Motor Control, 2015, 19, 191 -206 http://dx.doi.org/10.1123/mc.2014-0010 © 2015 Human Kinetics, Inc.
original article
Perception of Self-Motion and Regulation of Walking Speed in Young-Old Adults Marie-Jasmine Lalonde-Parsi and Anouk Lamontagne McGill University Whether a reduced perception of self-motion contributes to poor walking speed adaptations in older adults is unknown. In this study, speed discrimination thresholds (perceptual task) and walking speed adaptations (walking task) were compared between young (19–27 years) and young-old individuals (63–74 years), and the relationship between the performance on the two tasks was examined. Participants were evaluated while viewing a virtual corridor in a helmet-mounted display. Speed discrimination thresholds were determined using a staircase procedure. Walking speed modulation was assessed on a self-paced treadmill while exposed to different self-motion speeds ranging from 0.25 to 2 times the participants’ comfortable speed. For each speed, participants were instructed to match the self-motion speed described by the moving corridor. On the walking task, participants displayed smaller walking speed errors at comfortable walking speeds compared with slower of faster speeds. The young-old adults presented larger speed discrimination thresholds (perceptual experiment) and larger walking speed errors (walking experiment) compared with young adults. Larger walking speed errors were associated with higher discrimination thresholds. The enhanced performance on the walking task at comfortable speed suggests that intersensory calibration processes are influenced by experience, hence optimized for frequently encountered conditions. The altered performance of the young-old adults on the perceptual and walking tasks, as well as the relationship observed between the two tasks, suggest that a poor perception of visual motion information may contribute to the poor walking speed adaptations that arise with aging. Keywords: aging, gait speed, locomotion, visual cues, virtual reality
Mobility impairments are prevalent among the elderly population, as evidenced by 8–19% of noninstitutionalized older adults having difficulty walking or requiring the assistance of another person or an assistive device to walk (Dawson, Hendershot, & Fulton, 1997; Leon & Lair, 1990). Older adults typically adopt slower comfortable walking speeds compared with young adults, with a rapid decline of 12–16% per decade being observed as from the critical age of 62 years (Himann, The authors are with the School of Physical & Occupational Therapy, McGill University, Montreal, Quebec. Address author correspondence to Anouk Lamontagne at [email protected]. 191
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
192 Lalonde-Parsi and Lamontagne
Cunningham, Rechnitzer, & Paterson, 1988). The capacity to reach fast, maximal walking speeds is even further altered (Bohannon, 1997; Cunningham, Rechnitzer, Pearce, & Donner, 1982; Himann et al., 1988; Murray, Kory, & Clarkson, 1969), which suggests a reduced ability to make the speed adjustments that are necessary to respond to contextual demands (e.g., speeding up to cross a street or catch a train). Beyond anthropometric characteristics, factors that are the most commonly suggested to contribute to slow walking speed in older adults include reduced lower extremity muscle strength and power, reduced maximal aerobic capacity, presence of health problems and pain (Bendall, Bassey, & Pearson, 1989; Bohannon, 1997; Buchner et al., 1996; Cunningham et al., 1982; Cuoco et al., 2004). While some reports suggest that an altered perception of visual stimuli may also contribute (Hassan, Lovie-Kitchin, & Woods, 2002; Kuyk & Elliott, 1999; Sakari et al., 2010), little is known about the influence of visual information perception on the control of walking speed in older adults. Successful ambulation through our environment requires the integration of sensory information from visual, vestibular, and proprioceptive inputs. In particular, controlling self-motion is contingent upon the information about our perceived speed of movement. We derive speed information from optic flow (OF), a pattern of motion generated at the retina by self-motion through an environment (Gibson, 2009), and use it to monitor and update our walking velocity. Previous studies have shown that healthy young and older adults, when instructed to maintain a comfortable walking speed while viewing changing OF speeds, respond similarly by slowing down in response to faster OFs and speeding up with slower OFs (Konczak, 1994; Prokop, Schubert, & Berger, 1997; Schubert, Prokop, Brocke, & Berger, 2005). After several minutes of walking, this out-of-phase modulation of walking speed attenuates, possibly due to a recalibration process between proprioceptive and visual information (Prokop et al., 1997). Such experimental paradigm, where participants are instructed to maintain their walking