SEARCH

SEARCH BY CITATION

Keywords:

  • human locomotion;
  • visual context;
  • locomotor aftereffect;
  • treadmill;
  • motor adaptation

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

A firm linkage exists between a motor command and its expected feedback. When we are exposed to a conflict between expected and actual feedback in a new context, we form a new linkage between action and perception, which may be further strengthened by prolonged experience. In this paper, we attempt to identify whether the linkage between treadmill locomotion and visual processing in relation to optic flow is strengthened in experienced users of treadmills. Yabe and Taga (2008) showed that ambiguous apparent motions are perceived to be moving downward more frequently when the stimuli are shown in front of the observers' feet on a treadmill when walking compared with when standing. Here, their experimental data was reanalyzed in relation to the experience of using the treadmill. The result revealed that habitual treadmill exercise reduced the difference in perceived direction of visual motion between the walking and standing conditions. It should be noted that the treadmill users showed perceptual “downward” bias for both the standing and walking conditions. The results suggest that treadmill users tend to activate the habitual linkage between treadmill locomotion and perception of optic ground flow even when they are just standing on a treadmill.

Recently, growing evidence has confirmed that how people move may affect how they see (Ishimura & Shimojo, 1994; Maruya, Yang, & Blake, 2007; Wexler, Panerai, Lamouret, & Droulez, 2001; Wohlschläger, 2000; Yabe & Taga, 2008). These studies show that motor planning or execution plays an important role in resolving conflicts between incompatible visual perceptions. For example, using two-alternative forced-choice (2-AFC) tasks, Yabe and Taga (2008) reported “treadmill capture”, in which ambiguous apparent-motion stimuli of horizontal gratings placed in front of participants' feet were perceived to move downward more frequently during walking on a treadmill than during standing on it. There may be a tight linkage between forward walking and backward (i.e. downward in front of the feet) optic flow, which may serve as a constraint to reduce the degree of freedom in spatial vision during self-motion. A schematic of this model is shown in Figure 1 (left).

image

Figure 1. Forward locomotion in real life (left) usually produces backward optic flow. The backward flow implied from prior experience can serve as a constraint to disambiguate the direction of motion in the visual stimuli. We hypothesized that the participants with prolonged treadmill exercise experience (right) should adapt to the absence of optic flow during locomotion on a treadmill and thereby should lose the effect of “downward” bias of the perceived direction of motion.

Download figure to PowerPoint

Not only how a person moves but also how he/she had moved can affect visual perception. A linkage between a motor command and its sensory feedback can be learned during sensorimotor experience. Studies of letter processing have shown that writing experience alters the perception of apparent motion (Tse & Cavanagh, 2000) and the motor system is activated during the visual processing of letters (James & Atwood, 2009). How motor experience affects visual processing was investigated using well-controlled experiments in a recent study on motor learning, in which Brown, Wilson, Goodale, and Gribble (2007) showed that motor learning of a new environmental force promotes visual-motion processing. In their study, participants were asked to intercept a target moving in the rightward force field, after training to make a reaching movement in either the null, leftward, or rightward force field. The responses of the rightward force field group were enhanced, while those of the leftward force field group deteriorated. By eliminating the opportunity for participants to move their hand in the force field, the force direction had no influence on the stable hand, although the participants recognized and remembered the force adequately. Brown et al. (2007) proposed that once force information has been learned and adapted by the motor system, the visual system can use the information to predict the visual motion.

It is well known that brief treadmill running induces instant motor aftereffects and the phenomenon has been associated with recalibration of sensorimotor linkages. Anstis (1995) reported that participants who attempted to jog in place on solid ground jogged forward after 60 s of treadmill running. The aftereffect was greatest for zero delay, that is, immediately after running. When they started jogging in place after standing still, the forward motion dissipated over the course of 1–2 min. Anstis (1995)'s methods were improved by Durgin, Pelah, Fox, Lewis, Kane, and Walley (2005), who observed inadvertent forward motion without deceleration over 20 s with zero delay. They also performed experiments using a treadmill with an adaptation period of 20 s to 5 min and showed aftereffects with zero delay. Anstis (1995) also showed that one-legged hopping on the treadmill did not produce such aftereffects in the other leg and ruled out the effect of visual adaptation; he concluded that the aftereffects reveal recalibration to match motor output and feedback for the backward motion of the treadmill belt. In contrast, a possible visual adaptation during hopping was investigated by Durgin, Fox, and Hoon Kim (2003). Their interpretation of Anstis (1995)'s experiments was that there would have been no conflict because there was neither flow nor hop on treadmills for the nonadapted leg. Pelah and Barlow (1996) showed that participants who were instructed to maintain a constant “visual” speed while repeatedly walking a 5-m course after 20 min of treadmill running were seen to accelerate their pace. It was argued that the running-in-place aftereffect represents a visuomotor adaptation that takes place in the absence of normal optic flow (Durgin & Pelah, 1999). In most of the previous studies, after stepping off the treadmill, visuomotor expectancies that have been linked to motor commands of treadmill locomotion are thought to be under de-adaptation pressure and recalibrated again while standing or walking on solid ground. The aftereffect of brief treadmill running declines over the course of several minutes. However, few studies have described whether information about the moving surface is stored if the participant repetitively experiences treadmill running. Pelah and Barlow (1996) mentioned that long-term experience of exercise reduced the aftereffect, even though their tasks were only of the order of minutes.

