The Feasibility and Efficacy of a Serial Reaction Time Task that Measures Motor Learning of Anticipatory Stepping

Abstract

Background:

Motor learning has been investigated using various paradigms, including serial reaction time tasks (SRTT) that examine upper extremity reaching and pointing while seated. Few studies have used a stepping SRTT, which could offer additional insights into motor learning involving postural demands. For a task to measure motor learning, naïve participants must demonstrate a) improved performance with task practice, and b) a dose-response relationship to learning the task.

Research Question:

Is a stepping SRTT feasible and efficacious for measuring motor learning? Methods: In this prospective study, 20 participants stood on an instrumented mat and were presented with stimuli on a computer screen. They stepped to the corresponding positions on the mat as quickly as possible. Presented stimuli included random sequences and a blinded imbedded repeating sequence. Three days after completing the randomly assigned practice dose [high dose group (n=10) performed 4320 steps; low dose group (n=10) performed 144 steps], a retention test of 72 steps was performed. Feasibility was measured as the proportion of participants who completed the assigned practice dose without adverse events. Efficacy was measured as within-group performance improvement on the random sequences and on the repeating sequence (paired t-tests), as well as a dose-response relationship to learning both types of sequences (independent t-tests).

Results:

All participants (mean age 26.8 years) completed all practice sessions without adverse events, indicating feasibility. High dose practice resulted in performance improvement while low dose did not; a dose-response relationship was found, with high dose practice resulting in greater learning of the task than low dose practice, indicating efficacy.

Significance:

This stepping SRTT is a feasible and efficacious way to measure motor learning, which could provide critical insights into anticipatory stepping, postural control, and fall risk. Future research is needed to determine feasibility, efficacy, and optimal practice dosages for older and impaired populations.

1. Introduction

Motor learning is a set of internal processes leading to a relatively permanent change in the capability to perform a motor skill [1]. Motor learning is often measured via retention test, or as a comparison of pretest and retention test performance, in order to capture its ‘relative permanence;’ in contrast, motor performance is commonly measured via the level of proficiency observed during practice [1]. Motor learning has been investigated using a variety of experimental paradigms, including serial reaction time tasks (SRTT) [2–4]. In a SRTT, participants map a visuospatial stimulus to a corresponding response location [5]. Stimuli are presented in a series of either random or repeating sequences, although participants are blinded to the repeating sequence. The ability to improve responses to stimuli presented in random sequences is termed general learning, and represents learning of the general movement skill, whereas the ability to improve responses to a repeating sequence is termed repeated-sequence learning and represents learning of the general movement skill combined with learning of the underlying motor sequence [4]. The most common SRTT paradigms consist of upper extremity reaching and pointing while seated [2–4].

While research using upper extremity SRTTs has provided insight into ways to optimize motor learning of fine motor reaching and pointing tasks [2–4], to our knowledge there is only one other SRTT paradigm that includes a task with any type of standing postural demands [6–8], and it is relatively new. This current lack of knowledge and methodology, moreover, has clinical implications. Motor practice, such as is provided during rehabilitation, can be used to help people with motor deficits learn or relearn motor skills, including those related to stepping responses, balance, and fall risk reduction [9]. But the lack of evidence specific to learning these types of movements hampers that progress. Developing a stepping SRTT can improve our understanding of how practice affects motor learning of anticipatory stepping, and will lead to new insights that could enhance recovery for people with such deficits.

To our knowledge, only our prior studies [10,11] and those of Du et al [6–8] have used a SRTT that incorporates stepping while standing, and none have formally reported its feasibility and efficacy for measuring motor learning. Additionally, while Du et al shows evidence of the postural demands of their stepping SRTT, including center of mass displacement, their task setup appears to be limited to the confines of the lab environment, and no data have been provided relating performance on that task to fall risk [6–8]. In contrast, the SRTT described in the current paper uses the same equipment and motor task methodology (although not the same step sequences) as previous studies that were able to detect fall risk in older adults based on their step response times [12], and that were used to show that home-based step training in older adults could improve step response times, postural sway, and dualtask ability compared to a control group [13]. Thus, a SRTT built into this existing framework, with the added convenience of using equipment that is feasible for use in a home setting, could provide important insights into anticipatory stepping, postural control, and fall risk, which seated upper extremity SRTTs cannot do.