speed, leads to unintentional speed modulations that reflect the ability of the central nervous system (CNS) to respond to mismatching sensory cues (e.g., visual vs. body-based cues). As evidenced in a recent study in stroke survivors, however, such ability is poorly associated with the participants’ ability to achieve voluntary and appropriately scaled speed modifications while using OF information (Lamontagne, Fung, McFadyen, & Faubert, 2007). Whether older adults can use OF cues to make volitional walking speed adjustments remains unknown. A number of studies have observed alterations in motion detection accompanying normal aging, with deteriorations starting to appear in the young-old age range (65–74 years) (Bennett, Sekuler, & Sekuler, 2007; Gilmore, Wenk, Naylor, & Stuve, 1992; Trick & Silverman, 1991). Specifically, the ability to discriminate the speed of visual motion information, displayed as visual gratings or moving dots, is compromised with advanced age (Habak & Faubert, 2000; Norman, Ross, Hawkes, & Long, 2003; Raghuram, Lakshminarayanan, & Khanna, 2005; Snowden & Kavanagh, 2006). None of the existing studies, however, have examined speed discrimination abilities while using meaningful and ecological displays that would resemble the visual sceneries experienced in daily life. Moreover, whether a reduction in the ability to discriminate speed from OF cues in older age contributes in part to the locomotor alterations in this population is still unclear. MC Vol. 19, No. 3, 2015
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
Perception and Modulation of Walking Speed 193
The primary aim of this study was to examine and compare abilities of young and older adults to discriminate OF speeds and to make walking speed adjustments in response to changing OF speeds while immersed in a meaningful, ecological virtual environment (VE). A secondary aim was to investigate the relationship between speed discrimination and walking speed modulation abilities in the older adults. Because changes in visual-perceptual, balance and locomotor abilities are noticeable in persons in their sixties and seventies, and since these abilities further deteriorate as participants enter the ‘old’ age range (≥75 years) (Dobbs et al., 1993; P. Haibach, Slobounov, & Newell, 2009; P. S. Haibach, Slobounov, Slobounova, & Newell, 2007; Himann et al., 1988), this study specifically targeted ‘young-old’ adults between the ages of 60 and 74 years (Zizza, Ellison, & Wernette, 2009). We hypothesized that OF speed discrimination thresholds would be increased in the young-old adults, and that their ability to adjust their walking speed while using OF speed information would be reduced in comparison with young adults. We further predicted that OFdriven walking speed modulation would be associated with OF speed perception.
Methods Participants Twelve healthy young-old adults (68 ± 4.39 years; 63–74 years; 6 males) and twelve healthy young adults (23 ± 2.41 years; 19–27 years; 6 males) participated in this study. Participants were recruited from the greater Montreal and Laval communities through flyers, and from volunteers at the Jewish Rehabilitation Hospital. The project was approved by the Ethics Committee of the Centre for Interdisciplinary Research in Rehabilitation of Greater Montreal and all participants provided written informed consent. Eligible participants were between the ages of 60 and 74 years or between the ages of 19 and 29 years, to be considered as an older adult or young adult, respectively. Additional inclusion criteria included regular community ambulation, as defined as walking a minimum of 45 min per week, and normal or correctedto-normal visual acuity (20/20 vision) as assessed with a Snellen chart. Exclusion criteria included the presence of orthopedic, rheumatological, cardiopulmonary, and central or peripheral neurological conditions interfering with locomotion, as assessed during a screening interview. Participants with cognitive deficits (Montreal Cognitive Assessment score < 26) (Nasreddine, Phillips, & Chertkow, 2012), a history of dizziness (Dizziness Handicap Inventory score ≥ 30) (Treleaven, 2006), as well as walking and balance impairments (Functional Gait Assessment score ≤ 22) (Wrisley & Kumar, 2010), were also excluded.
Study Design This study used a quasi-experimental design in which participants were assessed on their ability to discriminate OF speed (perceptual experiment) and to make walking speed adjustments (walking experiment). Participants were assessed over 2 sessions taking place within the same week. The order of the experiments was randomized. A baseline clinical assessment screened for the exclusion criteria described earlier and also included an evaluation of the participants’ anthropometric characteristics. MC Vol. 19, No. 3, 2015
194 Lalonde-Parsi and Lamontagne
Comfortable and maximal walking speed over 10m was also assessed, using the mean of 3 trials per condition.
