (Table 1). However, there were no significant differences across the vision conditions when comparing across the two speeds (asymmetrical vs symmetrical, F(1,7) = 2.43, P = 0.16, ηp2 = 0.26), or in the slow condition only (asymmetrical vs symmetrical, F(1,7) = 2.39, P = 0.17, ηp2 = 0.25). The time effect was significant at the slow speeds, showing a significant linear trend for increased HR as a function of time, F(1,7) =114.25, P < 0.001, ηp2 = 0.94.

Back to Top | Article Outline
Rating of perceived exertion.

There were no significant effects involving vision; however, as with the other physiological measures, there was a trend for the symmetrical condition to be perceived as the least effortful, particularly under fast speeds (Table 2). The asymmetrical condition was perceived as the most effortful. Main effects for time for both the slow condition and when comparing across two subsequent blocks for the slow and fast conditions provided evidence of the typical effect of fatigue on level of perceived exertion, F(3,27) = 24.22, P < 0.001, ηp2 = 0.73 and F(1,9) = 25.46, P < 0.001, ηp2 = 0.74, respectively. There was also a significant interaction across speed and time, F(1,9) = 5.13, P = 0.05, ηp2 = 0.36, showing that the perception of fatigue was more affected by time at the faster speed.

Back to Top | Article Outline
Kinematic Measures
Mean SD of lateral footfall variability.

At the slow speed, there were no significant effects pertaining to vision condition (n = 6). However, when vision condition was compared across the two speeds, a significant speed × vision interaction effect was observed. This effect was due to the difference in footfall variability between the dynamic and static conditions, F(1,5) = 10.20, P < 0.05, ηp2 = 0.67. This interaction effect is illustrated in Figure 2. At the fast speed, the dynamic conditions (in particular, the asymmetrical condition) were performed with more variability than the static, control condition.

Back to Top | Article Outline
Gait cycle frequency.

Analysis of gait cycle frequency (n = 6) yielded a main effect of speed when comparing the last block of the slow speed to the fast speed, F(1,5) = 26.55, P < 0.01, ηp2 = 0.84. There were no significant effects involving vision (Table 3).

Back to Top | Article Outline
Psychological Measures
Reaction time.

RT was slower under both dynamic conditions in comparison to the static condition, when comparing the RT from the last block of the slow speed to the fast speed, F(1,9) = 6.80, P < 0.05, ηp2 = 0.43. These data are illustrated in Figure 3. Vision did not interact with speed. For the slow speed condition, the difference between the vision conditions was not significant, despite a similar trend as detailed above, F(1,9) = 2.64, P = 0.14, ηp2 = 0.23.

Back to Top | Article Outline
Rating scale for mental effort.

The condition effects were not significant, but there was a significant three-way interaction across vision, speed, and time (n = 9). The asymmetrical condition was rated higher for mental effort compared to the symmetric, dynamic condition (Fig. 4). This effect was significant at the fast speeds and in the last block, F(1,8) = 10.85, P = 0.01, ηp2 = 0.58. Significant effects of speed, F(1,8) = 72.06, P < 0.001, ηp2 = 0.90, and time, F(1,8) = 62.53 P < 0.001, ηp2 = 0.89, confirmed the prediction that the faster speed would require more mental effort than the slower speed and that perceptions of mental effort would be affected by time and interference (i.e., fatigue). A significant time effect was observed when comparing across the four time blocks at the slow speed, F(3,24) = 87.72, P < 0.001, ηp2 = 0.92.

Back to Top | Article Outline


We examined whether treadmill running stability is mediated by incidental visual couplings with perceptual information. Stability was determined by examining various systems assumed to be reflective of coordination efficiency (i.e., metabolic, kinematic, and cognitive). In line with earlier research involving between-person coordination (e.g., (19-21,26)), it was predicted that incidental visual coupling with concurrent feedback of limb movements phased in visual symmetry with the body would afford a greater level of stability than asymmetrical limb movements or a static visual image. This stability was expected to be indexed through increased metabolic economy, reduced cognitive load, and decreased kinematic variability. Increases in metabolic demands, as a function of increases in both speed and time, were expected to accentuate any effects of the visual manipulations.

The vision conditions affected various measures of effort and stability of treadmill locomotion. Although the effects were not reliably observed across all dependent measures, a trend was evident across variables for the asymmetrical, dynamic coupling condition to be perceived and measured as the most effortful and variable in comparison to the static condition and the symmetrical, dynamic condition. For measures of V˙O2 consumption (mL·kg·min−1), the symmetrical, dynamic condition resulted in a more economical performance than the control condition, supporting our original predictions. However, this effect was not observed for HR or for RPE. Moreover, the dynamic conditions generally resulted in slower RT and increased variability in footfall when compared to the static condition, which would not unequivocally support a "stability-enhancing" effect for unintentional, in-phase coupling.