Studies of use-dependent neuroplastic changes in humans [14] and animals [15,16] demonstrate that large amounts of practice lead to reorganization of the cortex and improved task performance. These studies show that increased practice results in greater task improvements, provided that the practiced task is challenging, progressive, and requires skill [17–19]. Similarly, meta-analyses suggest a positive dose-response relationship with practice [20,21], such that more practice results in better learning of motor tasks. Therefore, a research task that is adequately challenging, progressive, and skill-based to measure motor learning constructs should demonstrate a similar dose-response relationship.

As this stepping SRTT had not yet been tested for its ability to assess motor learning, it was important to first test its feasibility (including safety) and efficacy in a population without motor impairment, including age-associated motor changes. Thus, the purpose of this study was to test the feasibility and efficacy of a standing stepping SRTT for measuring motor learning, in healthy young adults. It is noted that several criteria are required in order for a task to efficaciously test motor learning: 1) participants must initially be naïve to the task, 2) practicing the task must lead to improved task performance, and 3) more task practice should result in more learning than less task practice (i.e., a dose-response relationship to learning). We therefore hypothesized that participants would complete the entire study without experiencing adverse events (feasibility), and that we would find step response time performance improvement as well as a dose-response relationship between the amount of practice and step response time learning (efficacy).

2. Methods

2.1. Design

In this prospective cohort study, participants were randomly assigned to one of two groups (NCT02898701). This study was approved by the University of Utah Institutional Review Board, and participants provided written informed consent prior to enrolling. A trained assessor performed initial assessments, after which a sealed envelope was opened to determine each participant’s group assignment: i.e., low dose (LD) or high dose (HD) of practice. Blocked randomization was used to ensure equal group size. Trained study staff supervised all practice sessions and performed retention tests.

2.2. Participants

Between September 2016 and May 2017, eligible participants aged 18–45 years were recruited from the Salt Lake area. Individuals were excluded if they were non-English speaking, or if they had any of the following: acute medical problems, uncorrected visual impairment, cognitive impairment (Montreal Cognitive Assessment score <26) [22], or any other conditions affecting mobility or balance that could influence their ability to perform the motor task (e.g., arthritic, orthopedic, metabolic, vestibular). Demographics and baseline measures of motor and cognitive function were collected to characterize the sample, including Montreal Cognitive Assessment, self-selected and fast gait speeds via the 10 Meter Walk Test [23,24], Four Square Step Test [25], and mini Balance Evaluation Systems Test [26].

A within-group effect size of 1.71 (Cohen’s d) has been reported for motor learning of a seated upper extremity SRTT [27] in a motor-impaired cohort (basal ganglia stroke). However, since this study includes between-group differences (i.e., high vs. low practice dosage), we expected a more modest effect size of approximately 1.35. A priori power analysis using this effect size, along with an alpha level of .05, a desired 80% power, and the use of independent t-tests suggested a sample of 10 participants per group (G*Power 3.1, Heinrich-Heine-University Düsseldorf, Germany). Thus, 20 participants were recruited.

2.3. Apparatus and task

The motor task (Figure 1) was modeled after other well-established SRTTs [2,4], with the primary difference being the effector and position used (i.e., lower extremities while standing, instead of upper extremities while sitting). In this SRTT, participants stood on an instrumented step mat [12,13], were presented with a stimulus on a computer screen, and were instructed to step to the corresponding position on the mat as fast as they could safely step (Figure 1). The step mat was part of the open-source DDR game Stepmania (www.stepmania.com) and was modified using custom code to create the SRTT sequences. The mats were manufactured for frequent use in the context of gaming. Approximately 66.35 Newtons of force (i.e., equaling a 6.77 kg stationary object) was required for the mat to register a step. Prior to each participant beginning testing or training, independent confirmation of the mat sensitivity was established by study personnel.

Figure 1.

Showing the instrumented step mat used for the stepping SRTT, with a participant mid-step. The participant was instructed to begin by standing in double limb support with one foot on each of the 2 rear arrows (“home” position). Each time a stimulus was presented, the participant stepped with their ipsilateral foot to the appropriate target arrow (i.e., left, right, left front, or right front) as fast as they could safely do so. This figure is reprinted from the Journal of Neurologic Physical Therapy, 43, G.N. Olivier, S.S. Paul, K.R. Lohse, C.S. Walter, S.Y. Schaefer, L.E. Dibble, Predicting Motor Sequence Learning in People With Parkinson Disease, 33–41, Copyright (2019), with permission from Wolters Kluwer [7].