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
Equipment During both the perceptual and walking experiments, participants were assessed while wearing an NVisor SX60 head-mounted display (HMD). This HMD has a 1280 × 1024 pixel resolution and a 60° diagonal field of view (34° vertical × 44° horizontal). Participants viewed, within the HMD, a rich-textured virtual environment (VE) representing a corridor (Lamontagne et al., 2007) simulating forward self-motion at different speeds. The virtual corridor was 15m long with two archways located 10m apart at 3m and 13m of forward displacement. For the walking experiment, participants walked on a self-paced treadmill that was fitted with horizontal low-friction sliding handrails, permitting comfortable arm movements in the sagittal plane (Figure 1). The sliding handrails also allowed the participants’ position to change along the antero-posterior axis and were thus unlikely to provide a position feedback that influenced the control of walking speed. The treadmill’s servo-controlled motor allowed participants to modify their walking speed based on a real-time algorithm involving both the distance measured by an electro-potentiometer (tethered to the pelvis of the user) and the treadmill speed (derived from a microcontroller). The belt accelerated when the participant moved forward on the treadmill, and decelerated the belt when the participant moved backward on the treadmill, and thus adjusted treadmill speed in response to changes in the walking
Figure 1 — A participant walking on the self-paced treadmill while viewing the virtual environment through the head-mounted display. Also shown is the potentiometer tethered to the subject, which allowed controlling the treadmill’s speed in real time based on the anteroposterior position of the participant. A safety harness worn by the participant was suspended from overhead. MC Vol. 19, No. 3, 2015
Perception and Modulation of Walking Speed 195
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
speed of the user. All participants wore a safety harness suspended overhead the treadmill to provide support in the event of a fall. Using the CAREN 3.8.11 software from Motek, the speed of the VE was controlled in real-time and matched the speed of the treadmill during the habituation period and to determine the participants’ comfortable walking speed on the treadmill. For the actual walking experiment, a nonimmersive paradigm was used, where the VE visualized in the HMD was not influenced by the head movements or by the walking speed of the participant. Participants were fitted with earphones and listened to white and pink noise during the walking experiment to avoid auditory cues that could aid in self-speed estimation.
Setup & Procedure Speed Discrimination Task. Participants completed the evaluation while seated and watched the virtual corridor in the HMD. For this experiment only, however, the two archways delineating the 10m distance were removed so as to control for distance cues that could aid in speed discrimination (Sun, Lee, Campos, Chan, & Zhang, 2003). Speed discrimination thresholds were measured with a twoalternative forced choice paradigm for a standard reference speed of 1.0 m/s. The participant was presented with two visual scenes depicting forward self-motion through the virtual hallway: a standard stimulus (set to 1.0m/s) and a test stimulus. The presentation order of the standard and test stimuli was randomized. The test stimulus always moved faster than the standard and its speed varied from trial to trial using a two-down one-up staircase paradigm (Lakshminarayanan, Raghuram, & Khanna, 2005), in which the relative difference between the standard and test stimulus speeds (also referred to as the Weber fraction) decreased following two consecutive correct responses, and increased following every incorrect response. The staircase stopped after 16 reversals. The Weber fractions initially changed by 0.2 log units until the second reversal, by 0.1 log units until the fourth reversal, and by 0.05 log units for subsequent trials. Speed discrimination threshold was calculated as the geometric mean of the last eight reversals. The VE presentation time randomly varied and ranged from 500ms to 1000ms to discourage participants from estimating differences in speed by approximating the duration of the visual scene and an interstimulus interval of 1000 ms demarcated the two VE presentations. A static circular target placed at the end of the virtual corridor was also included to facilitate gaze fixation. While fixating the target at all times, participants were instructed to identify, verbally, which of the two scenes was moving faster by saying ‘one’ or ‘two’. They were given a series of 5–10 practice trials and the experiment was completed once the algorithm had identified a discrimination threshold. The algorithm and calculation of the threshold was programmed and recorded in CAREN 3.8.11.
Walking Experiment Before data collection, all participants underwent a familiarization period, which included walking on the treadmill without the HMD (≥ 2 min) followed by walking with the HMD and visualizing the VE (≥ min). During this familiarization session, MC Vol. 19, No. 3, 2015
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
196 Lalonde-Parsi and Lamontagne
the expanding OF speed of the VE was matched in real-time to the participant’s self-selected walking speed. After the familiarization period, the comfortable walking speed was measured as the average self-selected speed over 1 min of steady state walking, and the same procedure was repeated to measure fast walking speed. In the experiment itself, participants were presented with different optic flow speeds and were instructed to walk at the same speed as that perceived in the VE. The OF speeds were manipulated, in a random order, such that they corresponded to 0.25, 0.5, 0.75, 1, 1.25, 1.50, 1.75, or 2 times the individual’s comfortable walking speed. Two blocks of 8 trials (8 trial pairs) were completed, for a total of 16 trials. Note that the VE was static to start with and began moving only when participants had reached their comfortable walking speed. At that point, participants perceived themselves starting 3 m from the first archway and continued walking until a stop sign appeared 2 m after reaching the second archway. Information about walking speed, walking distance and OF speed were collected at 100Hz in CAREN and stored for off-line analysis. Rest periods were provided during the walking blocks and, as needed, between the walking trials. Five of the 24 participants showed mild symptoms of dizziness that could be related to motion sickness during the experiment.