Here, we examine whether long-term treadmill exercise causes a user to store the specific linkage between motion and perception, and to thereby change the magnitude of treadmill capture (Yabe & Taga, 2008), which is presumably based on a tight linkage between normal locomotion on the ground surface and downward optic flow. On the basis of the abovementioned studies on treadmill illusions, treadmill running or walking constructs a new linkage between locomotion on a treadmill and the absence of optic flow. We therefore hypothesize that if the information on the motor-perceptual linkage during each exercise is stored, the participants who are prolonged treadmill users should perceive “downward” visual motion fewer times than inexperienced participants. A schematic of this hypothesis is shown in Figure 1 (right). However, even for the experienced participants, the duration of exercise on treadmills is far shorter than the duration spent on solid ground. It is an open question whether the exercise on treadmills can change the treadmill capture, which may be based on the daily experience in life.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

We derived data from Experiment 2 of Yabe and Taga (2008) and reanalyzed it in terms of the participants' experience in treadmill exercise. In this section, we present an overview of the participants, stimuli, and experimental design (for details see Yabe & Taga, 2008). The exercise history was obtained by conducting a newly developed questionnaire survey.

Participants

Twenty-one healthy individuals participated in the experiment. They were all naïve to the purpose of the study. Each participant had normal or corrected-to-normal vision. Informed consent was obtained from all participants. Data from one participant was excluded from the analysis due to persistent unidirectional judgment during the sessions (in 19 of 22 trials). In addition, one female participant did not provide her history of treadmill exercise, and was thus excluded from the analysis. We analyzed data from 7 men and 12 women, 20–33 years old (mean age, 24 years) and 150–170 cm in height (mean height, 163 cm).

Stimuli

The stimuli were generated on a Windows computer and presented on a 15-in. LCD monitor (EIZO FLexScan L355). The refresh rate of the display was 75 Hz, and the resolution was 1024 × 768 pixels. As described in Figure 1a of Yabe and Taga (2008), the stimulus display was a horizontal grating pattern drawn with four-cycle sinusoidal luminance presented within a square window of side 14.85 cm, mounted on a black background. The visual angle and luminance varied depending on the stationary positions on the treadmill and the height of the participants. The visual angle was approximately 5.9–7.0 deg. The maximum value of the luminance was approximately 20–25 cd/m2 and the minimum was approximately 1.8–1.9 cd/m2; the luminance contrast was between 84% and 92%. The stimulus image was repainted ten times per second. At the time of repainting, the grating was shifted by a constant distance. The magnitude of the shift affected the perceived direction of the gratings. When the shift was equal to half the width of one cycle of the gratings (1.86 cm; approximately 0.8 deg when the visual angle of the stimulus window was the mean, 6.4 deg), the spatial phase difference became 180 deg. Thus, a counterphase grating (i.e. a truly directionally ambiguous grating) was presented with the temporal frequency of 5 Hz (Figure 1b, middle, Yabe & Taga, 2008). If the perceived direction of motion was unidirectional, its apparent speed was approximately 8.0 deg/s. With decreasing spatial phase difference from 180 deg, the apparent motion direction was biased upward (Figure 1b, top, Yabe & Taga, 2008). In contrast, with increasing spatial phase difference, the direction was biased downward (Figure 1b, bottom, Yabe & Taga, 2008). In either case, the velocity of the apparent motion slows down as the spatial phase difference recedes from 180 deg and approaches 0 deg or 360 deg. The minimum absolute speed was 6.0 deg/s, which was implemented by the shift of 135 deg or 225 deg, given that an element of the gratings in a frame is regarded as fused with the closest one in the next (Figure 1b, Yabe & Taga, 2008). We presented no fixation point on the stimulus in order to not give a reference point for judging the motion direction.