The step mat included six pressure pads in the shape of arrows, of which two represented the “home” position (side-by-side at the back of the step mat) and four represented the targets (one to the right of home position, one to the left of home position, and two side-by-side directly in front of home position). Participants started by standing in double limb support with one foot on each of the two home position arrows at the rear of the mat while looking at an image of the step mat projected onto a computer screen. Participants stepped to the corresponding target arrow on the step mat each time a stimulus was presented on the computer screen. They were instructed to step using the ipsilateral foot relative to the target location (e.g., the left foot stepped to targets to the left or left front of the participant). When the requisite amount of force was applied to the correct target, the stimulus dimmed on the computer screen and participants returned their stepping foot to home position. For subsequent stimuli to appear on screen, the participant first had to apply enough force to the correct target with one foot, and then had to apply enough force to both of the home pressure pads simultaneously (i.e., with one foot on each home pressure pad). Step response time was collected and defined as time from stimulus presentation to foot pressing on target [1]. In the event that a participant applied force to an incorrect target, the computer screen would show the incorrect target change color (from white to gray) for the duration of force application, while the target stimulus remained on the screen (in bright green) until the participant applied force to the correct target.

The task was comprised of sequences of 12 steps that were either random or followed a repeating pattern. A computer randomization program created hundreds of random 12-step sequences, ensuring that participants encountered a random sequence only once during the entire study protocol. In contrast, participants experienced the repeating 12-step sequence once during every practice trial, although they were blinded to its presence. Within each 12-step sequence, the four target arrows were equally represented such that each target was presented three times during every sequence (whether that sequence was random or repeating).

2.4. Procedure

One practice trial comprised two consecutive 12-step sequences (one random, one repeating) presented in random order (Figure 2). One block of practice comprised 6 trials (i.e., 144 steps). Participants were provided a 25 second standing rest break between trials, and a 4-minute seated rest break between blocks. One complete day of practice included 6 blocks of practice, lasting approximately 90 minutes. During seated rest breaks, participants were provided with visual and verbal feedback [28–30] of their median step response time for the preceding block. Specifically, their median step response time was written on a dry erase board, and participants were verbally told “for that block of practice, on average, it took you X.XXX seconds to complete each step; your goal during the next block of practice is to step even faster, if you can safely do so.” After completing their assigned practice dose, participants had two days of no practice before returning on the third day post-practice for a retention test consisting of three trials of the SRTT.

Figure 2.

A schematic summary of the organization of the serial reaction time task (SRTT), and participant flow through practice and retention testing. Abbreviations: HD: high dose of practice. LD: low dose of practice. ^The order of the random and repeating sequences was randomized for each trial within a block. Note: the pretest consisted of the first 3 trials of each group’s first day of practice (i.e., first 72 steps), while end of acquisition was the last 3 trials of each group’s last day of practice (i.e., last 72 steps). Following two days of no practice, the retention test (i.e., 3 trials, totaling 72 steps) occurred on the third day post-practice.

Pretest performance was defined as mean step response time on the first three trials of the first day of practice. End of acquisition performance was defined as mean step response time on the final three trials of the last day of practice. (Note: In the LD group, because each participant’s practice is completed in a single day, their first and last day of practice are the same day; for the HD group, the fifth day of practice is their last day.) Retention performance was defined as mean step response time during retention testing. For all three timepoints, calculations were performed separately for the two different types of sequences: random and repeating.

To test whether changes in motor performance are affected by dose of motor practice (i.e., a dose-response), we randomly assigned participants to a LD group (1 block of 6 trials of task practice over one training day, equaling 144 total steps) or a HD group (6 blocks of 6 trials of daily task practice for five training days, equaling 4320 total steps) (Figure 2). During this practice, participants’ first 72 steps served at pretest, while their last 72 steps served as end of acquisition. Because it is known that, generally, “more is better” when it comes to motor practice [31–33], the HD group was provided a very high dose of practice to ensure substantial opportunity to learn the task, and an amount similar to that which has been shown to produce behavioral and cortical changes (~4500 repetitions) [34]; while the LD group was provided with a very small amount of practice, though still more than is provided during typical physical and occupational therapy treatments [35].

2.5. Data analysis

Visual inspection of the data revealed some outlying response time data points due to technical difficulties. Extreme outliers (step response time > 2.5 seconds) were excluded from analysis to eliminate equipment and computer errors but preserve steps that included participant errors, resulting in 0.35% of data being removed and treated as missing. Descriptive statistics provided baseline characteristics of the participants. For each type of sequence (random and repeating) post hoc independent samples t-tests compared pretest performance on the SRTT between the two groups.