Data Analysis Speed discrimination threshold was calculated as the geometric mean of the last eight reversals, as described in the procedure section. The walking data were imported in Matlab and analyzed with a custom-made routine. Speed modulation variables, further detailed below, were calculated based on the average walking speed observed between the two archways of the virtual corridor (10 m), to avoid acceleration and deceleration phases. For each OF condition, the difference between the actual speed of walking and that of the optic flow was calculated (Δ Speed). The ratio of walking speed to OF speed (W/OF Speed Ratio) was also calculated. Each ratio value was then converted to base-10 logarithm (log (W/OF Speed Ratio)) to yield the walking speed error. The slope characterizing the relationship between walking speed errors (log(W/OF Speed Ratio)) and the base-10 log transformation of OF speed ratios (log (OF Speed Ratio) was calculated for each individual and for the average values of each group. In the present context, where intercept values were found to be ~0, the slope values represent an overall estimation of the walking speed errors across all OF speeds: the steeper the slope, the larger the walking speed errors; at variance, a gentle slope signifies that walking speed errors are smaller, or closer to zero. A T-test for independent samples was used to compare speed discrimination thresholds (discrimination task) between young and older adults. A repeated measure mixed model analysis of variance was conducted to determine whether walking speed errors differed between age groups (old vs. young) and across OF speed ratios (8 conditions, from 0.25 to 2). Post hoc comparisons with Bonferroni adjustments were conducted. Pearson’s correlation coefficients were used to quantify the relationship between the performance on the perceptual task (speed discrimination thresholds) and that on the walking task (slope values). Note that the discrimination threshold of one young female was excluded from the analysis MC Vol. 19, No. 3, 2015
Perception and Modulation of Walking Speed 197
because of difficulty concentrating during the task. The level of significance was set to p < .05. Statistical analyses were performed using SPSS Version 20.
Results
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
Speed Discrimination Results obtained for speed discrimination thresholds with respect to age group are illustrated in Figure 2. Significant differences were observed between the two age groups (t(21) = 2.08, P = 0.004), with older adults presenting higher discrimination thresholds (Weber Fraction = 0.18 ± 0.08, mean± 1 SD) than young adults (0.11 ± 0.04).
Walking Speed Adjustments Overground walking speed of the older adults (1.29 ± 0.13 m/s) was similar to that of the young adults (1.38 ± 0.18 m/s) at comfortable pace, but was significantly slower (P = 0.003) in the older group (1.71 ± 0.19 m/s) than in the younger group (1.93 ± 0.12 m/s) for the fast walking condition. Mean maximal walking speeds reached during the treadmill experiments for the young in both walking paradigms (1.69 ± 0.25m/s) and older adults (1.31 ± 0.30m/s) fell below the maximal overground walking speed of the participants. Figure 3 shows the walking speeds exhibited by one young and one older participant as a function of the different OF speeds. Both participants adjusted their walking speed to match that of the OF but their performance varied as a function of the OF speed. The participants, especially the older adult illustrated in this example, walked faster than the targeted speed at slow OFs, and slower than
Figure 2 — Bar graph representing the mean (± 1 SD) speed discrimination thresholds for the young and older participants. Thresholds are expressed as Weber fractions. Speed discrimination thresholds were found to be significantly larger in the older adults compared with the young adults (p < .01). MC Vol. 19, No. 3, 2015
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
198 Lalonde-Parsi and Lamontagne
Figure 3 — Walking speed plotted against optic flow (OF) speed in one young adult and one older participant. The dotted line represents an ideal scenario, where walking speeds perfectly match the different OF speeds. In this example, the participants had a similar comfortable walking speed, such that they were exposed to identical OF speed changes.
