Keeping in step with the young: Chronometric and kinematic data show intact procedural locomotor sequence learning in older adults.

Sequence learning in serial reaction time tasks (SRTT) is an established, lab-based experimental paradigm to study acquisition and transfer of skill based on the detection of predictable stimulus and motor response sequences. Sequence learning has been mainly studied in key presses using visual target stimuli and is demonstrated by better performance in predictable sequences than in random sequences. In this study, we investigated sequence learning in the context of more complex locomotor responses. To this end, we developed a novel goal-directed stepping SRTT with auditory target stimuli in order to subsequently assess the effect of aging on sequence learning in this task, expecting that age-related performance reductions in postural control might disturb the acquisition of the sequence. We used pressure-sensitive floor mats to characterise performance across ten blocks of trials. In Experiment 1, 22 young adults demonstrated successful acquisition of the sequence in terms of the time to step on the target mat and percent error and thus validated our new paradigm. In Experiment 2, in order to contrast performance improvements in the stepping SRTT between 27 young and 22 old adults, motion capture of the feet was combined with the floor mat system to delineate individual movement phases during stepping onto a target mat. The latencies of several postural events as well as other movement parameters of a step were assessed. We observed significant learning effects in the latency of step initiation, the time to step on the target mat, and motion parameters such as stepping amplitude and peak stepping velocity, as well as in percent error. The data showed general age-related slowing but no significant performance differences in procedural locomotor sequence learning between young and old adults. The older adults also had comparable conscious representations of the sequence of stimuli as the young adults. We conclude that sequence learning occurred in this locomotor learning task that is much more complex than typical finger-tapping sequence learning tasks, and that healthy older adults showed similar learning effects compared to young adults, suggesting intact locomotor sequence learning capabilities despite general slowing and normal age-related decline in sensorimotor function.

Introduction

The learning of structured sequences of keypresses as in the serial-reaction-time-task (SRTT) paradigm [1] is an established, lab-based paradigm to investigate the acquisition of motor and cognitive skills. The SRTT paradigm is one of the best approaches to illustrate procedural and implicit knowledge formed during the acquisition of a sensorimotor skill. In this experimental paradigm, individual manual key presses are required in response to a seemingly random sequence of visual or auditory cues. For example, four adjacent reaction buttons are spatially assigned to four stimulus positions on the computer screen. Performance with seemingly irregular but with actually predictable stimulus sequences is contrasted against performance with randomized, unpredictable series of stimulus sequences. Exposure to such sequences across several blocks results in a performance curve, for example in terms of decreasing response latency, demonstrating the gradually increasing ability to predict the upcoming sequence. Nissen and Bullemer [1] demonstrated that the production of the motor sequence typically becomes more automatic as learning progresses. In order to separate general practice-related performance improvements and sequence-specific learning, a block with purely randomized stimuli is often inserted in the experiment to uncover differences in performance between late blocks with a fixed sequence and the random block [“negative transfer”; see, e.g., 2–5 for reviews]. Any regularity contained within the stimulus sequence will result in improvements in response latencies and error rates beyond a performance level achieved by just practicing the response task itself. Instead, participants acquire the ability to predict upcoming stimuli and to anticipate any subsequent responses [1,6].

An important aspect in sequence learning of the SRTT is the question of whether participants require conscious awareness of the regularities that are embedded in a sequence [7]. Interestingly, participants are not necessarily able to report any hidden regularity as they learned the sequence without being aware of it [8] and the ability to report a sequence explicitly is not necessary for the learning of a sequence to occur [2,4,9, for reviews]. It has been demonstrated, however, that being consciously aware of the sequence facilitates its acquisition [e.g. 10]. In addition, multimodal contingent consequences of any responses during learning enhance acquisition of explicit knowledge about a stimulus sequence [11-13]. Haider and coworkers interpreted this enhancement by way of the ideomotor principle of action control as well as an increased sense of agency, which might trigger a causal inference process [11,13].

However, it is notable that the typically used motor task in sequence learning consisted of relatively simple actions such as finger key presses, which do not provide a rich sensorimotor context as observed in gross movements such as goal-directed upper-limb reaching or stepping with the limbs. As finger key presses are actions of relative short duration, they do not encompass the same amount of feedback-driven control during movement execution as in movements of body segments or the full body with much longer movement durations. Gross movement tasks on the other hand provide an abundance of movement degrees of freedom and thus multitude of possible solutions. Therefore, practice effects may express themselves on several potential parameters, e.g. spatiotemporal. By measuring not only the onset latencies but also movement times and the motion kinematics during sequence learning, however, it may be possible to better distinguish improved motor performance due the acquisition of a sequence from the optimisation of movements due to practice [14,15].

Our central goal of the present study consisting of two experiments was to evaluate the effects of older age on performance in a locomotor SRTT task, which required full body movements. The first experiment validated a novel locomotor sequence learning task. In the second experiment, using this novel, more complex locomotor task, we then assessed whether there are age-related differences in sequence acquisition. In locomotor sequences, the motor system must ensure that the body balance remains stable when a person takes one or more steps. The adaptation of stepping behaviour to dynamic environmental stimuli, especially if these occur in the form of sensory or mechanical perturbations that could destabilize body balance, is an important sensorimotor capability to avoid physical harm. The detection of regularities in environmental dynamics allows the preparation of postural adjustments to minimize the impact of any balance disturbance. This learning to predict the sequence of environmental stimuli is important to prevent an unintended destabilization of balance and a fall [16]. The complexity of postural and locomotor responses stems from the planning of sequences of several postural adjustments to ensure postural stability before any voluntary response can be executed. Anticipation and prediction play important roles in keeping body balance stable during standing or walking as disturbances of body balance, either self-imposed or externally caused, can be derived from context-specific knowledge gained through experience.

Sequence learning in a more complex locomotor task such as stepping has not been investigated in detail until now but experimental paradigms in terms of either postural or locomotor sequence learning have been developed in recent years. These studies indicated that sequence learning is likely to occur in more complex movement contexts. It seems, however, that the biomechanical constraints imposed on the body during full body movements, such as keeping body balance stable during stepping, may interfere with those processes of sequence learning as they are normally observed in manual SRTTs. For example, Du et al. [17] utilized a lower-limb foot-tapping SRTT, which more closely resembled the original manual SRTT. Du and Clark [18] concluded that the implicit sequence knowledge arose from statistical learning during foot-tapping sequence learning. This conclusion was based on the observation that responses to stimuli were only correlated with the single preceding stimulus or response, so that participants learned the probabilities by which a certain stimulus predicted a subsequent stimulus. Indications of chunking were also observed in the foot-tapping SRTT but were interpreted as masquerading biomechanical task constraints, such as improved strategies to control body weight shifts [18].

The study of locomotor sequence learning is important for understanding how older adults deals with such complex motor tasks. Older adults show an increasing fall risk [19]. Locomotor performance in older adults can inter-individually be restricted by many factors, such as reduced muscle strength and impaired cognitive and sensorimotor functions, which may lead to reductions in preferred gait speed, step and stride length, cadence, duration of the swing and single stance phases, and generally decreased joint range of motion, but also increases in postural stiffness, greater attentional demands, and stronger visual dependency [20]. Older adults with increased fall risk show longer stepping latencies in a choice stepping reaction time test by elongating anticipatory postural adjustments, thereby prioritizing dynamic stability over response time [21]. This observation corresponds to the “posture-first” principle commonly observed in older adults when confronted with an additional cognitive load in multitasking situations [i.e. 22]. Reduced APA amplitude and increased duration or absent APA components correlate with an increased fall risk in older adults [23,24]. In addition, Cohen et al. [25] demonstrated that step initiation in a reaction time task is delayed in older adults’ step initiations, possibly due to less efficient inhibition of erroneously response tendencies [26-28].