Feasibility was measured as the proportion of participants who completed the assigned practice dose without adverse events, which were defined as injury, falls, pain, or discomfort. Efficacy was defined as general improvement and repeated-sequence improvement in both groups, as well as a dose-response relationship (i.e., better general learning and repeated-sequence learning in the HD group compared to the LD group). These outcomes were calculated: general improvement (end of acquisition performance compared to pretest performance on the random sequences); repeated-sequence improvement (end of acquisition performance compared to pretest performance on the repeating sequence); general learning (retention performance minus pretest performance on the random sequences); repeated-sequence learning (retention performance minus pretest performance on the repeating sequence). Improvement within both groups was determined using paired t-tests, while learning was compared between groups using independent samples t-tests. Post-hoc effect sizes were calculated for all significant findings using Cohen’s d, in which ≥0.2, ≥0.5, and ≥0.8 was considered small, moderate, and strong, respectively [36]. Alpha level was set at .05. Because this was a study of feasibility and efficacy, post hoc adjustments for multiple comparisons were not included. Analyses were conducted using JMP Pro 13 (SAS, Cary, NC, USA).

For each group (LD and HD), all variables (i.e., general improvement, repeated-sequence improvement, general learning, and repeated sequence learning) were tested for normality using the Shapiro-Wilk test. All variables met the assumption of normality with the exception of general learning and repeated sequence learning in the HD group. In the HD group, one participant performed random and repeating sequences worse at retention compared to pretest, causing a skew in the distribution of HD learning scores. When this participant was excluded, the remaining HD participants’ data were normally distributed. However, we did not exclude this participant from tests of our primary outcomes because 1) t-tests are very robust to departures from normality [37], and 2) since this participant performed poorly at retention, including their data in the analyses represents a more conservative estimate of the true population means, thereby reducing the risk of Type I statistical error.

3. Results

3.1. Participant Characteristics

Twenty adults (9 female, mean age 26.8 years) participated. Table 1 shows that the participants’ cognitive and gross motor functions were within normal limits, and that the groups were similar to one another at baseline with respect to demographic and clinical measures. However, compared to the HD group, the LD group performed the SRTT significantly faster at pretest for both the random (p=.0003) and repeating (p<.0001) sequences.

Table 1.

Baseline characteristics by group.

Outcome	HD (n = 10)	LD (n = 10)	All (n = 20)
Age (years)	25.9 (6.3)	27.7 (4.0)	26.8 (5.2)
Sex (F)	4 (40%)	5 (50%)	9 (45%)
MoCA (0–30)*	29.3 (1.1)	29.0 (0.8)	29.2 (0.9)
Self-selected gait speed (m/s)*	1.44 (0.27)	1.49 (0.17)	1.46 (0.22)
Fast gait speed (m/s)*	2.13 (0.34)	2.43 (0.50)	2.28 (0.45)
4SST time (s)	5.1 (1.1)	4.9 (1.4)	5.0 (1.3)
mini-BEST (0–28)*	27.0 (0.8)	27.6 (0.7)	27.3 (0.8)
Pretest Random Sequences, Step response time (s)	.984 (.105)†	.887 (.055)†	.935 (.107)
Pretest Repeating Sequence, Step response time (s)	.999 (.106)†	.881 (.065)†	.940 (.120)

Note. Data presented as mean (SD) or n (%).

The symbol (^†) indicates that post-hoc t-tests were performed on these two measures and found that the HD and LD groups were significantly different from one another, with respect to these pretests.

Abbreviations: HD: high dose of practice group. LD: low dose of practice group. MoCA: Montreal Cognitive Assessment. 4SST: Four Square Step Test. mini-BEST: Mini Balance Evaluations Systems Test.

^*Higher score indicates better performance.

3.2. Feasibility

All participants (100%) in both groups completed all practice sessions. Prior to retention testing, participants in the LD group each performed 144 total steps, while those in the HD group each performed 4320 total steps. None of the pre-defined adverse events were reported by any participants in either group.

3.3. Efficacy

Figure 3 illustrates all efficacy outcomes, including performance improvement and learning (Figure 3A,B), as well as each group’s performance at each timepoint for each type of sequence (Figure 3C,D).

Figure 3.