the targeted speed at fast OFs. Speed matching was the best at 1.0 m/s, which corresponded to the participant’s comfortable walking speed. Similar findings were observed when considering all participants, with larger differences between actual and targeted speed being observed at slow and fast OFs, and smallest difference at an OF ratio of 1 (Figure 4a). At the slowest OF speed (OF ratio = 0.25), mean W/OF speed ratios ranged from 1.8 (young adults) to 3.4 (older adults), indicating that the actual walking speed was 1.8–3.4 times faster compared with the targeted OF speed (Figure 4b). W/OF speed ratios of » 1.0 (young = 1.0; older = 0.9) and
original article
Perception of Self-Motion and Regulation of Walking Speed in Young-Old Adults Marie-Jasmine Lalonde-Parsi and Anouk Lamontagne McGill University Whether a reduced perception of self-motion contributes to poor walking speed adaptations in older adults is unknown. In this study, speed discrimination thresholds (perceptual task) and walking speed adaptations (walking task) were compared between young (19–27 years) and young-old individuals (63–74 years), and the relationship between the performance on the two tasks was examined. Participants were evaluated while viewing a virtual corridor in a helmet-mounted display. Speed discrimination thresholds were determined using a staircase procedure. Walking speed modulation was assessed on a self-paced treadmill while exposed to different self-motion speeds ranging from 0.25 to 2 times the participants’ comfortable speed. For each speed, participants were instructed to match the self-motion speed described by the moving corridor. On the walking task, participants displayed smaller walking speed errors at comfortable walking speeds compared with slower of faster speeds. The young-old adults presented larger speed discrimination thresholds (perceptual experiment) and larger walking speed errors (walking experiment) compared with young adults. Larger walking speed errors were associated with higher discrimination thresholds. The enhanced performance on the walking task at comfortable speed suggests that intersensory calibration processes are influenced by experience, hence optimized for frequently encountered conditions. The altered performance of the young-old adults on the perceptual and walking tasks, as well as the relationship observed between the two tasks, suggest that a poor perception of visual motion information may contribute to the poor walking speed adaptations that arise with aging. Keywords: aging, gait speed, locomotion, visual cues, virtual reality
Mobility impairments are prevalent among the elderly population, as evidenced by 8–19% of noninstitutionalized older adults having difficulty walking or requiring the assistance of another person or an assistive device to walk (Dawson, Hendershot, & Fulton, 1997; Leon & Lair, 1990). Older adults typically adopt slower comfortable walking speeds compared with young adults, with a rapid decline of 12–16% per decade being observed as from the critical age of 62 years (Himann, The authors are with the School of Physical & Occupational Therapy, McGill University, Montreal, Quebec. Address author correspondence to Anouk Lamontagne at [email protected]. 191
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
192 Lalonde-Parsi and Lamontagne
Cunningham, Rechnitzer, & Paterson, 1988). The capacity to reach fast, maximal walking speeds is even further altered (Bohannon, 1997; Cunningham, Rechnitzer, Pearce, & Donner, 1982; Himann et al., 1988; Murray, Kory, & Clarkson, 1969), which suggests a reduced ability to make the speed adjustments that are necessary to respond to contextual demands (e.g., speeding up to cross a street or catch a train). Beyond anthropometric characteristics, factors that are the most commonly suggested to contribute to slow walking speed in older adults include reduced lower extremity muscle strength and power, reduced maximal aerobic capacity, presence of health problems and pain (Bendall, Bassey, & Pearson, 1989; Bohannon, 1997; Buchner et al., 1996; Cunningham et al., 1982; Cuoco et al., 2004). While some reports suggest that an altered perception of visual stimuli may also contribute (Hassan, Lovie-Kitchin, & Woods, 2002; Kuyk & Elliott, 1999; Sakari et al., 2010), little is known about the influence of visual information perception on the control of walking speed in older adults. Successful ambulation through our environment requires the integration of sensory information from visual, vestibular, and proprioceptive inputs. In particular, controlling self-motion is contingent upon the information about our perceived speed of movement. We derive speed information from optic flow (OF), a pattern of motion generated at the retina by self-motion through an environment (Gibson, 2009), and use it to monitor and update our walking velocity. Previous studies have shown that healthy young and older adults, when instructed to maintain a comfortable walking speed while viewing changing OF speeds, respond similarly by slowing down in response to faster OFs and speeding up with slower OFs (Konczak, 1994; Prokop, Schubert, & Berger, 1997; Schubert, Prokop, Brocke, & Berger, 2005). After several minutes