The effect of ageing does not influence motor skill acquisition in general but rather in a manner specific to the characteristics of the motor task to be learnt [29]. This applies to sequence learning too. Urry et al. [30] reported preserved SRTT performance in terms of manual response latency gains during sequence learning in older adults. Similarly in manual tasks, Bhakuni and Mutha [31] found no learning reductions in older adults, while Verneau et al. [32] observed ageing-related reductions for explicit learning of a sequence but not for implicit learning. In more general terms, no ageing effects were found for learning in SRTTs with relatively simple sequences [e.g. 33,34–37]. With more structurally complex sequences, however, learning in a SRTT declines with older age [e.g. 38,39–41]. Increases in complexity may be expressed not only in the sequences but also in the types of movements, so that when the required movements or actions become more complex, ageing effects become more likely too. For example, Hayes et al. [42] showed that older adults need more time to learn a sequence in a postural tracking task.

By moving away from a more cognitive focus on motor sequence learning involving simple manual button presses, we evaluated sequence learning in a locomotor activity as a function of ageing by recording performance in locomotor sequence learning. This included movement parameters such as latencies of step initiation and stepping on the target, but also step lengths, step velocities, and movement times in addition to traditional parameters (response latencies and error rates). Based on the extant evidence, we expected that participants would acquire knowledge of the regularities in the sequences and optimize their movement production in advance. This should be visible in performance curves and learning effects by means of reduced latencies and error rates. In addition, we expected that the performance limitations observed and the prioritization of postural stability during walking would limit the ability to acquire knowledge about specific locomotor sequences in older adults. According to this hypothesis, older adults should demonstrate a reduced ability to acquire a complex sequence of stimuli and motor responses. On the other hand, locomotor activities provide rich and meaningful state-dependent and contextual, multimodal sensorimotor cues, by which movement sequences could be discretized into specific sensorimotor events. This could facilitate the extraction of regularities in complex but predictable sequences. For example, reversals in movement direction may represent distinct sensorial signals indicative of regular sensorimotor state alterations useful for sequence segmentation and which older adults might benefit from.

Experiment 1

This experiment was planned to establish the novel methodology for the investigation of a locomotor SRTT demonstrating the basic locomotor sequence learning effects, and enable us to calculate an adequate statistical power for Experiment 2. Further, observation of participants’ behaviour allowed us consider suitable performance parameters for the evaluations of age-related changes in locomotor sequence learning.

Methods

Participants

The age inclusion criterion was an age between 18 and 35 years. We recruited 22 healthy younger adults (18 females and 4 males; average age = 22.0 +/- 3.3 years) from the RWTH Aachen University. The research ethics review board of the medical faculty granted ethical approval for this study (EK 322/19). All participants provided written informed consent before inclusion in the study according to the Declaration of Helsinki.

Stimuli and task

The administration of the experimental stimuli was programmed in PsychoPy2 [43]. The target stimuli 1, 2, 3, and 4 (1 and 3 were target fields for the left foot; 2 and 4 for the right foot) were presented auditorily as verbal number heard stimuli spoken by a male voice via stereo headphones (“eins”, “zwei”, “drei”, “vier” in German; JVC stereo headphones HA-S31M) and presented with a stimulus duration of approximately 470 ms and at a sound pressure level of about 65 dB. Three different types of steps were possible as a single response to an auditory stimulus: a forward step, a backward step, a step with the same foot on the spot. A step on the spot did not occur as a direct stimulus repetition but could be the result of an interspersed step of the other foot, for example such as a right-left-right triple (2-3-2). In response to the presentation of a single stimulus, participants were instructed to place their foot onto one of four numbered target fields (pressure-sensitive mats) as fast as possible following stimulus presentation. The accuracy of stepping onto the correct target was emphasized, while it was declared unimportant, whether or not a target was hit with the full foot.

In contrast to a manual SRTT, where response keys are typically released immediately after having been pressed, this locomotor SRTT did not require participants to return to a default, neutral placement of the feet after each single step. Instead, they were instructed to remain in the current standing position. The maximum response latency for a trial was 3 s with an inter-trial-interval following a response of 1 s. Responses had to be performed within the normal trial period or a trial was registered as incorrect. Following an incorrect stepping response, an “oops” sound was played as error feedback. The error feedback duration was 400 ms with an additional 100 ms of silence. The target of an actual step had to be coded “online” manually by the experimenter during the response-stimulus interval. This was necessary due to the high sensitivity of the pressure mats, which would detect weight shifts onto the standing leg also and would register these events as steps by standing leg additionally to the actual step.

The four numerical stimuli were arranged in sequences of 12 numbers so that two different second-order conditional sequences [44] were employed counterbalanced across participants (sequence 1: 2-4-1-2-3-2-1-4-3-1-3-4, sequence 2: 2-3-4-1-2-4-2-1-3-1-4-3). These sequences had been used in previous experiments [10,45]. Immediate stimulus repetitions were excluded in both sequences and both were comparable regarding the number of reversal patterns with each sequence (e.g. 2-3-2, 3-1-3, 2-4-2, 1-3-1). Randomized sequences were subject to the following constraints. Each number occurred with the same frequency. Further, randomized sequences did not contain any repetitions of identical elements (e.g. 2–2). Ascending or descending full runs (e.g. 1-2-3-4 or 4-3-2-1) and partial runs (e.g. 1-2-3-2-1 or 4-3-2-3-4) were possible.

Each sequence was presented 6 times per block, so that for a single block 72 individual trials were tested. In total 10 blocks were assessed. While the sequence followed a fixed pattern of the respective assigned sequence in the Blocks 3 to 8 and the 10th block, randomized stimulus sequences of digits were presented in the first, second and 9th block. For these three blocks, a randomized sequence was generated each, so that every participant received a unique set of three randomized blocks. Participants performed a practice block of 12 random stimuli at the start of the experiment to familiarize themselves with the experimental procedure. After data collection had been completed, a structured follow-up interview was conducted with each participant to ask whether they had detected any regularities on the sequence of target stimuli or whether they could even recall parts of or their entire sequence explicitly. The questioning began with asking for anything remarkable about the experiment and specific details, if a participant had to report an observation. In the next step, participants were asked whether they had detected any regularities and for further details if they affirmed this. If a participant mentioned the impression of a sequence of targets or steps, then the participant was asked to recall any elements of the sequence. A reported partial sequence was only scored as accurate, if it consisted of at least 4 elements in their correct order.

Experimental setup and procedure

Four target fields (pressure-sensitive mats; 23 x 16 cm) were placed on the lab floor arranged in a 2 x 2 squared configuration (distances between mats: 27 cm in anteroposterior direction, 18 cm in mediolateral direction; stepping area: 73 x 50 cm). Participants were instructed to step onto these target fields with both feet according to the sequence as given by the auditory stimuli. Fig 1 shows the square configuration of four pressure-sensitive mats. The pressure-sensitive mats provided response latencies using a customized response-acquisition system (1000 Hz). The mats on participants’ left side were marked with an odd number (1, 3), while the mats on the right were marked with an even number (2,4). This arrangement was compatible with the “markedness of response codes” (MARC) effect [46]. Participants’ starting position at the beginning of a block was with the left foot on mat 1 and with the right foot in mat 2.

Fig 1

Experimental setup of the pressure-sensitive target mats used in Experiments 1 and 2.

