Human Movement Science 21 (2002) 961972 www.elsevier.

com/locate/humov

Ageing eects on the attention demands of walking

W.A. Sparrow
a

a,*

, Elizabeth J. Bradshaw b, Ecosse Lamoureux c, Oren Tirosh a

School of Health Sciences, Deakin University, 221 Burwood Hwy, Burwood 3125, Vic., Australia b New Zealand Academy of Sport-Northern Region, Auckland, New Zealand Department of Ophthalmology, Center for Eye Research Australia, University of Melbourne, Melbourne 3001, Australia

Abstract Attention demands of walking were determined in six male and six female young adults (mean 26.3 yr) and 12 gender-matched healthy, active older adults (mean 71.1 yr) using a dual task procedure with a reaction time (RT) secondary task. In three conditions an auditory stimulus, a visual stimulus and both stimuli (auditory/visual) were presented. Relative to no-walking baselines, increased RT was found in all conditions revealing an attentional cost of normal walking. When participants traversed the laboratory walkway and also positioned one of their feet within a target area on the ground, attention demands were greater than in unconstrained walking. In the targeting task, RTs to the visual stimulus were longer than for the auditory stimulus due to the interference associated with viewing both the stimulus monitor and the foot-target. Older participants RTs in the visual and auditory/visual conditions, but not in the auditory condition, were signicantly longer than for the young group in both walking tasks but RTs for young and older adults were not dierent in no-walking baseline trials. Inspection of mean RT functions at time intervals following gait initiation (03000 ms from the rst step) suggested a uctuating attentional cost of walking with increased demands associated with contingencies requiring step length regulation. The ndings have applied signicance in demonstrating the possibility of increased falls and pedestrian accident risk in older individuals in dual task situations such as road crossing. 2002 Elsevier Science B.V. All rights reserved.

Corresponding author. Tel.: +61-3-9244-6334; fax: +61-3-9244-6017. E-mail address: sparrow@deakin.edu.au (W.A. Sparrow).

0167-9457/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0167-9457(02)00154-9

962

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

PsycINFO classication: 2330 Keywords: Attention; Perceptual motor coordination; Visual perception; Ageing

1. Ageing eects on the attention demands of walking Our everyday experience of walking is that it usually requires little attention, possibly because it is so highly practiced. Despite the relative ease with which we control the complex movements of walking it may not be reasonable to describe it as automatic. Smyth, Collins, Morris, and Levy (1996) suggested that automatic processing, albeit in cognitive tasks, is demonstrated when the task does not inhibit the capacity for doing other things. Adopting this criterion, walking is not proposed to be automatic and Mulder, Berndt, Pauwels, and Nienhuis (1993) have already shown that when concurrently performing a mental calculation walking speed reduced in older individuals but not in young controls. Similarly, Lajoie, Teasdale, Bard, and Fleury (1996) presented evidence of higher reaction time (RT) in older individuals when walking but it was unclear from their report as to whether RT increased signicantly relative to no-walking control RT trials. One research question emerging from the above observations was whether normal unobstructed walking has an attentional cost. A second issue was whether the attention demands of walking would vary as a function of the diculty or challenge posed by the gait task. It was anticipated that walking at a comfortable speed when unconstrained would have relatively low attention demands. Conversely, as the gait task became more challenging it was expected that the attentional requirements would increase. With the exception of the Lajoie et al. (1996) experiments and a study by Bardy and Laurent (1991) there are few accounts of the attention demands of walking, or other gait tasks, by which we can evaluate these hypotheses. In the experiment reported here the cognitive demands of walking were investigated using a foot-targeting task that required participants to place their foot, while walking, within a designated narrow area on the oor. In previous work it was found that the step length regulation required to do this task was particularly challenging even to healthy active older individuals (Halliday, Sparrow, & Begg, 1999). The common feature of such approach-to-target tasks is the ability to regulate step length to either place the foot on the object (as in a stair tread) or to position it in front of or beyond the obstacle, such as in stepping over something. Previous work has revealed how visual information is used to regulate stride parameters in order to control foot position to meet the spatial constraints of a target (Bradshaw & Sparrow, 2001; Berg, Wade, & Greer, 1994; Lee, Lishman, & Thomson, 1982). It was hypothesised that there is an attentional cost of processing visual input from the foot-target and regulating the motor output to modify step parameters. If, therefore, the primary task demands associated with the spatial constraints of foot placement were increased attentional resources available to a secondary task should be reduced. In Bardy and Laurents (1991) experiment, primary task demands were manipulated by having participants approach large and small cardboard targets suspended in a corridor with the requirement to touch them with the nose. Auditory RT was