of walking, this out-of-phase modulation of walking speed attenuates, possibly due to a recalibration process between proprioceptive and visual information (Prokop et al., 1997). Such experimental paradigm, where participants are instructed to maintain their walking speed, leads to unintentional speed modulations that reflect the ability of the central nervous system (CNS) to respond to mismatching sensory cues (e.g., visual vs. body-based cues). As evidenced in a recent study in stroke survivors, however, such ability is poorly associated with the participants’ ability to achieve voluntary and appropriately scaled speed modifications while using OF information (Lamontagne, Fung, McFadyen, & Faubert, 2007). Whether older adults can use OF cues to make volitional walking speed adjustments remains unknown. A number of studies have observed alterations in motion detection accompanying normal aging, with deteriorations starting to appear in the young-old age range (65–74 years) (Bennett, Sekuler, & Sekuler, 2007; Gilmore, Wenk, Naylor, & Stuve, 1992; Trick & Silverman, 1991). Specifically, the ability to discriminate the speed of visual motion information, displayed as visual gratings or moving dots, is compromised with advanced age (Habak & Faubert, 2000; Norman, Ross, Hawkes, & Long, 2003; Raghuram, Lakshminarayanan, & Khanna, 2005; Snowden & Kavanagh, 2006). None of the existing studies, however, have examined speed discrimination abilities while using meaningful and ecological displays that would resemble the visual sceneries experienced in daily life. Moreover, whether a reduction in the ability to discriminate speed from OF cues in older age contributes in part to the locomotor alterations in this population is still unclear. MC Vol. 19, No. 3, 2015
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
Perception and Modulation of Walking Speed 193
The primary aim of this study was to examine and compare abilities of young and older adults to discriminate OF speeds and to make walking speed adjustments in response to changing OF speeds while immersed in a meaningful, ecological virtual environment (VE). A secondary aim was to investigate the relationship between speed discrimination and walking speed modulation abilities in the older adults. Because changes in visual-perceptual, balance and locomotor abilities are noticeable in persons in their sixties and seventies, and since these abilities further deteriorate as participants enter the ‘old’ age range (≥75 years) (Dobbs et al., 1993; P. Haibach, Slobounov, & Newell, 2009; P. S. Haibach, Slobounov, Slobounova, & Newell, 2007; Himann et al., 1988), this study specifically targeted ‘young-old’ adults between the ages of 60 and 74 years (Zizza, Ellison, & Wernette, 2009). We hypothesized that OF speed discrimination thresholds would be increased in the young-old adults, and that their ability to adjust their walking speed while using OF speed information would be reduced in comparison with young adults. We further predicted that OFdriven walking speed modulation would be associated with OF speed perception.
Methods Participants Twelve healthy young-old adults (68 ± 4.39 years; 63–74 years; 6 males) and twelve healthy young adults (23 ± 2.41 years; 19–27 years; 6 males) participated in this study. Participants were recruited from the greater Montreal and Laval communities through flyers, and from volunteers at the Jewish Rehabilitation Hospital. The project was approved by the Ethics Committee of the Centre for Interdisciplinary Research in Rehabilitation of Greater Montreal and all participants provided written informed consent. Eligible participants were between the ages of 60 and 74 years or between the ages of 19 and 29 years, to be considered as an older adult or young adult, respectively. Additional inclusion criteria included regular community ambulation, as defined as walking a minimum of 45 min per week, and normal or correctedto-normal visual acuity (20/20 vision) as assessed with a Snellen chart. Exclusion criteria included the presence of orthopedic, rheumatological, cardiopulmonary, and central or peripheral neurological conditions interfering with locomotion, as assessed during a screening interview. Participants with cognitive deficits (Montreal Cognitive Assessment score < 26) (Nasreddine, Phillips, & Chertkow, 2012), a history of dizziness (Dizziness Handicap Inventory score ≥ 30) (Treleaven, 2006), as well as walking and balance impairments (Functional Gait Assessment score ≤ 22) (Wrisley & Kumar, 2010), were also excluded.
Study Design This study used a quasi-experimental design in which participants were assessed on their ability to discriminate OF speed (perceptual experiment) and to make walking speed adjustments (walking experiment). Participants were assessed over 2 sessions taking place within the same week. The order of the experiments was randomized. A baseline clinical assessment screened for the exclusion criteria described earlier and also included an evaluation of the participants’ anthropometric characteristics. MC Vol. 19, No. 3, 2015
194 Lalonde-Parsi and Lamontagne
Comfortable and maximal walking speed over 10m was also assessed, using the mean of 3 trials per condition.