Efficacy of the stepping serial reaction time task (SRTT) via Improvement and Learning. All error bars illustrate standard error. A) Improvement for each participant was calculated: End of Acquisition scores minus Pretest scores. Mean improvement within each group is shown for the repeating sequence (i.e., repeated-sequence improvement) and for the random sequences (i.e., general improvement). The asterisks (*) indicate statistically significant within-group differences, based on paired t-tests and an alpha level of .05. B) Learning for each participant was calculated: Retention scores minus Pretest scores. Mean learning scores for each group are shown for the repeating sequence (i.e., repeated-sequence learning) and the random sequences (i.e., general learning). The asterisks (*) indicate statistically significant between-group differences, based on independent t-tests and an alpha level of .05. Mean step response times are also shown for each group at each testing timepoint for C) the repeating sequence and D) the random sequences.

3.3.1. Performance Improvement

Improvement was measured as pretest performance compared to end of acquisition performance. The HD group demonstrated significant (p=.002) general improvement (i.e., the random sequences), while the LD group did not (p=.29) (Table 2, Figure 3A), which translates to a strong practice effect (Cohen’s d=1.43) for the HD group on the random sequences. Similarly, the HD group demonstrated significant (p<.0001) repeated-sequence improvement (i.e., the repeating sequence), while the LD group did not (p=.46) (Table 2, Figure 3A), which equates to a strong practice effect (Cohen’s d=2.75) for the HD group on the repeating sequence.

Table 2.

Performance improvement on the stepping SRTT.

Outcome	Group	Pretest	End of Acquisition	p-value	Effect Size (Cohen’s d)
General Improvement (seconds)	High Dose	.984 (.105)	.832 (.107)	.002*	1.43
	Low Dose	.887 (.055)	.864 (.090)	.29	0.32
Repeated-sequence Improvement (seconds)	High Dose	.999 (.106)	.747 (.077)	<.0001*	2.75
	Low Dose	.881 (.065)	.872 (.081)	.46	0.12

Note. Data presented as mean (SD). Improvement was calculated via paired t-tests comparing End of Acquisition scores to Pretest scores. Scores and comparisons for each group are shown for the repeating sequence (i.e., repeated-sequence improvement) and for the random sequences (i.e., general improvement).

The asterisks (*) indicate statistically significant within-group differences, based on an alpha level of .05.

3.3.2. Learning

Learning was measured as the difference in performance from pretest to retention test. There was a significant difference in general learning (i.e., the random sequences, p=.03) and in repeated-sequence learning (i.e., the repeating sequence, p=.01) between the HD and the LD groups (Table 3, Figure 3B). This translates to strong dose-response learning effects of the random sequences (Cohen’s d=1.04) as well as of the repeating sequence (Cohen’s d=1.21).

Table 3.

Learning of the stepping SRTT.

Outcome	High Dose (n=10)	Low Dose (n=10)	p-value	Effect Size (Cohen’s d)
General Learning (seconds)	−.152 (.082)	−.057 (.100)	.03*	1.04
Repeated-sequence Learning (seconds)	−.190 (.105)	−.062 (.106)	.01*	1.21

Note. Data presented as mean (SD). Learning scores for each participant were calculated: Retention scores minus Pretest scores. Mean learning scores for each group are shown for the repeating sequence (i.e., repeated-sequence learning) and the random sequences (i.e., general learning).

The asterisks (*) indicate statistically significant between-group differences, based on independent t-tests and an alpha level of .05.

4. Discussion

4.1. Primary Findings

The purpose of this study was to test the feasibility and efficacy of a novel stepping SRTT as a measure of motor learning of a task with postural demands. In this study, all the participants completed their entire assigned dose of task practice without experiencing any adverse events, demonstrating that this stepping SRTT was feasible for healthy young adults. While boredom was not explicitly measured, the researchers anecdotally observed that some participants in the HD group spontaneously reported boredom during their fourth or fifth practice session. Efficacy of this task was also demonstrated, as evident by our findings that a) practicing the task resulted in performance improvement, provided that participants were given enough practice, and b) there was a dose-response relationship, such that HD practice resulted in better learning of the task than LD practice. These findings held true for both general improvement and learning (i.e., of the random sequences) and repeated-sequence improvement and learning (i.e., of the repeating sequence). The within-group effect sizes were large for performance improvement of the random and the repeating sequences for the HD group, and the between-group effect sizes were also large for learning of both types of sequences.