The spatial layout and configuration was identical in both experiments. Reflective markers for the optoelectronic motion capture system were placed on participants’ feet and lower extremities in Experiment 2 only.

Design and outcome parameters

Predictability of the stimulus sequence (random vs. predictable sequence) was the independent within-subject variable. By means of the pressure-sensitive mats, participants’ percent error and the duration when the pressure-sensitive target mat (T-TARGET) was triggered by a step were extracted from any stepping responses detected. In order to determine any sequence-specific learning effects within participants, the extent of the increase in response times and error rates from Block 8 to Block 9 and the decrease from Block 9 to 10 was interpreted as a measure of how well the respective sequence was internalized. Thus, the sequence-specific learning effect was determined as the difference between the average of Blocks 8 and Blocks 10 and performance in the random Block 9.

Statistical analysis

The first trial of each block was excluded from analysis. In addition, for the calculation of the average response latencies, both any error trials and correct trials following an error trial were excluded to avoid the influence of post-error slowing on latencies. All statistical computations were performed in R Studio 1.1.456. Paired t-tests were applied to test for a significant change in the observed learning effects against a hypothetical learning effect of 0. An alpha level of 5% was used to determine statistical significance. In order to determine a group average learning rate, we fitted an exponential decreasing non-linear regression function to the averaged performance curves across those blocks with a non-random sequence (Blocks 3 to 8 and Block 10) from which we obtained the function parameters A (intercept), B (time constant) and C (asymptote). For a considerable number of individual participants fitting the exponential function did not provide adequate fits (explained variance < 0.75%) so that the function fitting was conducted on a group level only.

In order to investigate if the sensorimotor complexity of a specific step interacted with learning effect, each step was classified in terms of its direction (forward step, backward step, step with the same foot on the spot) and whether a step was performed with the same or different leg with respect to the stepping leg in the previous trial. Both target sequences did not allow a step on the spot, when the previous step was performed with the same foot. Thus, instead of six different step types, only five step types could occur in each sequence. In an ad hoc ordering, (1) stepping backwards with a different foot than previously was considered the most complex as it required to shift body weight to the other foot and take a step without visual feedback of the stepping target. On a continuum between these two ends from more to less complex, we ordered (2) stepping on the spot following a different foot, (3) stepping forward following a different foot, and (4) stepping backward with the same foot as the previous step. Finally, we considered (5) stepping forward with the same leg as in the previous trial the least complex. For the learning effects of both primary outcome parameters, stepping complexity was included in separate ANOVAs as a 5-level within-subject factor and age group as between-subject factor. Bonferroni-adjusted single comparisons were performed between step complexity conditions, where necessary.

Results and discussion

Of the 22 participants 17 (77%) reported the subjective impression that some regularities in the sequence were present but only three (14%) individuals were able to report a correct partial sequence of at least 4 elements with a maximum of 9 elements. Table 1 summarizes the descriptive statistics and statistical results for the paired t-tests and the parameters for the exponential learning curves across the 7 blocks containing repeated sequences for Experiment 1.

Table 1

Descriptive statistics, sequence-specific learning effects and performance curve parameters for the participants in Experiment 1.

	Mean Blocks 8 and 10 (SD)	Mean Block 9 (SD)	Learning effect (AV, SD)	Learning effect statistical comparisons	Intercept	Asymptote	Time constant
	Young (n = 22)	Young (n = 22)	Young (n = 22)	All vs. 0 (df = 21)	Young (n = 22)	Young (n = 22)	Young (n = 22)
PE (%)	1.83 (1.33)	4.23 (2.62)	2.40 (2.29)	t = 4.92, p < 0.001 dz = 1.05	3.39	1.68	0.38
T-TARGET (ms)	892 (142)	954 (152)	56 (57)	t = 4.58, p < 0.001 dz = 0.98	943	897	0.94

PE: Percent error.

The response latency in terms of the time point after stimulus onset, at which a step triggered the pressure-sensitive target mat (T-TARGET), showed a strong sequence-specific learning effect (t(21) = 4.58, p < 0.001, dz = 0.98). A strong learning effect was also found for the percent errors (PE) (t(21) = 4.92, p < 0.001, dz = 1.05). Fig 2A and 2B show the average performance of all participants across the 10 blocks.

Fig 2

Performance curves of the (A) stepping response latencies and (B) percent error. In the Blocks 1, 2 and 9, the sequence of stimuli was random, while in the remaining Blocks 3 to 8 and 10 the respective target sequence was presented 6 times. T-TARGET: time point step onto pressure-sensitive mat. Error bars represent 95% confidence intervals.

Step complexity affected the learning effect in terms for percent error (F(1,4) = 4.78, p = 0.002, partial eta^2 = 0.19). Post-hoc single comparisons indicated that stepping forward with a different foot (lowest learning effect: mean = -1.1%, SD 2.2) showed the opposite tendency to stepping backwards with a different foot (greatest learning effect: mean = 4.4%, SD 5.1; p < 0.001) and stepping backwards with the same foot (mean = 2.7%, SD 4.0; p = 0.004), both conditions of which showed positive learning effects. A similar, numerical tendency was observed for stepping on the spot with a different foot (mean = 2.9%, SD 5.5; p = 0.08). In contrast, T-TARGET was not was affected by step complexity (F(4,84) = 0.38, p = 0.82, partial eta^2 = 0.02). All step complexity conditions showed positive learning effects with the lowest effect when stepping forwards with the same foot (mean = 28.1 ms, SD 110.0) and the greatest learning effect when stepping backwards with the same foot (mean = 51.7 ms, SD 99.0).

The above results demonstrate that sequence learning is expressed in the context of a more complex task, such as goal-directed stepping. The strong effect sizes for both the response latency and the error rate indicate that the sensorimotor complexity of the motor task did not diminish the acquisition of a sequence. The statistical structure of the sequences used in our study, in terms of their sequential transitions, were the same as in Zirngibl and Koch [10]. Therefore, it appears as if the rate of individuals with a high degree of explicit knowledge of the sequences was considerably lower in this experiment compared to the data reported in Zirngibl and Koch [10], where 28% to 35% of the participants acquired a high degree of explicit knowledge of at least 5 correct elements depending on the response mode (verbal > manual). It is important to consider, however, that Zirngibl and Koch [10] used manual and vocal responses and therefore imposed different motor demands. It is possible, therefore, that the demands of the locomotor task imposed a processing load, which prevented the knowledge of the sequence to reach a conscious level.

In Experiment 2, we used our novel locomotor sequence task to investigate if sensorimotor and cognitive changes associated with normal ageing would interfere with sequence learning. In addition, we were interested in subdividing the stepping responses into its subcomponents in order to assess the earliest stage of movement execution at which a sequence-specific learning effect could be found. This would provide us further insights into the movement parameters that would be influenced selectively by the acquired sequence knowledge.

Experiment 2

The major aim of this experiment was to investigate age-related differences in locomotor sequence learning. The basic methodology of this experiment was the same as described in Experiment 1. In the following, we will describe how Experiment 2 extended the methodological approach of Experiment 1. Older adults were expected to demonstrate limitations in their stepping performance, which would restrict their acquisition of the specific sequences.

The age inclusion criterion for the younger adults was an age between 18 and 35 years and an age between 65 and 80 for the older adults. The experiment included 27 younger adults (13 females and 14 males; average age = 25.9 +/- 3.9 years, weight = 69.2 +/- 12; height = 173.1 +/- 10.1; Body Mass Index = 23.1 +/- 3.2) and 22 older adults (13 females and 9 males; average age = 72.8 +/- 4.8 years, weight = 74.3 +/- 11.6; height = 166.2 +/- 9.4; Body Mass Index = 26.8 +/- 2.9) who were recruited from the general public. Like Experiment 1, Experiment 2 was reviewed and approved by the ethics committee of the RWTH Aachen University (EK 322/19). All participants provided written informed consent before inclusion in the study according to the Declaration of Helsinki.