Measures of V˙O2 have been shown to differentiate across relative phase patterns in a bimanual task requiring manipulation of two independent bicycle ergometers (11,23). However, the most economic patterns were not always the most kinematically stable. Lay et al. (11) concluded that modes lower in metabolic energy may compete with, rather than predict, dynamically stable modes. They found the antiphase coordination more efficient in terms of HR and oxygen consumption, whereas the in-phase coordination had the most stable pattern in terms of variability in relative phase (i.e., kinematics). Results from the current experiment extend these conclusions. Although the participants in our study were always performing antiphase movements, the visual-spatial constraints additionally affected the efficiency of this movement pattern, particularly at higher frequencies.

Incidental or unavoidable coordination with symmetrical motion seems to afford a more economical running style when compared with asymmetrical and static conditions. In comparison to the asymmetrical condition, this type of symmetrical mirror coupling is also reported to feel less mentally demanding, as evidenced in the RSME data. In postexperimental debriefing, participants expressed "discomfort" experienced under the "disorientating" asymmetrical conditions, in comparison to the "less effortful" symmetric and static conditions. One explanation for these findings might be in terms of a more psychologically relaxed running style afforded by the more familiar symmetrical coupling. Indeed, Caird et al. (7) have shown that participants who undertook psychological relaxation training before treadmill running were able to lower their HR by 2.5% and decrease oxygen consumption by 7.3%, although fitness levels did not change. However, as discussed later, the symmetrical coupling did not result in a more kinematically stable movement, which was initially expected to be indicative of a relaxed running style.

The above findings concur with research showing that visual scenes characterized by stimuli moving in the same direction (i.e., in-phase or isodirectional) are perceptually more salient than other motion structures (e.g., (4,5,17)). According to the "isodirectional principle," visual perception of limbs moving in the same direction enhances the stability of the in-phase pattern, whereas perception of limbs moving in the antiphase pattern reduces the stability of relative phase patterns (5). Furthermore, perceived variability in 180° mean relative phase increases as a function of movement frequency, while perception of 0° mean relative phase does not change as frequency increases (31). Data generated in the current study warrant further investigation into the effects of fatigue on perceived variability.

Despite the metabolic differences between conditions, the kinematic data did not show evidence of enhanced stability when participants were symmetrically coupled. However, control conditions did seem more stable. One explanation for this finding may lie in the RT data. Responding to LED stimuli in the dynamic displays was arguably a more difficult task than under static conditions because the moving images would make stimuli detection more difficult. The additional demands associated with the RT task for the dynamic displays would make any clear comparisons with the control condition difficult. In fact, because of the secondary task effects associated with the RT task, it is an even more telling result to find V˙O2 advantages for the dynamic conditions. However, the present study did not examine performance in the absence of a whole-body visual representation. This precludes an assumption that symmetrical coupling leads to more efficient running than observed under normal (no information) conditions. Instead, we considered the static image to be a more important control condition. This was based on the evidence in psychology literature showing the detrimental effects on performance created by an inward attentional focus when seeing an image of oneself performing (e.g., (12,30)). Therefore, the control in the present study ensured that self-attention was not a potential confound.

In conclusion, we have shown perceptual-spatial manipulations can affect various indices of stability during treadmill running. Although predicted effects were not reliably found across all measures, there is evidence to support the conclusion that treadmill running is more efficient when viewing mirrorlike images of symmetrical limb movements over asymmetrical and static images. Our findings highlight the complex interactions between motor and perceptual constraints. However, further research is required to better understand the energetic costs associated with incidental visual couplings and what might be the "preferred" state. Possibly, advanced understanding of these issues could inform the practice of motor development across varied domains. Our initial data indicate that treadmill running in front of a mirror, as is common in a gym, induces an economic running state. For some people (such as the elderly), this might be desirable, but for others, running in front of a mirror might not deliver the physiological demands required. On the other hand, runners may find it beneficial to use these perceptual affordances to conserve energy while running in synchrony behind a competitor in a race.

The results of the present study do not constitute endorsement by ACSM.