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

963

measured during the targeting tasks and in control conditions while both sitting and walking unconstrained without targets. Bardy and Laurent (1991) clearly demonstrated a progressive increase in RT from the sitting control condition to unconstrained walking with progressively longer latencies for the large and small targets. In addition, as shown in Fig. 1, Bardy and Laurent (1991) found that RT increased as participants homed-in on the target, with considerably faster RTs during the early stages of the approach. Attention demands in similar gait tasks might, therefore, be expected to vary as a function of target constraints, such as their size, and where in the approach the probe is presented. It was considered important in this experiment to rst determine whether there is an attention demand of normal unconstrained walking and, second, to show whether the constraints of regulating step parameters on approaching a target impose an additional cognitive cost. The third aim was to demonstrate the eects of dierent presentation times of the RT stimulus to show how attention demands might uctuate during the approach to the target. While there are few previous accounts of the attention demands of walking in older people there are frequent references to how safety in older individuals might be compromised by the attentional demands of walking. In research on road crossing in older pedestrians, for example, Oxley, Fildes, Ihsen, Charlton, and Day (1997) concluded that reduced capacity to shift attention might explain the diculties associated with road crossing while at the same time looking out for oncoming vehicles. Patla (1993) discussed the increased dependence on vision when walking as reected in longer visual sampling by older adults. Similarly Anderson, Nienhuis, Mulder, and Hulstijn (1998) reported the common clinical observation of older adults looking at the ground more when walking than younger adults do. The proposal that deciencies in attentional resource sharing compromise safe locomotion is further supported by Brown, Shumway-Cook, and Woollacott (1999) who found that stepping is more attention demanding for older adults. In the experiment reported here the attention demands of walking were determined for both young and older individuals using both auditory and visual probe RT. The auditory stimulus was employed as a secondary

Fig. 1. Probe RT as a function of position in the sequence. 0 represents contact with the target. The mean duration of each interval (position) was 1200 ms. Data from Bardy and Laurent (1991) Fig. 2. Reprinted with permission of authors and publisher from: Bardy and Laurent (1991). Perceptual and Motor Skills 1991.

964

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

task with the expectation that it would have no eects on walking due to interference but would inuence gait parameters due to information processing demands. The visual RT task was expected to inuence walking variables due to the requirement to divert vision from foot placement to the stimulus. In addition to presenting the visual and auditory stimuli in dierent conditions a third task manipulation was employed in which either a visual or an auditory stimulus was presented randomly within the same condition. During walking in the everyday environment it might be expected that the modality of the external stimulus would be unanticipated, it could be either visual or auditory. The auditory/visual condition was, therefore, included in order to determine the eects of age on uncertainty concerning the stimulus modality.

2. Method 2.1. Participants Six men and six women aged 6478 yr (mean 71.1 yr) were recruited from the local community and 12 gender-matched young participants 2032 yr (mean 26.3 yr) were obtained from the academic community of Deakin University. Participants undertook informed consent procedures approved and mandated by the Deakin University Research Ethics Committee and also completed a screening questionnaire mandated and approved by the Deakin University Research Ethics Committee to ensure that they had a no overt balance or mobility dysfunction or medical conditions that would suggest lack of capacity to safely and eectively undertake the gait tasks. 2.2. Apparatus and procedure There were two gait tasks, unconstrained walking (walking) in which participants walked the 8 m walkway at a comfortable (preferred) speed, and a targeting task (targeting). The latter required walking at a comfortable speed but also stepping to place either foot between target strips xed perpendicular to the direction of travel 30 mm wider than shoe length. The target strips were balsa wood, semi-circular in crosssection, 20 mm wide and of high contrast (bright yellow) relative to the oor (Fig. 2). Participants rst performed each gait task without the RT task to determine a baseline walking speed. Following these trials was either a visual or an auditory baseline RT task undertaken while standing at the start of the walkway to determine the attention demand of the RT tasks alone. In the RT task participants were instructed to press a hand-held response button immediately the stimulus was presented. The auditory stimulus was the built-in computer chime and the visual stimulus was a red letter R presented on a computer monitor positioned at the end of the walkway. A total of 15 trials was presented, the rst three of which were practice. The timing apparatus was operated from the computer keyboard with a randomised delay of 03000 ms prior to stimulus onset. The experimenter gave the verbal command ready as a warning signal at the same time as the key was pressed to initiate the timing system.