Sample sizes in more recent publications that studied aging effects on sequence learning in the upper limb (fingers, hand, arm), especially studies reporting reduced [47-49] or even enhanced [31] learning performance in older adults above 65 years, ranged from 15 up to 40 participants per age group. Therefore, we expected that a sample size of at least 22 participants per age group would provide sufficient statistical power to discover large age effects. Based on the recruited sample sizes, we performed a post-hoc power calculation [G*Power 3.1.9.7; 50] to determine the hypothetical power for finding a difference between the two groups in terms of their learning effects. This resulted in a statistical power of 0.77 for finding a large between-group effect size (dz = 0.8) at an alpha of .05.

All older adults completed a survey regarding information on health and lifestyle and were assessed with the Falls Efficacy Scale, the history of the fall and the german adaptation of the Activities-specific Balance Confidence scale (Mini Mental State Examination = 28.6 +/- 1.3; Falls Efficacy Scale = 18.2 +/- 2.5; Activities-specific Balance Confidence = 96.2 +/- 3.9). No participant reported any symptoms of acute pain during the testing session. Generally, older participants were recruited from a research database of the geriatric hospital hosting the gait lab. No participants were included, if a diagnosis of a neurological diseases such as apoplexy, Parkinson’s disease, multiple sclerosis, epilepsy, or a rheumatological or autoimmune disease with an acute attack and/or therapy with antibody therapy were reported. All participants had to show free movement of the upper and lower extremities, which had to be possible without pain (e.g. due to osteoarthritis) and no apparent cognitive deficit [≤ 24 points in Mini-Mental State Examination; 51]. As additional exclusion criteria, participants were not tested if they reported a severe visual impairment (e.g. acute glaucoma attack, blindness on both sides or an unstable, wet macular degeneration), a severe hearing impairment (e.g. severe hearing impairment, residual hearing impairment and deaf), or acute, exacerbated Chronic Obstructive Pulmonary Disease, uncontrolled cardiovascular disorders (e.g. acute cardiac decompensation with New York Heart Association stage 4, no recent heart attack), or dependency on walking aids such as rollators, walking sticks.

In addition to the pressure-sensitive mats used in Experiment 1, a motion capture system was included consisting of a 10 camera-based optoelectronic system (Qualisys Medical AB, Oqus 500+, Gothenburg, Sweden; 120 Hz). The presentation of the auditory stimuli was recorded as stereo signals in 2 additional analog channels of the motion capture system. This enabled segmentation of the motion capture recordings of a single block into the constituting individual stimulus trials.

In order to enable kinematic analysis of participants’ steps in terms of their 3D trajectories, passive, hyper-reflective markers were placed at anatomically predefined locations on the lower extremities as well as on the pelvis and upper extremities of all participants. Clusters of 4 markers were mounted on rigid plastic plates, which were placed on each segment of the lower extremities. The kinematic and kinetic data were processed in MATLAB (2019b, The MathWorks, Inc., Natick, USA). Kinematic data were spline interpolated from 120 Hz to 2400 Hz and subsequently merged with the kinetic data. All timeseries data were smoothed using a generic dual-pass 4th-order Butterworth lowpass filter (10 Hz cut-off).

The predictability of a stimulus sequence (random vs. predictable sequence) served as an independent within-subject variable and age group was included as an independent between-subject variable (younger adults vs. older adults). In addition to participants’ percent error, several types of response latencies were determined based on stepping responses detected in the two data modalities of the pressure-sensitive mats and the foot kinematics. (i) The earliest response event that was extracted was the initiation of a step detected by the motion onset of a foot marker (T-STEP; 3 SD velocity threshold above baseline). This was followed by the (ii) time point when the step onto the pressure-sensitive target mat (T-TARGET) was completed. In between these events, additional movement parameters such as the step amplitude (STEP-AMP), step duration (STEP-DUR) and the time point (T-PVEL) and magnitude of peak velocity (M-PVEL) of the step trajectory were extracted to better understand sequence learning-related adjustments in the control of the stepping movements. Fig 3 shows illustrative traces indicating the three major motion events in a single forward step of the left foot of a single participant.

Fig 3

Illustrative traces indicating the three major motion events in a single forward step of the (A) left foot of a single participant. During this step, the (B) right foot remains static, while the left foot is stepping forward. T-STEP: time point step onset; T-TARGET: time point step onto pressure-sensitive target mat; AP: anteroposterior.

As in Experiment 1, the first trial of each block was excluded from analysis. In addition, for all outcome parameters except percent error, error trials and the trials following an error trial were excluded to avoid the influence of post-error slowing on latencies and an alpha level of 5% was used to determine statistical significance. We used a 1-dimensional Chi^2 test with Yates correction to assess the significance of the frequencies of correctly reported partial sequences in the group of older adults based on expected frequencies derived from the frequencies observed in the group of younger adults. The critical Chi^2 value for significance was selected for testing the unidirectional alternative hypothesis that older adults would perform worse than the younger adults.

As it was possible that a participant performed more than one step in response to a single stimulus and as the pressure-sensitive mats registered a single step following a stimulus only, stepping sequences detected by the pressure-sensitive mats were manually compared to the stepping sequences detected by the motion capture system. All trials in which the detected steps could not be matched between the two data modalities based on runs of three consecutive steps were excluded from data analysis. Thus, 21% of all trials could not be assigned with certainty and therefore were excluded from the statistical analysis.

As in Experiment 1, paired t-tests were applied to test for a significant change in the observed learning effects within participants against a hypothetical learning effect of 0, while differences between the two age groups were also tested by calculation of two-sample t-tests. In addition, mixed repeated-measure analyses of variance (ANOVAs) with sequence condition as within-subject factor (average of sequence Blocks 8 and 10 vs. random Block 9) and age group as between-subject factor were calculated for the primary outcome parameters T-STEP, T-TARGET.

Linear regressions and Pearson correlations were calculated between all performance parameters and participants’ person specific characteristics in order to detect interdependencies and the specificity of learning effects. In order to protect against the problem of multiple testings, a conservative alpha level of 0.1% was applied for statistical significance. As in Experiment 1, for a considerable number of individual participants fitting the exponential function did not provide adequate fits (explained variance < 0.75%) so that the function fit was conducted on a group level.

In order to investigate if the sensorimotor complexity of a specific step interacted with age group or learning effect, each step was classified as described for Experiment 1. For the learning effects of the all outcome parameters, stepping complexity was included in separate mixed ANOVAs as a 5-level within-subject factor and age group as between-subject factor. Bonferroni-adjusted single comparisons were performed between step complexity conditions where necessary.

After being asked if they detected any regularities in the sequence of target stimuli or could recall parts or the entire sequence, 19 (71%) of the younger and 15 (68%) of the older participants declared that they felt that regularities in the sequence were present. The frequency of this subjective impression did not differ between age groups (Chi^2(1) = 0.06, p>0.1). Of the younger adults, 6 (22%) reported a partial sequence of at least 4 elements correctly, while 2 individuals (7%) were able to report the full sequence. Of the older adults, 2 individuals (9%) reported a correct partial sequence between 4 and 6 items only. Nevertheless, the relative proportion of conscious recall of the sequence was not different between both age groups (Chi^2(1) = 1.62, p>0.1). Also in terms of the average number of elements recalled, a difference between the young adults (mean = 1.68, SD 3.46) and the older adults (mean = 0.64, SD 1.71; t(41.19) = 1.39, p = 0.17) was not observed.