Experimental design

Participants were either standing or walking on a treadmill (Nihon Kohden Aeromill STM-1420; Figure 1a, Yabe & Taga, 2008). The treadmill speed was set at 91.66 cm/s. Under the walking condition, participants were told to use the handrails only if absolutely necessary for safety purposes. No participants actually used the handrails during the experiment. Apparent motion stimuli of a horizontal sine wave grating with shifts of the phase were displayed on a 15-in. LCD monitor in front of the participants' feet. In each trial, the shift of the sine wave grating was randomly chosen from 11 variations between 135 deg and 225 deg, with a counterphase grating that shifts 180 deg, five upward ones that shift less than 180 deg, and five downward ones that shift more than 180 deg. We darkened the room to make the participants' feet and the texture of the treadmill belt as invisible as possible. The belt was colored solid dark green and its luminance was 0.005 cd/m2 at most. It had an indented patterned surface with furrows 0.6 cm apart, which corresponds to approximately 0.2 deg for a 160-cm-tall participant. It is unlikely that participants could see the belt motion with peripheral viewing (Anderson, Mullen, & Hess, 1991). Each participant determined the position on the treadmill at which they would stand or walk throughout the experiment so that they could watch the display clearly. Participants were told that they would be shown displays of a moving horizontal grating, and that “the gratings move upward or downward.” Participants were also told to view the entire stimuli and not to fixate on their fringes. In each trial, the grating pattern was presented for 3 s, followed by a 10-s presentation of a central white cross on a black background as a fixation target for gazing. As soon as the display switched from the stimulus to the screen of the fixation cross, the participants fixated on the cross continuously and made a forced-choice judgment about whether the motion direction of the gratings presented just prior moved downward or upward. All responses were given orally and entered into a computer by an experimenter without any feedback. Each observer took part in one practice session and five experimental sessions. One experimental session consisted of a standing condition block and a walking condition block, which were performed in a counterbalanced ABBA order. Each block consisted of 22 trials and represented 11 distinct displays shown twice in random order. The stimuli were sine-wave gratings whose phase shift was randomly chosen from 11 varieties with a counter-phase grating, and five upward and five downward ones between 122.4 deg and 237.6 deg. The total amount of time required for all sessions was approximately 90 min. The participants stood on a board bridged over the running treadmill during the standing condition and the treadmill was kept at 91.66 cm/s, that is, at the same speed as during the walking condition.

Prior to the start of the experimental tasks, the participants answered a question assessing whether or not they had any experience in treadmill exercise. The participants were categorized as non-treadmill runner (nTR) if they had experienced treadmill walking three times or less. This limit was chosen to include the participants who have happened to use a treadmill but have not used one for exercise habitually. Participants experienced in habitual treadmill exercise were categorized as treadmill runner (TR). The two groups of nTR and TR differ in their exercise history; the minimum number of exercise sessions experienced by the TR participants is 16 times (see Table 2).

Table 2.  Reported experience of treadmill exercise
Participant no.Exercise period (months)Frequency (per month)Running time (h)Break (months)Total no. of exercise sessionsTotal duration (h)
  1. Note. Exercise period, Frequency, Running time, and Break were obtained from Questions 1, 2, 3, and 4 of Appendix I, respectively. The total number of exercise sessions times was obtained by multiplying Exercise period with Frequency. The total duration was estimated by multiplying the Total number of exercise sessions and Running time.

02280.536168
03124204896
04180.750.75613.510.13
052480.33019263.36
1018122.2520216486
132120.3378247.92
173121123636

Detailed history of treadmill exercise

We conducted a questionnaire survey via email (Appendix) only among TR participants to collect details on their experience of treadmill exercise.

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

The demographics of the study population are given in Table 1. There were no differences between the TR and nTR groups in terms of age (two tailed t-test, p = .79), height (two tailed t-test, p = .34), sex (χ2 = 0.003, 1; p = .96), and vision (χ2 = 2.77, 1; p = .1).

Table 1.  Details of participants
CharacteristicsNon-treadmill runnersTreadmill runners
N118
Age (mean ± SD; range), years24.3 ± 3.80 (20–33)23.9 ± 1.96 (22–28)
Sex (number of women)75
Height (mean ± SD; range), cm162.3 ± 7.48 (150–170)165.13 ± 4.09 (158–170)
Vision (number of unaided participants)72

Comparison between nTR and TR

We examined the probability of trials in which the grating appeared to move “downward” (averaged for all participants) as a function of the physical shift of the grating. Figure 2a,b shows the group-averaged probability functions under the walking and standing conditions.

image

Figure 2. The graphs on the left show the probability of the “downward” responses as a function of the shift of the sine wave by mean and SE, averaged across participants who were either (a) non-treadmill runners (nTR) or (b) treadmill runners (TR). The distance of the shift of the sine wave is the physical bias of the stimulus display represented in spatial phase difference. The displacement at 180 deg makes a counterphase grating. (inline image) The probability under the walking condition. (inline image) The probability under the standing condition. The graphs at the right show the point of subjective equality (PSE) under the standing condition and that under the walking condition for (c) nTR and (d) TR participants.