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

965

Fig. 2. Experimental set-up showing walkway, force plates, timing gates and RT equipment. The force plates were used to calculate walking speed and the timing gates were employed to maintain walking speed from the baseline trials to the dual task conditions with the RT task.

In dual task conditions, the walking tasks were undertaken while at the same time performing the RT task as described above except that participants were instructed to start walking on the verbal ready command. The response button was attached to the computer via a light exible cable and held in the left hand so that the cable did not obstruct when walking from right to left. The response button was also held in the left hand in the baseline no-walking conditions. Following the ready signal the participants rst step onto a pressure-sensitive mat activated the RT software with the same randomized delay as for the baseline trials. The walkway was traversed in approximately 25003000 ms, the auditory and visual stimuli therefore probed the participant at any time from gait initiation, on contact with the stimulus mat, up to the end of the walkway. In the dual task conditions six no-stimulus catch trials were included within the block of 12 stimulus trials and walking speed was maintained within 10% of the baseline condition using experimenter feedback from the timing gates (Fig. 2). Trials either faster or slower than 10% of baseline were not excluded but the participant was asked to either walk faster or slower on the following trial. Data collection took approximately 2 h in two 1-h visits to the laboratory with one visit comprising all conditions for either walking or targeting. Presentation order of the gait tasks was counterbalanced with half of the participants in each age group undertaking the unconstrained task rst and the targeting task on the second visit and vice versa. Within each gait task, presentation order of conditions (visual, auditory and auditory/visual RT) was completely counterbalanced across participants to ensure that practice did not confound interpretation of the RT data, with two participants undertaking each of the six condition orders. The baseline RT condition always immediately preceded the corresponding gait task such that the baseline RT condition moved with the walking task to minimize

966

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

practice eects. A sequence of conditions was, for example, Walking Baseline, Visual RT Baseline, Walking Visual RT, Auditory RT Baseline, Walking Auditory RT, Auditory/visual RT Baseline, Walking Auditory/visual RT. 2.3. Design and analysis Repeated measures multivariate analysis of variance (ANOVA) was used to determine Age (young and older), Task (walking and targeting) and Condition (single and dual) eects on group mean auditory, visual and auditory/visual RT and walking speed. Gender was not included in the analysis. RTs below 100 ms and above 3000 ms were excluded. The walking speed analysis was undertaken to conrm that group mean walking speed in the baseline condition was not dierent from the dual task conditions. To reveal how the gait tasks may have inuenced the attentional demands of the task from gait initiation to the end of the walkway individual RTs were assigned to six time bins corresponding to six stimulus delay intervals of 500 ms and the mean RT calculated for each interval following stimulus onset; that is, 0500, 5011000, 10011500, 15012000, 20012500, and 25013000 ms. No statistical analysis of the time-delay data was undertaken.