We observed general slowing and generally less accurate performance in the group of older participants but no group differences in sequence-specific learning. Fig 4 shows the performance curves across the 10 practice blocks of the experiment for the three major response stages and percent error as function of age group. Table 2 provides the function parameters for the performance curves across the 7 blocks with repeated sequences (Blocks 3 to 8 and 10). The function intercepts and asymptotes of the performance curves indicate general slowing in the older adults but function slopes were similar between the two age groups for the latency of stepping onto a target mat and the latency of step initiation. Overall, the function slopes indicate that just a single block of exposure resulted in at least 37% of the final performance improvements.

Fig 4

Performance curves of the (A and B) response latencies for the main stepping events and (C) percent error. In the Blocks 1, 2 and 9, the sequence of stimuli was random, while in the remaining Blocks 3 to 8 and 10 the respective target sequence was presented 6 times. T-STEP: time point step onset; T-TARGET: time point step onto pressure-sensitive mat. Error bars represent 95% confidence intervals.

Table 2

Performance curve parameters for the participants in Experiment 2.

	Intercept		Asymptote		Time constant
	Young (n = 27)	Old (n = 22)	Young (n = 27)	Old (n = 22)	Young (n = 27)	Old (n = 22)
PE (%)	5.38	6.34	3.99	3.12	0.67	0.94
T-TARGET (ms)	1116	1224	1038	1133	0.53	0.43
T-STEP (ms)	622	761	551	679	0.46	0.52

In the final three blocks (Blocks 8, 9, and 10), the latencies of a step onto the pressure-sensitive target mat (T-TARGET) showed a main effect of age group (F(1,47) = 8.27, p = 0.006, partial eta^2 = 0.15), with the young adults responding faster than the older adults, and a main effect of sequence condition (F(1,47) = 35.35, p<0.001, partial eta^2 = 0.43), with faster responses in the sequence blocks than in the random block. Clearly, no interaction between the two factors was present (F(1,47) = 0.0008, p = 0.98, partial eta^2<0.001). The same pattern occurred for the onset latency of a step (T-STEP; group effect: F(1,47) = 13.74, p<0.001, partial eta^2 = 0.23; sequence condition: F(1,47) = 21.42, p<0.001, partial eta^2 = 0.31; interaction: F(1,47) = 1.75, p = 0.19, partial eta^2 = 0.04) and the percent error (group effect: F(1,47) = 13.41, p<0.001, partial eta^2 = 0.22; sequence condition: F(1,47) = 4.56, p = 0.04, partial eta^2 = 0.09; interaction: F(1,47) = 0.71, p = 0.41, partial eta^2 = 0.01) with less errors in younger adults and when the stimuli were predictable. Fig 5 shows the response latencies for the three major response stages and percent error for the final three blocks as a function of age group and sequence condition.

Fig 5

Box plots for the (A and B) latencies for the main stepping events and (C) percent error as a function of age group and the sequence condition for the final three blocks (average of sequence Blocks 8 and 10 vs random Block 9). T-TARGET: time point step onto pressure-sensitive mat; T-STEP: time point step onset. The whiskers of the boxes indicate variability outside the upper and lower quartiles. Relative outliers may be plotted as individual points.

Performance reductions in the random Block 9 compared to the sequence Blocks 8 and 10 are clearly visible for the latency of stepping onto the target mat, the latency of step initiation, and the percent error (Figs 4 and 5). Table 3 summarizes the descriptive statistics and statistical results of the t-tests for all movement parameters extracted. All movement parameters except for the stepping duration showed significant sequence-specific learning effects (all t(48) > 2.07, all ps < 0.04, all dz > 0.30). Acquisition of the sequences resulted in reduced latencies of step initiation and stepping on the target mat, longer stepping amplitudes, higher peak velocities during the swing phase at an earlier time point, and an earlier peak velocity of the lateral CoP shift. No differences in sequence-specific learning effects between the age groups were found (all unsigned t(all df>40) < 1.54, all ps > 0.13, all d < 0.45; Table 3).

Table 3

Descriptive statistics, age comparison and sequence-specific learning effects in Experiment 2.

	Mean Blocks 8 and 10 (SD)		Mean Block 9 (SD)		Sequence-specific learning effect (AV, SD)		Sequence-specific learning effect statistical comparisons
	Young (n = 27)	Old (n = 22)	Young (n = 27)	Old (n = 22)	Young (n = 27)	Old (n = 22)	Young vs. old	All vs. 0 (df = 48)	Young vs. 0 (df = 26)	Old vs. 0 (df = 21)
PE (%)	1.70 (2.28)	5.33 (5.26)	1.89 (2.21)	6.26 (6.05)	0.56 (2.10)	1.30 (3.89)	t = -0.79p = 0.43d = 0.24	t = 2.07, p = 0.04 dz = 0.30	t = 1.39p = 0.18dz = 0.27	t = 1.56p = 0.13dz = 0.33
T-TARGET (ms)	1039 (113)	1136 (150)	1096 (98)	1195 (131)	58 (68)	58 (68)	t = -0.03p = 0.98d = 0.008	t = 6.05, p < 0.001 dz = 0.86	t = 4.42 p < 0.001 dz = 0.85	t = 4.04 p < 0.001 dz = 0.86
T-STEP (ms)	556(124)	687(116)	607 (112)	717 (112)	52(59)	29(63)	t = 1.31p = 0.20d = 0.38	t = 4.75, p < 0.001 dz = 0.68	t = 4.59 p < 0.001 dz = 0.88	t = 2.14 p = 0.04 dz = 0.46
STEP-AMP(mm)	404 (19)	379 (37)	336 (30)	323 (40)	-69 (30)	-58 (32)	t = -1.21p = 0.23d = 0.35	t = -14.54, p < 0.001 dz = 2.08	t = -12.11 p < 0.001 dz = 2.33	t = -8.52 p < 0.001 dz = 1.82
STEP-DUR (ms)	706 (129)	744 (128)	701 (143)	744 (126)	-6 (56)	2 (47)	t = -0.50p = 0.62d = 0.14	t = -0.31,p = 0.76dz = 0.04	t = -0.52p = 0.61dz = 0.10	t = 0.18p = 0.86dz = 0.04
T-PVEL (ms)	863 (149)	1023 (149)	921 (144)	1055 (124)	58 (68)	31 (77)	t = 1.32p = 0.20d = 0.38	t = 4.43, p < 0.001 dz = 0.63	t = 4.45 p < 0.001 dz = 0.86	t = 1.88p = 0.07dz = 0.40
M-PVEL (mm/s)	856 (201)	794 (136)	746 (134)	735 (142)	-113 (105)	-60 (129)	t = -1.54p = 0.13d = 0.45	t = -5.28, p < 0.001 dz = 0.75	t = -5.59 p < 0.001 dz = 1.08	t = -2.18 p = 0.04 dz = 0.46

Across the entire group of participants (n = 49), the sequence-specific learning effects in all temporal parameters showed strong correlations, while learning effects in movement parameters such as step duration, step amplitude and peak velocity during stepping were less strongly or uncorrelated (Table 4). Additional linear regression analyses between the sequence-specific learning effects in the two main movement parameters (T-STEP, T-TARGET) and individual characteristics of a person such as age, Body Mass Index, and the Mini Mental State Examination, Falls Efficacy Scale and Activities-specific Balance Confidence scores in the group of older adults did not show any correlations between learning effects and parameters (Table 5).