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
Equipment During both the perceptual and walking experiments, participants were assessed while wearing an NVisor SX60 head-mounted display (HMD). This HMD has a 1280 × 1024 pixel resolution and a 60° diagonal field of view (34° vertical × 44° horizontal). Participants viewed, within the HMD, a rich-textured virtual environment (VE) representing a corridor (Lamontagne et al., 2007) simulating forward self-motion at different speeds. The virtual corridor was 15m long with two archways located 10m apart at 3m and 13m of forward displacement. For the walking experiment, participants walked on a self-paced treadmill that was fitted with horizontal low-friction sliding handrails, permitting comfortable arm movements in the sagittal plane (Figure 1). The sliding handrails also allowed the participants’ position to change along the antero-posterior axis and were thus unlikely to provide a position feedback that influenced the control of walking speed. The treadmill’s servo-controlled motor allowed participants to modify their walking speed based on a real-time algorithm involving both the distance measured by an electro-potentiometer (tethered to the pelvis of the user) and the treadmill speed (derived from a microcontroller). The belt accelerated when the participant moved forward on the treadmill, and decelerated the belt when the participant moved backward on the treadmill, and thus adjusted treadmill speed in response to changes in the walking
Figure 1 — A participant walking on the self-paced treadmill while viewing the virtual environment through the head-mounted display. Also shown is the potentiometer tethered to the subject, which allowed controlling the treadmill’s speed in real time based on the anteroposterior position of the participant. A safety harness worn by the participant was suspended from overhead. MC Vol. 19, No. 3, 2015
Perception and Modulation of Walking Speed 195
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
speed of the user. All participants wore a safety harness suspended overhead the treadmill to provide support in the event of a fall. Using the CAREN 3.8.11 software from Motek, the speed of the VE was controlled in real-time and matched the speed of the treadmill during the habituation period and to determine the participants’ comfortable walking speed on the treadmill. For the actual walking experiment, a nonimmersive paradigm was used, where the VE visualized in the HMD was not influenced by the head movements or by the walking speed of the participant. Participants were fitted with earphones and listened to white and pink noise during the walking experiment to avoid auditory cues that could aid in self-speed estimation.
Setup & Procedure Speed Discrimination Task. Participants completed the evaluation while seated and watched the virtual corridor in the HMD. For this experiment only, however, the two archways delineating the 10m distance were removed so as to control for distance cues that could aid in speed discrimination (Sun, Lee, Campos, Chan, & Zhang, 2003). Speed discrimination thresholds were measured with a twoalternative forced choice paradigm for a standard reference speed of 1.0 m/s. The participant was presented with two visual scenes depicting forward self-motion through the virtual hallway: a standard stimulus (set to 1.0m/s) and a test stimulus. The presentation order of the standard and test stimuli was randomized. The test stimulus always moved faster than the standard and its speed varied from trial to trial using a two-down one-up staircase paradigm (Lakshminarayanan, Raghuram, & Khanna, 2005), in which the relative difference between the standard and test stimulus speeds (also referred to as the Weber fraction) decreased following two consecutive correct responses, and increased following every incorrect response. The staircase stopped after 16 reversals. The Weber fractions initially changed by 0.2 log units until the second reversal, by 0.1 log units until the fourth reversal, and by 0.05 log units for subsequent trials. Speed discrimination threshold was calculated as the geometric mean of the last eight reversals. The VE presentation time randomly varied and ranged from 500ms to 1000ms to discourage participants from estimating differences in speed by approximating the duration of the visual scene and an interstimulus interval of 1000 ms demarcated the two VE presentations. A static circular target placed at the end of the virtual corridor was also included to facilitate gaze fixation. While fixating the target at all times, participants were instructed to identify, verbally, which of the two scenes was moving faster by saying ‘one’ or ‘two’. They were given a series of 5–10 practice trials and the experiment was completed once the algorithm had identified a discrimination threshold. The algorithm and calculation of the threshold was programmed and recorded in CAREN 3.8.11.
Walking Experiment Before data collection, all participants underwent a familiarization period, which included walking on the treadmill without the HMD (≥ 2 min) followed by walking with the HMD and visualizing the VE (≥ min). During this familiarization session, MC Vol. 19, No. 3, 2015
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
196 Lalonde-Parsi and Lamontagne
the expanding OF speed of the VE was matched in real-time to the participant’s self-selected walking speed. After the familiarization period, the comfortable walking speed was measured as the average self-selected speed over 1 min of steady state walking, and the same procedure was repeated to measure fast walking speed. In the experiment itself, participants were presented with different optic flow speeds and were instructed to walk at the same speed as that perceived in the VE. The OF speeds were manipulated, in a random order, such that they corresponded to 0.25, 0.5, 0.75, 1, 1.25, 1.50, 1.75, or 2 times the individual’s comfortable walking speed. Two blocks of 8 trials (8 trial pairs) were completed, for a total of 16 trials. Note that the VE was static to start with and began moving only when participants had reached their comfortable walking speed. At that point, participants perceived themselves starting 3 m from the first archway and continued walking until a stop sign appeared 2 m after reaching the second archway. Information about walking speed, walking distance and OF speed were collected at 100Hz in CAREN and stored for off-line analysis. Rest periods were provided during the walking blocks and, as needed, between the walking trials. Five of the 24 participants showed mild symptoms of dizziness that could be related to motion sickness during the experiment.