3. Results The top panel of Fig. 3 presents walking velocity in the baseline (no RT) condition and averaged over the dual task conditions for walking and targeting. The ANOVA results revealed neither signicant main eects nor interactions for walking speed, conrming that in both walking and targeting there was no dierence, for either group, between the baseline condition and the dual task conditions. Any eects on RT reported below would not, therefore, be due to reduced attention to the (primary) gait task. As revealed in Fig. 3 for both age groups, auditory RT increased signicantly from baseline in both unobstructed walking and targeting F 1; 20 62:1, p < 0:01. As expected, the targeting task engendered a signicantly higher auditory RT (286 ms) than unobstructed walking (262 ms), F 1; 20 37:6, p < 0:01. The baseline auditory RTs were not dierent for the two participant groups, the ANOVA conrmed that there was no eect of age and no signicant interactions. In summary, auditory RT increased signicantly from baseline in both walking and targeting, targeting was associated with higher RTs than walking but there were no eects of age. For visual RT the ANOVA results indicated signicant eects of age (F 1; 22 14:3, p < 0:01) task (F 1; 22 16:2, p < 0:01, and condition F 1; 22 94:4, p < 0:01). There were, however, Age Condition (F 1; 22 22:2, p < 0:01) and Task Condition (F 1; 22 7:8, p < 0:01) interactions. It can be seen from Fig. 3 that visual RT increased considerably in the older group while undertaking the targeting task, whereas young participants showed considerably less increase from baseline in the targeting task. Using two one-way ANOVAs to compare the interaction

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

967

Fig. 3. Walking velocity (top panel) and RTs for visual, auditory, and auditory/visual RTs for young and older participants in no-walking baselines and in dual task conditions for walking and targeting tasks. All pairs of baseline and dual task RT means within each age group are signicantly dierent (p < 0:05) except for young participants in the auditory/visual condition.

means it was found that the Age Condition eect was due to age group dierences only in the dual task condition (older 559 ms, young 312 ms F 1; 22 20:7, p < 0:01) with no age dierences in baseline RT (older 257 ms, young 247 ms). The Task Condition interaction was also due to the two gait tasks showing no

968

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

signicant dierence in baseline (walk 245 ms, target 257 ms) while the targeting task elicited signicantly higher visual RT in the dual task condition (walk 338 ms, target 409 ms F 1; 23 12:7, p < 0:01). For visual RT, to summarise, there were group eects with the older group showing considerably increased RT in both walking and targeting dual task conditions. Baseline RT, as with the auditory stimulus, was consistent across conditions and not dierent between age groups. The pattern of results for auditory/visual RT in Fig. 3 was similar to the visual task, with age, task, and condition main eects (respectively F 1; 22 5:7, p < 0:05, F 1; 22 18:1, p < 0:01, and F 1; 22 22:0, p < 0:01) and a Task Condition interaction (F 1; 22 10:5, p < 0:01). As for visual RT the signicant interaction was due to the tasks revealing no dierence in baseline (walk 285 ms, target 303 ms) whereas the targeting task again engendered signicantly longer

Fig. 4. Mean RT as a function of time from gait initiation during walking and targeting. Baseline RT data are the mean values for the RT trials performed while stationary on the walkway prior to undertaking the walking task and in a separate condition prior to the targeting task.

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

969

RT in the dual task condition (walk 318 ms, target 407 ms F 1; 23 16:5, p < 0:01). In Fig. 4 are presented the mean RTs for the six 500 ms stimulus delay intervals for walking, targeting, and for the baseline trials. The mean RTs for each time interval reect the attention demands of the gait task as the participant progressed along the walkway. The elevation in RTs during both gait tasks relative to baseline can be seen and it is important to note that the curves for the baseline trials are essentially at. Inspection of all six panels of Fig. 4 demonstrates that in walking, for both participant groups, the RT functions are also relatively at except for somewhat longer RTs in at short (0500 ms) stimulus delays. There was, for example, a considerable elevation in visual RT in the rst time interval for the older participants during walking. All panels of Fig. 4 reveal consistently slower RTs for the older participants throughout both walking and targeting. With respect to the stimulus delay analyses for visual RT, a most interesting observation was that in the targeting task there was an increase for older participants close to the target, at approximately 20002500 ms, then a decline in the nal stage of the task. This feature of the data can also be seen for the young group but less accentuated. The auditory RT data (top right) also show a slight elevation in older participants at the same stimulus delay as for the visual RT but the curve for the young group is essentially at. The auditory/visual data also reveal elevated RTs around the target in the older group, similar to the visual RT data. In summary, the analyses in Fig. 4 revealed elevated RT relative to baseline in both tasks, consistently lower RT for the young participants throughout the trials, and a marked eect of the targeting task on visual RT for elderly individuals, with evidence of a small elevation in auditory RT associated with targeting.