Table 4

Correlations between sequence-specific learning effects of all movement parameters for the group of all participants in Experiment 2.

	T-TARGET	T-STEP	STEP-DUR	STEP-AMP	T-PVEL	M-PVEL
	R, p
T-STEP	0.66, <0.001
STEP-DUR	0.37, 0.008	0.05, 0.74
STEP-AMP	-0.15, 0.30	-0.21, 0.14	-0.02, 0.87
T-PVEL	0.71, <0.001	0.88, <0.001	0.30, 0.03	-0.17, 0.24
M-PVEL	0.23, 0.12	0.16, 0.28	0.18, 0.23	0.11, 0.47	0.27, 0.06

Table 5

Correlations between sequence-specific learning effects and individual characteristics of the older adults in Experiment 2.

	Age		BMI		MMSE		FES		ABC-D
	R2 (%)	p	R2	p	R2	p	R2	p	R2	p
T-STEP (ms)	<0.001	0.99	0.02	0.95	10	0.14	4	0.35	1.8	0.56
T-TARGET (ms)	16	0.06	3	0.41	5	0.34	5	0.34	12	0.11

BMI: Body Mass Index; MMSE: Mini Mental State Examination; FES: Falls Efficacy Scale; ABC-D: German adaptation of Activities-specific Balance Confidence scale.

The statistical results for the outcome parameters regarding an effect of stepping complexity on the learning effects are presented in Table 6. In general, ANOVAs for all specific outcome parameters on the learning effects in terms of the difference between the random block 9 and the average of sequence blocks 8 and 10 did not find any interactions between age group and step complexity. T-STEP showed the tendency for an effect of complexity (F(4,188) = 2.24, p = 0.09, partial eta^2 = 0.05), while T-TARGET was affected by complexity (F(4,188) = 3.52, p = 0.01, partial eta^2 = 0.07). Neither of the two parameters showed an effect of age group (age group: both F(1,47)< = 2.52, both ps> = 0.12, both partial eta^2 = 0.05) or an interaction between age group and step complexity (both F(4,188)< = 1.21, both ps> = 0.31, both partial eta^2 = 0.03). Post-hoc single comparisons for T-TARGET indicated that the learning effect tended to differ only between stepping on the spot following a step of the other foot (lowest learning effect: mean = 33.9 ms, SD 126.1) and taking a backward step following a step with the same foot (greatest learning effect: mean = 100.7 ms, SD 131.6; p = 0.08). The learning effects in STEP-AMP, T-PVEL and M-PVEL were influenced by step complexity as well. For STEP-AMP, single comparisons indicated the stepping forward with the same foot resulted in the strongest learning effect compared to all other step complexities (mean = -53 mm, SD 105; all p < 0.01; Fig 6A). For the other stepping complexities, learning effects were not different from 0. Another pattern was observed for M-PVEL (Fig 6B). A learning effect when stepping forward with the same foot (mean = 188 mm/s, SD 418) as well as stepping backward with a different foot (mean = 155 mm/s, SD 259) meant an increase in the peak velocity. No learning effect was observed for stepping in the spot with a different foot (mean = -0.29 mm/s, SD 319), while stepping forward with a different foot (mean = -99.8 mm/s, SD 188) and stepping backward with the same foot (mean = -396 mm/s, SD 511) lead to reductions in peak velocity. Fig 6 summarizes the effect of stepping complexity on learning effect.

Table 6

Effect table for the effect of step complexity and age group on sequence learning effects.

	PE (%)	T-TARGET (ms)	T-STEP (ms)	STEP-AMP (mm)	STEP-DUR (ms)	T-PVEL (ms)	M-PVEL (mm/s)
Step complexity (dfn = 4, dfd = 188)	F = 1.25,p = 0.29,partial eta^2 = 0.03	F = 3.52, p = 0.01, partial eta^2 = 0.07	F = 2.24,p = 0.09,partial eta^2 = 0.05	F = 12.23, p < 0.001, partial eta^2 = 0.21	F = 2.50,p = 0.06,partial eta^2 = 0.05	F = 3.29, p = 0.02, partial eta^2 = 0.07	F = 20.89, p < 0.001, partial eta^2 = 0.31
Step complexity x age group (dfn = 4, dfd = 188)	F = 2.08,p = 0.11,partial eta^2 = 0.04	F = 1.21,p = 0.31,partial eta^2 = 0.03	F = 0.77,p = 0.52, partial eta^2 = 0.02	F = 0.18,p = 0.69,partial eta^2 = 0.004	F = 1.35,p = 0.26,partial eta^2 = 0.03	F = 0.35,p = 0.81,partial eta^2 = 0.007	F = 1.62,p = 0.2,partial eta^2 = 0.03

Fig 6

Box plots of the sequence learning effects for stepping amplitude (A; STEP-AMP) and average peak stepping velocity (B; M-PVEL) as a function of the direction of a step and the foot of the previous step. The whiskers of the boxes indicate variability outside the upper and lower quartiles. Relative outliers may be plotted as individual points.

The latencies for T-TARGET appeared slower for the young adults in Experiment 2 compared to the young adult participants in Experiment 1, even though the sequence and the task requirements were completely similar. Therefore, a mixed analysis of variance with experiment as between-subject factor and the series of all blocks as within-factor was computed, which confirmed a significant effect of block (F(9,423) = 16.03, p<0.001, partial eta^2 = 0.25) with respect to the generally observed decreasing performance curves and faster responses for the participants who took part in Experiment 1 (F(1,47) = 24.63, p<0.001, partial eta^2 = 0.34). An interaction between both factors was not found (F(9,423) = 1.39, p = 0.23, partial eta^2 = 0.03). The percent error did not differ between the groups of young adults in Experiments 1 and 2. Only an effect of block was found (F(9,423) = 8.30, p<0.001, partial eta^2 = 0.15; group: F(1,47) = 1.23, p = 0.27, partial eta^2 = 0.03; interaction: F(9,423) = 1.09, p = 0.37, partial eta^2 = 0.02). As a motion capture system was not available for Experiment 1, no other motion parameters could be compared between both groups. Hence, the young participants in Experiment 1 were generally faster than those in Experiment 2, but the important result is that the learning effects were very similar across the young adults in Experiment 1 and 2. The settings differed between the first and second experiment, but it does not seem as if the changed environmental context of the gait lab setting emphasized accuracy of stepping over speed in the younger adults in Experiment 2 as the two groups did not differ in terms of the progression of percent error. The young adults in Experiment 2 were slightly older and had been recruited from the general public, while the younger adults in Experiment 1 were recruited from a body of psychology students and therefore possibly more experienced in reaction-time paradigms. The more extensive experimental procedure in Experiment 2 might also have motivated the participants to adopt a more conservative trade-off between the response latencies and the spatial stepping accuracy according to Fitts law [longer latencies with less variable step positions; 52]. A direct comparison between Experiment 1 and 2 regarding the spatial variability of the stepping was not possible due to the aforementioned lack of a motion capture system in Experiment 1.