Download figure to PowerPoint

When the physical shift was 180 deg (counterphase gratings), the downward probability was .59 (SD = 0.21) under the walking condition and .35 (SD = 0.18) under the standing condition for nTR. The difference between the walking and standing conditions for nTR was significant, two-tailed paired t-test, t(10) = 3.49, p < .01. The downward probability was .65 (SD = 0.19) under the walking condition and .54 (SD = 0.24) under the standing condition for TR. We found no significant difference between the walking and standing conditions for TR, t(7) = 1.84, p = .1.

The probability of the “downward” responses was calculated and fitted with a logistic function for each participant. From the fitted functions, the point of subjective equality (PSE, i.e. the probability of the “downward” responses equal to .5) was estimated for each participant (Figure 2c,d). A statistical analysis of nTR showed that the PSE, (176.3°, SD = 9.23) under the walking condition was significantly different from that (183.0 deg, SD = 7.15) under the standing condition, two-tailed paired t-test, t(10) = 3.31, p < .01. For TR, the PSE (176.53 deg, SD = 4.68) under the walking condition was not significantly different from that (178.12 deg, SD = 6.73) under the standing condition, two-tailed paired t-test, t(7) = 0.73, p = .49.

Correlation between amount of treadmill capture and history of using the treadmill

The responses to the questionnaire by the TR participants are summarized in Table 2. We could not contact a male TR participant, who was excluded from further analysis. The total number of exercise was obtained by multiplying the exercise period and frequency for each TR participant. By multiplying the total number of exercise sessions and the running time, the total duration of exercise was estimated.

We plotted the PSE difference between the walking and standing conditions against the total number of exercise sessions and the total duration of exercise for each TR participant (Figure 3a,b). There is a negative correlation between the total number of exercise sessions and the PSE difference, although it is not significant, r = −0.71, p = .07. There was no significant correlation between the total duration of exercise and the PSE difference, p = 0.462. The PSE of the walking condition (Figure 3c,d) did not show a significant correlation with either parameter. The PSE of the standing condition showed a negative correlation with the total number of exercise sessions, r = −0.85, p < .05, and with the total duration of exercise, r = −0.84, p < .05, as shown in Figure 3e,f.

image

Figure 3. Scattergram to visualize intersubject correlation between the point of subjective equality (PSE) statistics and the statistics of exercise history. Data from seven participants from the treadmill runners group are plotted, with one point corresponding to one subject. (a–b) Intersubject scattergram between the PSE difference and (a) total number of exercise sessions and (b) total duration of exercise. The solid line indicates linear regression: y = −0.04x + 3.303. (c–d) Intersubject scattergram between the PSE under the walking condition and (c) total number of exercise sessions and (d) total duration of exercise. (e–f) Intersubject scattergram between the PSE under the standing condition and (e) total number of exercise sessions and (f) total duration of exercise. The solid lines indicate linear regression: (e) y = −0.045x + 179.796 and (f) y = −0.022x + 178.550.

Download figure to PowerPoint

There was no significant effect of the length of the break from treadmill exercise (i.e. how many months had passed since the last exercise) on the PSE difference (p = .86) between the two conditions, the PSE of the standing condition (p = .261), and the PSE of the walking condition (p = .146), as shown in Figure 4.

image

Figure 4. Scattergram to visualize intersubject correlation between the point of subjective equality (PSE) statistics and break of exercise. Data from seven participants are plotted; one point corresponds to one subject. Intersubject scattergram (a) between the PSE difference and break, (b) between the PSE under the standing condition and break, and (c) between the PSE under the walking condition and break.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

In this study, we have hypothesized that habitual treadmill exercise modulates the linkage between motion and perception, resulting in influences on the perceived direction of ambiguous motion during walking or standing on a treadmill (treadmill capture). We obtained details of treadmill exercise history from participants to examine the effect of the amount of exercise on treadmill capture. The analysis indicated that those participants without experience of treadmill exercise perceived directionally ambiguous motions of shifting frames of sinusoidal horizontal gratings in front of their feet as moving downward more frequently while walking than while standing on a treadmill. In contrast, the participants who were experienced in treadmill exercise did not show the effect that the gratings are perceived downward more frequently. Although this supports the hypothesis that visual perception on a treadmill is affected by the prolonged experience of locomotion on treadmills, it does not support our hypothesis that long-term experience of exercise on treadmills reduces the perceptual downward bias of flow of motion under the walking condition. The data indicates that the experience of treadmill exercise did not affect the response while walking, but affected it while standing on a treadmill.