From pretest to retention test, the HD group improved by .152 (SD .082) and .190 (SD .105) seconds for general learning and repeated-sequence learning, respectively. This amount of retention may represent a clinically meaningful change in step response time, as previous findings from a similar task found that older adult fallers had step response times that were .154 seconds slower than those of non-fallers [12]. This should be interpreted with caution, however, given the difference in study populations (i.e., healthy young, versus community-dwelling adults).

These findings, including general and repeated-sequence improvement in the HD group as well as the dose-response relationship to learning seen between the two groups, are consistent with previous studies of SRTTs [2–4] and dosing of motor practice [14–21]. Like those studies, our behavioral findings may be related to underlying neuroplastic changes, possibly including cortical reorganization [14–16,38]. However, our findings are among the first using a stepping SRTT, and suggest that this novel task is challenging, progressive, and skill-based enough to be used for testing other practice variants and their effects on motor learning.

The LD group did not show significant performance improvement in either type of sequence, which could be due to under-dosing of practice, as these participants performed only 3% of the practice dose performed by the HD group. Current motor rehabilitation research suggests the importance of quantifying practice dosage as the number of repetitions of practice performed (which is consistent with how this study defined practice dosage) as opposed to defining it by the time spent practicing [31,35,39]. Studies also show that standard-of-care physical and occupational therapy includes very few repetitions of purposeful task practice addressing balance and postural control deficits [35,40], specifically reporting an average of 27 repetitions per session [35]. Knowing that people with motor deficits may require higher practice doses than healthy young adults [41–44], and in order to ensure the LD group was not under-dosed compared to standard-of-care, we chose a LD practice dosage well beyond what is reported in studies describing current clinical practice [35].

While SRTTs have been explored extensively as a way to study performance improvement and learning, versions that include standing and stepping have not. Balance impairment and falls are common, dangerous, and costly, particularly among older adults [45] and people with neurologic conditions [46–48], warranting the development of research paradigms that can assess motor improvement and learning of posturally-demanding tasks. This study describes a potentially feasible and efficacious paradigm for doing so.

4.2. Limitations and Future Directions

The authors acknowledge that this study included a small sample size, although it was sufficient to identify statistically significant performance improvement in the HD group, a dose-response relationship to learning, and large effect sizes. Possibly related to sample size, pretest performance on the SRTT differed between our two groups, despite random assignment to groups. This may have inflated our effect sizes, as the LD group performed better at pretest than the HD group, possibly contributing to a ceiling effect. The difference in pretest performance between the two groups was unrelated to any measured variable, but could have been related to something unmeasured, such as motivation, competitiveness, or interest in the motor task. These factors, and their relationship to ceiling effects on this task, should be investigated in future work. Additionally, due to funding constraints precluding us hiring interpreters, we excluded individuals who could not speak English, which may limit generalizability to more diverse populations.

While this study demonstrated a dose-response relationship associated with this task, it investigated two very different practice dosages (i.e., a very low versus very high dose of practice). There may be a more ideal dose somewhere between those two extremes, and future work should investigate what the optimal practice dosage is for various participant populations. Similarly, compared to the HD group, the LD group had significantly less exposure to study staff, and to motor practice in general. To address this, an alternative paradigm could transition the LD group into practicing an unrelated motor task once they complete their assigned dose of SRTT practice, thereby manipulating practice dosage of the SRTT while keeping the two groups standardized with regards to the amount of motor practice, in general, as well as exposure to study personnel. Additionally, this study compared two practice dosages in which all participants in each group performed the exact same number of repetitions. Given the interindividual variability of motor performance improvement [49], personalized methods of dosing practice may be more appropriate.

This practice paradigm was feasible and efficacious in an unimpaired cohort; however, we acknowledge that older adults and people with motor impairments (e.g., stroke, Parkinson disease) more commonly experience anticipatory stepping and postural control deficits, as well as falls. After this initial and important first step in examining a stepping SRTT as a measure of motor learning of a task with postural demands, future work will investigate its utility in aging, motor impaired, and cognitively impaired populations, as well as individualized practice dosages among various populations. Additionally, future work should investigate whether practice of this stepping SRTT is related to other forms of motor learning, such as transfer of learning to untrained, functional tasks requiring postural control, which is particularly salient to rehabilitation of impaired populations.

Highlights.

• A standing serial reaction time task is safe for measuring postural motor learning

• This type of task can efficaciously measure postural motor learning

• Measuring postural motor learning is important for recovery of postural control