Data Analysis Speed discrimination threshold was calculated as the geometric mean of the last eight reversals, as described in the procedure section. The walking data were imported in Matlab and analyzed with a custom-made routine. Speed modulation variables, further detailed below, were calculated based on the average walking speed observed between the two archways of the virtual corridor (10 m), to avoid acceleration and deceleration phases. For each OF condition, the difference between the actual speed of walking and that of the optic flow was calculated (Δ Speed). The ratio of walking speed to OF speed (W/OF Speed Ratio) was also calculated. Each ratio value was then converted to base-10 logarithm (log (W/OF Speed Ratio)) to yield the walking speed error. The slope characterizing the relationship between walking speed errors (log(W/OF Speed Ratio)) and the base-10 log transformation of OF speed ratios (log (OF Speed Ratio) was calculated for each individual and for the average values of each group. In the present context, where intercept values were found to be ~0, the slope values represent an overall estimation of the walking speed errors across all OF speeds: the steeper the slope, the larger the walking speed errors; at variance, a gentle slope signifies that walking speed errors are smaller, or closer to zero. A T-test for independent samples was used to compare speed discrimination thresholds (discrimination task) between young and older adults. A repeated measure mixed model analysis of variance was conducted to determine whether walking speed errors differed between age groups (old vs. young) and across OF speed ratios (8 conditions, from 0.25 to 2). Post hoc comparisons with Bonferroni adjustments were conducted. Pearson’s correlation coefficients were used to quantify the relationship between the performance on the perceptual task (speed discrimination thresholds) and that on the walking task (slope values). Note that the discrimination threshold of one young female was excluded from the analysis MC Vol. 19, No. 3, 2015
Perception and Modulation of Walking Speed 197
because of difficulty concentrating during the task. The level of significance was set to p < .05. Statistical analyses were performed using SPSS Version 20.
Results
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
Speed Discrimination Results obtained for speed discrimination thresholds with respect to age group are illustrated in Figure 2. Significant differences were observed between the two age groups (t(21) = 2.08, P = 0.004), with older adults presenting higher discrimination thresholds (Weber Fraction = 0.18 ± 0.08, mean± 1 SD) than young adults (0.11 ± 0.04).
Walking Speed Adjustments Overground walking speed of the older adults (1.29 ± 0.13 m/s) was similar to that of the young adults (1.38 ± 0.18 m/s) at comfortable pace, but was significantly slower (P = 0.003) in the older group (1.71 ± 0.19 m/s) than in the younger group (1.93 ± 0.12 m/s) for the fast walking condition. Mean maximal walking speeds reached during the treadmill experiments for the young in both walking paradigms (1.69 ± 0.25m/s) and older adults (1.31 ± 0.30m/s) fell below the maximal overground walking speed of the participants. Figure 3 shows the walking speeds exhibited by one young and one older participant as a function of the different OF speeds. Both participants adjusted their walking speed to match that of the OF but their performance varied as a function of the OF speed. The participants, especially the older adult illustrated in this example, walked faster than the targeted speed at slow OFs, and slower than
Figure 2 — Bar graph representing the mean (± 1 SD) speed discrimination thresholds for the young and older participants. Thresholds are expressed as Weber fractions. Speed discrimination thresholds were found to be significantly larger in the older adults compared with the young adults (p < .01). MC Vol. 19, No. 3, 2015
Downloaded by Univ of Massachusetts Lib on 09/18/16, Volume 19, Article Number 3
198 Lalonde-Parsi and Lamontagne
Figure 3 — Walking speed plotted against optic flow (OF) speed in one young adult and one older participant. The dotted line represents an ideal scenario, where walking speeds perfectly match the different OF speeds. In this example, the participants had a similar comfortable walking speed, such that they were exposed to identical OF speed changes.
the targeted speed at fast OFs. Speed matching was the best at 1.0 m/s, which corresponded to the participant’s comfortable walking speed. Similar findings were observed when considering all participants, with larger differences between actual and targeted speed being observed at slow and fast OFs, and smallest difference at an OF ratio of 1 (Figure 4a). At the slowest OF speed (OF ratio = 0.25), mean W/OF speed ratios ranged from 1.8 (young adults) to 3.4 (older adults), indicating that the actual walking speed was 1.8–3.4 times faster compared with the targeted OF speed (Figure 4b). W/OF speed ratios of » 1.0 (young = 1.0; older = 0.9) and