4. Discussion In both young and older adults there was an attentional cost of walking as revealed in signicantly higher auditory RT compared to a no-walking baseline. It was also demonstrated that the foot-targeting task caused a signicantly greater elevation in auditory RT than did walking, suggesting that modications to step parameters elicited increased attention demands. Whether other aspects of gait task complexity are also associated with increased attention demand remains an empirical question but one hypothesis from the results here is that any requirement to undertake visually guided step length regulation will incur an attentional cost. The mean auditory and visual RTs for young and older participants were not dierent in the baseline conditions. The older individuals recruited for this study, therefore, had equally fast information processing abilities as measured by a simple (one-choice) RT task. This nding is consistent with our older participants reports that they were healthy and active. Previous studies have also indicated relatively fast RTs in physically active older people (Spirduso, 1975; Spirduso & Cliord, 1978). Spirduso and Cliord (1978) obtained simple visual RTs of 259 and 267 ms, respectively, for runners and racketball players aged 6070 yr (mean 64 yr) but a mean RT of 297 ms

970

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

for inactive individuals of the same age. The mean visual baseline RT of 257 ms for the older group employed here is, therefore, comparable to Spirduso and Cliords (1978) active groups. A focus of future research could be to show whether any eects of the gait tasks on RT are accentuated in older adults less healthy and physically active than those who participated here. It is important to emphasise that the visual RT task during gait involved structural interference in drawing on the same sensory input (Abernethy, 1988). Specically, the gait tasks and the visual RT task required the simultaneous use of vision, particularly in the targeting task, to obtain information regarding foot position. In everyday situations, however, competition between the visual demands of gait and secondary sources of visual information is commonplace and, therefore, independent of theoretical considerations, it is important to investigate. While the targeting task eects on auditory RT were weak, the auditory and visual RT delay data collectively suggest that attention demand will uctuate during the course of everyday locomotion, increasing as a contingency is approached and reducing when it has been negotiated. Bardy and Laurents (1991) auditory RT data as a function of time prior to the target (Fig. 1) are remarkably similar to those presented above (Fig. 4). The similarity extends to the elevated RTs at the beginning of the trials, possibly due to the attentional cost of gait initiation. The gait initiation interpretation is strengthened by the fact that in Fig. 4 there was no corresponding elevation in baseline RT at the rst delay interval (0500 ms). Bardy and Laurents (1991) data also showed a peak RT close to the target but because they stopped there the curves did not show a subsequent downward trend. The data presented here with those of Bardy and Laurent (1991) suggest that there is an attentional cost of gait initiation followed by reduced demand during steady-state walking, increased attentional costs are incurred as an obstacle or contingency is encountered followed by a decline when walking is again unconstrained. The ndings suggest that in everyday gait tasks increased attention demands would reduce the resources available for other (secondary) tasks. The high visual RT associated with the targeting task in the older individuals, for example, has implications for road crossing and other pedestrian activities in which at the same time as walking there is a requirement to attend to oncoming vehicles, pedestrians and any other potential hazards. In demanding gait tasks, competing visual information associated with a secondary stimulus could either lead to declines in gait task performance, increasing the risk of a fall or, alternatively, allocating increased attentional resources to the gait task may reduce the response time to a hazard. It was also expected that increased stimulus uncertainty in the auditory/visual condition would have elevated RT in older individuals while undertaking the gait tasks. The data did not support this hypothesis because the auditory/visual RTs for both gait tasks were lower than the visual RTs although no statistical tests were undertaken to compare mean RTs across conditions. Auditory/visual RT in the targeting task for the older participants was consistent with an eect of the visual stimulus presentations within that condition. It remains an open question, therefore, as to whether uncertainty concerning the stimulus modality has an interactive eect with age in dual task experiments.