The results of Experiment 2 demonstrate that the sensorimotor complexity of stepping does not impair procedural learning of sequences in older adults compared to younger adults despite a general slowing in the response latencies and reduced accuracy in the older adults. In terms of the duration of stepping onto the pressure-sensitive target mat (T-TARGET), a backward step following a step with the same foot showed numerically the greatest learning effect in both Experiment 1 and 2. Although the interaction between age group and step complexity was not significant, the average learning effect appeared numerically the greatest in older adults (mean = 133.0 ms, SD 164.0) when stepping backwards with the same foot. That aging does not reduce the capacity for procedural learning of the sequences is also expressed by any of the recorded person-related characteristics, such as age, Body Mass Index, cognitive level, or balance scores, not predicting the sequence-specific learning effects in the older adults. The ability to anticipate the stimuli in a sequence and prepare responses did not only affect the preparation of the steps in terms of the response latencies, but also the characteristics of the executed steps themselves, such as that sequence learning led to earlier, faster and longer steps.

Finally, the fact that the correlations indicated that peak stepping velocity and stepping amplitude were independent factors, however, indicates that sequence knowledge is utilized on different levels of locomotor control of taking a step. This conclusion is also supported by different patterns of the influence of stepping complexity on the two parameters.

General discussion

We developed a step initiation and goal-directed stepping (locomotor) SRTT paradigm in order to test sequence learning under more complex demands in postural and motor control, such as proactive postural stabilization before step initiation. Our novel paradigm allowed us to delineate the effect of an advanced age (i.e. ≥ 65 yrs) on locomotor sequence learning as we expected that age-related performance decrements in balance control and locomotion would impair the acquisition of the regularities inherent in the task. In two experiments, we demonstrated successful motor sequence learning in our goal-directed stepping SRTT. We found clear learning effects with respect to the latencies of two stepping events: the moment of step initiation and the conclusion of the step by the triggering of the target mat. Further parameters of the stepping movement were also adapted in the course of sequence acquisition resulting in faster and longer steps with the predictable sequence, although stepping duration was not altered by learning.

Importantly, despite alterations in sensorimotor function that occur with aging (i.e. chronological age), our results demonstrated that aging did not seem to have an obvious influence on the acquisition of the stepping sequence in terms of any latencies, movement parameters or error rates. Although tending to be more variable in their performance, the older adults demonstrated a comparable amount of learning of the sequences nevertheless, which also was not associated with any clinical measures of balance control and confidence or general cognitive aptitude. This contrasts with previous observations that implicit sequence learning becomes more limited in older age, especially with more complex sequences [38]. Thus, it may be the case that locomotor sequence learning is less susceptible to the effects of aging than the learning of sequences in other action and effector contexts (e.g. manual key press sequences). Alternatively, it could also be possible that effects of aging on sequence learning are reflecting the possible expertise in finger sequence movements of younger adults compared to less experienced older adults. Of course, it needs to be considered as well that smaller differences in locomotor sequence learning may actually exist between younger and older adults but that our study had not sufficient statistical power and sensitivity to detect these differences.

We did find indications of general slowing in the older adults but this did not affect the learning itself. Although the older adults appeared to be less likely to acquire a conscious, explicit representation of the sequence of stimuli or responses, a statistically significant difference was not found. This observation contrasts with age-related differences in explicit sequence monitoring reported by Caljouw et al. [53] in a postural visuomotor sequence learning task involving body weight shifts. Therefore, it may be that the influence of age on the acquisition of explicit sequence knowledge may be specific to the type of movement task or that our experiment was not sufficiently sensitive to uncover an age effect in the acquired explicit knowledge. Likewise, more subtle differences between the two age groups in their learning could exist but were not detected due to insufficient statistical power to find small or medium effects.

Our findings indicate that the biomechanical and behavioural complexity of a locomotor task such as goal-directed stepping does not impede the acquisition of any regularities in a sequence of stimuli. Du and Clark [18] suggested that the aggregation of regularities in their foot-tapping SRTT resulted from the biomechanical constraints imposed by the task, for example the reoccurrence of target positions harder to reach with a foot tap. The relevance of biomechanical constraints in sequence learning tasks was also indicated for responses in manual sequence learning. Differences in motor chunking between young and older adults in a manual sequence production task was attributed to biomechanical factors, such as increased stiffness of finger joints in older adults, which could be misinterpreted as chunking due to systematically and repeatedly occurring delayed latencies of finger with stiffer joints [47]. In other words, biomechanical constraints in the context of a particular SRTT imposed on participants may hide the expression of sequence learning. Compared to the foot tapping paradigm in Du and Clark’s study [18], we consider our SRTT less demanding in biomechanical and balance terms as our participants had to perform individual forward or backward steps only without returning to a starting posture so that phases of potential postural instability while standing on a single leg were shorter.

The observations in our study suggest at least two separable paths how the gradually improving prediction of the sequence shaped participants’ stepping behaviour. The first path concerns the ability to anticipate a stimulus and corresponding response and thereby to prepare an adequate step in advance. Correlations between the learning effects in our performance measures indicated that response time differences between the predictable and unpredictable sequences in the final three blocks of trials were strongly associated, starting with the earliest response measures, the timepoint of the anticipatory postural weight shifts. The second path relates to the execution of the stepping movements in terms of their spatiotemporal characteristics. Thus, it seems reasonable to assume that our participants prepared not only for the initiation of a step but also prepared the actual execution of a step, so that the stepping duration remained constant but the velocity and distance increased.

Flanagan and colleagues demonstrated that the sensorimotor control system learns to predict the consequences of an action faster than it optimizes the motor control during the action [54-56]. Mayr [57] argued for a dissociation between the implicit learning of spatial and non-spatial regularities attended to during sequence learning. Koch and Hoffmann [58] concluded that chunking occurred on the basis of relations between events and Koch and Hoffmann [59] reported that the learning of spatial event sequences is the dominant factor in implicit perceptual and motor learning. Correspondingly, Willingham et al. [60] presented findings suggesting that in implicit motor sequence learning the response location seems to play a more important role than the specific sequence of effectors activated. Thus, also in the locomotor stepping sequence learning of our present experiments, the sequence of location of target mats may be an important cue in addition to sensory effects linked to the sequence of final stance postures or the sequence of activated lower extremity muscles.

We envision that our paradigm could be used as a clinical tool for diagnosing impaired functional neuroplasticity and learning in ill-health older adults. Future research should be conducted to assess if geriatric populations with mild or more severe cognitive impairments are still capable of the observed locomotor sequence learning. It should also be determined if the potential inability to acquire the sequences might be used as a diagnostic indicator of the onset of cognitive deficits in older age. Additionally, our task might serve as a physical activity exercise in still mobile geriatric populations by the implementation of motivating performance feedback. Adapted for follow-up therapy aimed at promoting neural plasticity, it may even become a future approach to stimulate experience-dependent neural plasticity to counter age-related functional limitations and neurodegeneration. Research results into neuroplastic alterations in the human brain that occur during implicit and explicit sensorimotor learning are promising [61]. Therefore, it should be investigated if a clinical application of locomotor sequence learning in the context of neurorehabilitation is warranted.

In conclusion, we assessed locomotor sequence learning in a goal-directed stepping task in younger and older adults and observed learning in several parameters of locomotor performance. We did not observe effects of older age to restrict learning of the regularities embedded in the sequences.