It should be noted that there are considerable differences between the instant adaptation of treadmill locomotion in the previous studies and the effect of repetitive experience on treadmill capture in the present study. In the case of instant aftereffects of treadmill locomotion, participants inadvertently walked forward (Anstis, 1995) or accelerated (Pelah & Barlow, 1996) on solid ground soon after stepping off the treadmill, as if the ground surface, rather than only the treadmill belt, were moving. The motor adaptation to the motion of treadmill belt leads to a recalibration of motor command; that is, as the participant starts walking, the ground surface appears to go backward in any context. In the case of long-term treadmill exercise, users are repeatedly exposed to the process of adaptation and washout. This may produce a context-specific recalibration, which indicates that there appears to be no optic flow while walking on a treadmill. The context-specific recalibration is likely not to be diminished for years after they quit the exercise. Thus, long and short adaptation may induce different degrees of context-specificity of recalibration.

Context-specificity has been examined in several previous studies of locomotor aftereffects. Pelah and Barlow (1996) mentioned that the aftereffect of treadmill running was most compelling when walking on the treadmill itself after it has stopped. More recently, the striking phenomenon in which we experience an odd sensation on a broken escalator was investigated (Bunday, Reynolds, Kaski, Rao, Salman, & Bronstein, 2006; Fukui, Kimura, Kadota, Shimojo, & Gomi, 2009; Reynolds & Bronstein, 2003). Bunday et al. (2006), using a linear motor-powered sled in their laboratory, showed that fast locomotor aftereffects can be generated after participants walked on a moving sled only one or two times. Reynolds and Bronstein (2007) suggested that their “laboratory aftereffect” may differ from the broken escalator effect with respect to the time course. There may be a very strong association between the perceptual cues linked to the movement of real escalators, which are reinforced on a daily basis. Fukui et al. (2009) used a real escalator to investigate the occurrence of the odd sensation qualitatively. Comparing the kinematic properties during stepping onto a stopped escalator with those while stepping onto a moving one and onto wooden stairs, they found that postural forward sway, rather than inadequate leg movement, is essential for the perception of the odd sensation. They indicated that the odd sensation represents two different components of motor control: a voluntary component and an automatic control. When we step onto a broken escalator, our legs are controlled rather voluntary while our posture is largely controlled in an automatic fashion. Postural sway represents the automatically triggered habitual motor program specific to the context of an escalator.

The present study revealed that the history of treadmill exercise affects the responses under the walking and standing conditions differently, which can be explained by separate mechanisms that may work under each condition (Figure 5). While people spend most of their lives on solid ground, treadmill runners experience a different context. Our findings showed that treadmill exercise did not affect the visual response under the walking condition. This indicates that the lifetime experience of locomotion on solid ground can keep producing the preexisting effect that treadmill walking facilitates unidirectional perception of the counterphase gratings in front of the observers' feet in both non-treadmill runners and treadmill runners. The important finding of the present study is that the effect of experience of treadmill exercise was prominent under the standing condition. The reason why such an effect was only observed for the participants who have experienced prolonged exercise on treadmills can be related to the fact that they probably acquired the linkage between their locomotion and visual perception of the backward flow of the moving belt of a treadmill and that standing still on a treadmill provides a novel and “odd” context for them, just like walking on a broken escalator.

image

Figure 5. Schematic explanation of the results. Even if the participant has many years of treadmill exercise experience, the duration of exercise on treadmills (black arrow drawn in the arrow of “Real Life”) cover only part of the life spent on solid ground. Under the walking condition, the backward flow experienced daily during locomotion can contribute to disambiguating the direction of movements in the visual stimuli for both treadmill runners (TRs) and non-treadmill runners. Under the standing condition, the perceptual “downward” bias was observed only for TRs, which indicates the influence of context-specific adaptation to the locomotion on a moving surface.

Download figure to PowerPoint

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix

Appendix

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Appendix
Questionnaire
  • 1
    For how long have you or had you been exercising with a treadmill (in months)?
  • 2
    How frequently do you exercise in a month?
  • 3
    What is the typical duration of an exercise session?
  • 4
    Was the experiment conducted during the abovementioned exercise period? If you had already stopped exercising, how long was the break until the experiment was conducted? That is, how many months have passed since the last exercise?