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

971

One explanation for reduced secondary task performance with age is that increased attention compensates for age-related declines in sensory and cognitive processes (Shumway-Cook, Woollacott, Kerns, & Baldwin, 1997). As visual and kinaesthetic sensory input deteriorates, increased attention is used to heighten the signals from these systems (Shumway-Cook et al., 1997). While there does not appear to be a good test of the heightened attention hypothesis, it provides one of few general principles that might explain the reduced performance of older individuals in dual task situations involving gait. As a stimulus to future work we speculate that ageing is associated with increased attention to attenuated sensory input, as suggested above, but also increased attentional costs associated with output, specically modulating the neural commands to control step length during visually guided locomotion.

References
Abernethy, B. (1988). Dual-task methodology and motor skills research: Some applications and methodological constraints. Journal of Human Movement Studies, 14, 101132. Anderson, P., Nienhuis, B., Mulder, T., & Hulstijn, W. (1998). Are older adults more dependent on visual information in regulating self-motion than younger adults? Journal of Motor Behavior, 30, 104 113. Bardy, B. G., & Laurent, M. (1991). Visual cues and attention demand in locomotor positioning. Perceptual and Motor Skills, 72, 915926. Berg, W. P., Wade, M. G., & Greer, N. L. (1994). Visual regulation of gait in bipedal locomotion: Revisiting Lee, Lishman, & Thompson (1982). Journal of Experimental Psychology Human Perception and Performance, 20, 854863. Bradshaw, E. J., & Sparrow, W. A. (2001). Eects of approach velocity and foot-target characteristics on the visual regulation of step length. Human Movement Science, 20, 401426. Brown, L. A., Shumway-Cook, A., & Woollacott, M. H. (1999). Attentional demands and postural recovery: The eects of aging. Journal of Gerontology Series A: Biological Sciences and Medical Sciences, 54, M165M171. Halliday, D., Sparrow, W. A., & Begg, R. K. (1999). Foot targeting in young and elderly individuals: (I) Kinematic gait characteristics. In Proceedings of the Victorian IEEE conference of the engineering in medicine and biology society, February 2223 (pp. 165168). Clayton, Victoria: Monash University Press. Lajoie, Y., Teasdale, N., Bard, C., & Fleury, M. (1996). Attentional demands for walking: Age-related changes. In A.-M. Fernandez & N. Teasdale (Eds.), Changes in sensori-motor behavior in aging (pp. 235256). Elsevier Science B.V. Lee, D. N., Lishman, J. R., & Thomson, J. A. (1982). Regulation of gait in long jumping. Journal of Experimental Psychology, 8, 448459. Mulder, T., Berndt, H., Pauwels, J., & Nienhuis, B. (1993). Sensori-motor adaptability in the elderly and disabled. In G. E. Stelmach & V. Homberg (Eds.), Sensory-motor impairments in the elderly (NATO ASI series volume) (pp. 413426). Dordrecht: Kluwer. Oxley, J., Fildes, B., Ihsen, E., Charlton, J., & Day, R. H. (1997). Dierences in trac judgements between young and old pedestrians. Accident Analysis and Prevention, 29, 839847. Patla, A. E. (1993). Age-related changes in visually guided locomotion over dierent terrain. In G. E. Stelmach & V. Homberg (Eds.), Sensory-motor impairments in the elderly (NATO ASI Series volume) (pp. 231252). Dordrecht: Kluwer. Smyth, M. M., Collins, A. F., Morris, P. E., & Levy, P. (1996). Cognition in action (2nd ed.). Hove, East Sussex: Erlbaum.

972

W.A. Sparrow et al. / Human Movement Science 21 (2002) 961972

Shumway-Cook, A., Woollacott, M., Kerns, K. A., & Baldwin, M. (1997). The eects of two types of cognitive tasks on postural stability in older adults with and without a history of falls. Journal of Gerontology: Medical Sciences, 52, M232M240. Spirduso, W. W. (1975). Reaction and movement time as a function of age and physical activity level. Journal of Gerontology, 30, 435440. Spirduso, W. W., & Cliord, P. (1978). Replication of age and physical activity eects on reaction and movement time. Journal of Gerontology, 33, 2630.