Back to Top | Article Outline


1. Almasbakk B, Whiting HTA, Helgerud J. The efficient learner. Biol Cybern. 2001;84(2):75-83.
2. Bird S, Davison R, editors. Guidelines for the Physiological Testing of Athletes. 3rd edition. Leeds (UK): British Association of Sport and Exercise Sciences; 1997. pp. 108-11.
3. Bingham GP, Schmidt RC, Turvey MT, Rosenblum LD. Task dynamics and resources dynamics in the assembly of coordinated rhythmic activity. J Exp Psychol Human Percept Perform. 1991;17(2):359-81.
4. Bingham GP, Zaal FT, Shull JA, Collins DR. The effect of frequency on the visual perception of relative phase and phase variability of two oscillating objects. Exp Brain Res. 2001;136(4):543-52.
5. Bogaerts H, Buekers MJ, Zaal FT, Swinnen SP. When visuo-motor incongruence aids motor performance: the effect of perceiving motion structures during transformed visual feedback on bimanual coordination. Behav Brain Res. 2003;138:45-57.
6. Borg G. Borg's range model and scales. Int J Sport Psychol. 2001;32:110-26.
7. Caird SJ, McKenzie AD, Sleivert GG. Biofeedback and relaxation techniques improve running economy in sub-elite long distance runners. Med Sci Sports Exerc. 1999;31(5):717-22.
8. Fitch HL, Tuller B, Turvey MT. The Bernstein perspective: 3. Tuning coordinative structures with special reference to perception. In: Kelso JAS, editor. Human Motor Behaviour: An Introduction. Hillsdale (NJ): Lawrence Erlbaum Associates; 1982. p. 271-87.
9. Haken H, Kelso JAS, Bunz H. A theoretical model of phase transitions in human hand movements. Biol Cybern. 1985;51(5):347-56.
10. Kelso JAS. Dynamic Patterns: The Self-Organisation of Brain and Behaviour. Cambridge (MA): MIT Press; 1995. p. 46-52.
11. Lay BS, Sparrow WA, O'Dwyer NJ. The metabolic and cognitive energy costs of stabilizing a high energy interlimb coordination task. Hum Mov Sci. 2005;24:833-48.
12. Martin Ginis KA, Jung ME, Gauvin L. To see or not to see: the effects of exercising in mirrored environments on sedentary women's feeling states and self-efficacy. Health Psychol. 2003;22:354-61.
13. McArdle WD, Katch FI, Katch VL. Essentials of Exercise Physiology. 3rd edition. Baltimore (MD): Lippincott Williams & Wilkins; 2006. p. 448-53.
14. Mechsner F. A psychological approach to human voluntary movement. J Mot Behav. 2004;36(4):355-70.
15. Mechsner F, Knoblich G. Do muscles matter for coordinated action? J Exp Psychol Human Percept Perform. 2004;30(30):490-503.
16. Mechsner F, Kerzel D, Knoblich G, Prinz W. Perceptual basis of bimanual coordination. Nature. 2001;414:69-73.
17. Salesse R, Temprado JJ, Swinnen SP. Interaction of neuro-muscular, spatial and visual constraints on hand-foot coordination dynamics. Hum Mov Sci. 2005;24:66-80.
18. Salter JE, Wishart LR, Lee TD, Simon D. Perceptual and motor contributions to bimanual coordination. Neurosci Lett. 2004;363(2):102-7.
19. Schmidt RC, O'Brien B. Evaluating the dynamics of unintended interpersonal coordination. Ecol Psychol. 1997;9(3):189-206.
20. Schmidt RC, Carrello C, Turvey MT. Phase transitions and critical fluctuations in the visual coordination of rhythmic movements between people. J Exp Psychol Human Percept Perform. 1990;6:227-47.
21. Schmidt RC, Bienvenu M, Fitzpatrick PA, Amazeen PG. A comparison of intra- and interpersonal limb coordination: coordination breakdowns and coupling strength. J Exp Psychol Human Percept Perform. 1998;24(3):884-900.
22. Sparrow WA, Irizarry-Lopez VM. Mechanical efficiency and metabolic cost as measures of learning a novel gross motor task. J Mot Behav. 1987;19:240-264.
23. Sparrow WA, Lay BS, O'Dwyer NJ. Metabolic and attentional energy costs of interlimb coordination. J Mot Behav. 2007;39(4):259-75.
24. Sparrow WA, Newell KM. Metabolic energy expenditure and the regulation of movement economy. Psychon Bull Rev. 1998;5:173-96.
25. Swinnen SP, Wenderworth N. Two hands, one brain: cognitive neuroscience of bimanual skill. Trends Cogn Sci. 2004;8:18-25.
26. Temprado JJ, Laurent M. Attentional load associated with performing and stabilising a between-persons coordination of rhythmic limb movements. Acta Psychol. 2004;115:1-16.
27. Temprado JJ, Zanone P-G, Monno A, Laurent M. Attentional load associated with performing and stabilising preferred bimanual patterns. J Exp Psychol Human Percept Perform. 1999;25:1579-94.
28. Welsh TN, Almeida QJ, Lee TD. The effect of postural stability and spatial orientation of the upper limbs on interlimb coordination. Exp Brain Res. 2005;161(3):267-75.
29. Wilson AD, Collins DR, Bingham GP. Perceptual coupling in rhythmic movement coordination: stable perception leads to stable action. Exp Brain Res. 2005;164(4):517-28.
30. Wulf G. Attention and Motor Skill Learning. Champaign (IL): Human Kinetics; 2007. p. 107-19.
31. Zaal TJM, Bingham GP, Schmidt RC. Visual perception of mean relative phase and phase variability. J Exp Psychol Human Percept Perform. 2000;26(3):1209-20.
32. Zanone P-G, Monno A, Temprado JJ, Laurent M. Shared dynamics of attentional cost and pattern stability. Hum Mov Sci. 2001;20:765-89.
33. Zijlstra R. Efficiency in Work Behaviour. A Design Approach for Modern Tools. Delft, The Netherlands: Delft University Press; 1993. p. 82-3.


©2008The American College of Sports Medicine