Measuring cognitive task performance

Success in cognitive task performance was a primary outcome measure, along with gait speed, for this study. The ability to score the memory task and hence compare results across conditions was therefore important.

Due to anticipated gait speed differences, between participants, between walking conditions (i.e. with and without FES) and also over the course of the study, the time taken to walk a fixed distance would vary. The average speed at which new FES users walk, prior to provision of FES, ranges from 0.19 m/s (Bogataj et al., 1995) to 0.94 m/s (Granat et al., 1996). Furthermore, an average improvement in gait speed with FES use has been reported to be 0.13 m/s (Kottink et al., 2004). Consequently, if the slowest speed of 0.19m/s and the fastest of 1.07 m/s (i.e. 0.94 + 0.13 m/s) are used to calculate the time it would take for a range of study participants to cover a typical 10m length walkway, the range could be 9 to 53 seconds.

Assuming that gait and cognitive task performance were both to be measured in a laboratory of fixed length, the following possible design options were considered:

a) Fix the number of target words/images and vary the length of each walking trial to accommodate different walking speeds.

b) Vary the number of target words/images presented over a fixed length walking trial to accommodate different walking speeds.

c) Vary the frequency with which target words/images appear over a fixed length walking trial to accommodate different walking speeds.

d) Use a fixed number of targets or words/images for all participants and all conditions.

Solution a): To ensure the same degree of recall difficulty, each memory task should have the same number of items to be recognised either verbally or visually; leading to the use of a memory task of a fixed duration.

The restrictions of using a gait laboratory of fixed length would create a situation where faster participants would need to turn around to continue walking, whilst performing the memory task, potentially more than once. Turning, with the inherent need to decelerate and then accelerate, is considered more cognitively demanding than walking in one direction at a steady state (Herman et al., 2011) and presents significant practical challenges if using a visual task. It was agreed that turning during dual-tasking was not an acceptable option as this would contaminate the task performance results and hence varying the length of each walking trial was rejected.

An alternative was to use the largest of the University of Salford’s gait laboratories in which participants could walk in a straight line over a distance of up to 25m, potentially avoiding the need for participants to turn whilst walking. Application of a visual task in this environment however, would create challenges of delivery. From a practical perspective, the lab is heavily used for teaching, and is not ideally suited for people with stroke, being rather crowded with equipment and having tripod-mounted cameras. Additionally, faster walkers (i.e. 1.07 m/s) potentially would require a 20m length walkway to see ten figures, for example, delivered at 2 second intervals. Thus the size of the projected figures would need to be sufficiently large to be viewed at the beginning of a walk, for the fastest walkers starting the furthest distance away from the screen. To ensure consistency of task difficulty, between conditions and participants, the size of figures would need to have been the same for each test. At the end of the walk, in the case of a fast walker, the figures may then be overly large, creating difficulties for those participants with hemi-neglect, who would potentially not pay attention to the entire figure, thus increasing the difficulty of the memory task. As such, the use of the walkway lengths tailored to gait speed was rejected.

Solution b): An alternative potential solution to the problem of speed variation between participants and across conditions would be to vary the number of targets, according to the time spent walking. However, it was not clear whether it would be possible to model the effects of altering the number of items to be recalled on cognitive task difficulty. If this proved impossible, this would mean analysing the results for memory task performance across conditions, as speed changed with condition and across participants, would not be possible.

Solution c): A further alternative approach was to deliver a fixed number of targets and alter the frequency with which they were delivered. This could accommodate differing gait speeds and thus ensure that the targets were seen wholly during walking. Hyndman’s study (2006) had used this approach to deliver a 7-item list, closely matching the time taken to walk 5m with the time taken to deliver the entire list by altering the frequency of item delivery. Lists were prepared over durations of 5, 10, 15 and 20 seconds, thus the range in interval between items was 0.7 to 2.9 seconds. Task performance was analysed with no apparent allowance for differences within the group nor between the seated condition, presumably delivered over a standard time period, and the walking condition. The speed of the group is reported

in the dual-task condition as 0.5 ± 0.3m/s, thus the range of time spent walking 5m was approximately 5 to 50 seconds. This indicates that each of the 5 to 20 second lists were probably used in the group study, creating a variation in the task applied, and potentially confounding outcomes comparing task performance when seated with walking.

This approach would result in the gap between each target differing between walking conditions, and between participants. This inconsistency in the task would be difficult to account for when analysing the results.

Solution d): Having rejected potential options (a-c), due to the difficulty in taking account of variations in task length/frequency and consequences of varying walking length on task performance, it was decided to fix the number and frequency of targets and length of walkway. This approach would fix the difficulty of cognitive task performance across conditions and between participants and allow for comparison. As a consequence during some trials, when walking speeds were faster, some of the target items would be heard or seen whilst the participant stood still after walking the length of the walkway; during the slower trials, participants would potentially complete viewing the targets before reaching the end of the walk. This is further justified below.

The memory tasks require the participant to pay attention to the task during the acquisition phase, committing the target words and figures to short term memory, and retain the targets in short term memory between the end of the acquisition phase and testing. This is a continuous process as the memory task continues and more items are committed to and retained in memory, including whilst standing still if this were to occur (in the faster participants) and while walking, after the end of the task (in slower participants).

The number of targets chosen for the memory tasks was based on the length of walkway, the average walking speed for new FES users and the frequency at which targets would be delivered. With a 10 metre walkway and an average walking speed for FES users, as calculated from the studies included in Kottink’s review (2004), of 0.5 m/s, the duration of walking was calculated as 20 seconds. Thus, with a 2 second interval for target delivery (as described in section 4.2.3c), the number of targets was defined as 10 per memory task. The length of test in the context of placing a

sufficient cognitive load was tested during piloting, with a plan to assess the scores on memory task performance as indicative of appropriate difficulty under all conditions.