19 Oct 2021

Keeping in step with the young: chronometric and kinematic data show intact procedural locomotor sequence learning in older adults

PLOS ONE

Dear Dr. Johannsen,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

In addition to addressing all of the points raised by the two reviewers below, I would also like to raise the following:

a) While I agree with Reviewer 2 that the intro is well-written, it is my opinion that it would be beneficial to considerably reduce the introduction in length. For example, the SRTT paradigm is a common protocol and can be succinctly and clearly introduced in a sentence or two. Similarly, it would be advantageous to only introduce those concepts that are directly relevant to the current paper (e.g., different sequence representations, chunking, etc. were not assessed here and thus do no warrant such an elaborate introduction). A similar comment can be applied to the abstract. Not necessary here to mention the number of blocks or the specific block numbers used in the computation of learning.

b) It is unclear as to how results from  Experiment 1 were used for the power computation for Experiment 2. Experiment 1 did not assess age differences; and, ultimately, data from the available literature was used for the power comp (Howard & Howard). Regardless, the two groups then exceeded this desired sample size (27 and 22 in the two groups as opposed to 16 each). So there appears to be no data-driven justification for the precise sample in Experiment 2.

c) The following information (applicable to both experiments) is better fit for the data processing / reduction section: “.. the first trial of each block was excluded from analysis. In addition, for all outcome parameters except error proportions, error trials and the trials following an error trial were excluded to avoid the influence of post-error slowing on latencies”

d) If I understood correctly, the measure T-PWS was not valid for mats 3 and 4. Is this correct? If yes, this no longer seems like a useful dependent measure given that the large number of trials that were discarded.

e) For experiment 2, it would be advantageous to indicate whether the sequence specific learning effects (middle columns; Table 3) are significant WITHIN each age group. The last column collapses across the two groups and shows significant learning effects, but this appears to benefit from the large n (collapsed across young and old) and effects that are disproportionately larger in young (although not sig. different from older). In brief, it does not appear that the older adults exhibit significant learning per se on many metrics. This information would be of interest. Note that this does not take away from the main between-group comparisons, but would be worth specifying nonetheless.

f) I leave this up to the authors but I am not convinced the methods of the two experiments can not be presented within a single methods section followed by results of the two experiments.

Please submit your revised manuscript by Dec 03 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Bradley R. King

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match.

When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.

3. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“We appreciate the support received by Joeline Schulz and Johannes Quandel during data acquisition and data post-processing. Funded by the Federal Ministry of Education and Research (BMBF) and the Ministry of Culture and Science of the German State of North Rhine-Westphalia (MKW) under the Excellence Strategy of the Federal Government and the Länder. In addition, this study was supported by the Robert Bosch Stiftung in the context of the “Lehrstuhl für Geriatrie an der Medizinischen Fakultät der RWTH Aachen” (32.5.1140.0009.0).”

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Additional Editor Comments (if provided):

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Keeping in step with the young: chronometric and kinematic data show intact procedural locomotor sequence learning in older adults

This study examined a novel locomotor variant of the serial reaction time task (SRTT) paradigm, in which young and older adults stepped on targets in response to auditory cues, and analyzed sequence learning and age effects across multiple kinematic measures. Overall, sequence learning was observed for young and older adults, and was not affected by age. Specifically, in Experiment 1, young adults showed lower error rates and reduced time to step into the target zone for sequence blocks relative to random blocks. In Experiment 2, both young and older adults showed sequence learning effects for error rate, time to initiate a step, time to step into the target zone, step amplitude, latency and magnitude of peak step velocity, and latency of peak center-of-pressure shift velocity. None of the learning effects in Experiment 2 were affected by age.

The novel task applied in this study provides a useful extension of the SRTT literature, and aging effects thereon, to a more ecologically valid movement condition. The experiment appears to have been conducted with rigor, and the primary findings appear convincing. My concerns and comments are detailed below.

MAJOR

1) The locomotor SRTT has three different step types (forward step, backward step, foot on same spot), two of which (forward, backward) could have followed a step by the same foot, or by the opposite foot. It seems that these five conditions (forward/same foot, forward/different foot, backward/same foot, backward/different foot, same spot) present different postural and balance concerns and difficulty, which could result in different challenges to older adults. The finger movement SRTT literature demonstrates aging effects increase for higher-order sequence information and other forms of increased complexity (e.g., Howard et al., 2004, Psychology and Aging; Bennett et al., 2007, Journals of Gerontology - Series B); it may be then that in a locomotor SRTT, aging effects become apparent for the more difficult/complex movements. Did step type have any impact on sequence learning, or interact with age effects on sequence learning?

MINOR

2) Line 105: The quick mention of generalization feels out of place in the paragraph on chunking; I would recommend separating these concepts, and perhaps expanding on the generalization point.

3) Line 257: What is meant by “Each participant received an individualized set of three randomized sequences”? I presume that this means the sequences used had the same archetypal format across participants (i.e., A-C-D-B-A-C), but that the mapping of [1,2,3,4] to [A,B,C,D] is what was randomized three times for each participant. If this is instead referring to a loop within the random condition, that should be clarified, and potential predictability driven by that loop would need to be addressed in all analyses. Could you please clarify this point?

4) Line 309-310: This sentence is unclear; I believe it is missing a phrase (“In addition, for response latencies but not ???, error trials and…”)

5) Table 1: It would be good to include the random condition means (Block 9) in this Table for direct comparison with the sequence condition means. Also, assumedly EP means error proportion, but that is not specified- and if the units are % it cannot be a proportion.

6) Line 381: It is good that participants with severe hearing impairments were excluded, but what about non-severe aging-related hearing loss? Given that the cues were auditory, even moderate presbycusis could cause additional cue processing time for older adults that would affect responses in a manner unrelated to movement planning. Were audiometric thresholds obtained for older adults, or was hearing assessed in any other objective manner?

7) Figure 3: What do the red and blue traces in the CoP position time series plots differentiate? I am assuming ML and AP, but it should be labelled. Also, perhaps this is just an aspect of the coordinate system used for measurements, but if this Figure illustrates a forward step by the left foot, why are left foot displacement and velocity negative?

8) Line 471: Though the Chi-squared test did not reach significance, the descriptive statistics suggest an aging effect here. Moreover, for older adults the full range of number-of-elements-reported is given, whereas only the lower bound is mentioned for young adults. Did older and young adults differ on number of sequence elements correctly reported?

9) Line 496: After stating the lack of age x sequence condition interaction, it would be helpful to report the lower bound of the observed p-value (e.g., p > 0.2). This comment applies to all such statements of null effects intended to be meaningful in the Results sections.

10) Line 533: What was the reasoning for considering the T-PWS, T-STEP, and T-TARGET measures to be the main movement parameters? If these are the parameters of primary interest, why were multiple other measures included in analyses too?

11) Line 550: The comma between “participants” and “who” disrupts the meaning of the sentence. There are multiple odd commas like this throughout the manuscript- these should be identified and removed.

12) Line 553: Would be good to remind here that the reason T-STEP and T-PWS are not mentioned is that those measures were not possible in Experiment 1. You mention this later (line 580), but it would be useful here.

13) Line 562: I don’t recall seeing the T-PWS sequence effect being reported as marginal in the Results. I see in Table 3 the p-value for that comparison was 0.17, which is a stretch for including as marginal- the authors do hint here though that the effect was more reliable in young adults alone, which is not reported. Given the lack of age x sequence interaction though, such a breakdown would not be a licensed comparison. This claim should either be better defended, or removed.

14) Line 601: The given interpretation that locomotor sequence learning may be less susceptible to aging decline than finger movement sequence learning is certainly reasonable. An alternate interpretation the authors may want to consider is that the aging effects reported in the finger movement literature may instead reflect a relative expertise of young adults for finger sequence movements, rather than an aging-related decline, and that this aging-related difference in expertise is not observed for stepping movements.

Reviewer #2: The authors present a study that examined the effect of aging on locomotor sequence learning. The study views this topic from a different context by assessing locomotion and asks an interesting and important question. The introduction is well-written and describes the current literature and motivation of the for study well. However, the Results and Discussion sections are a bit confusing to understand and the statistical methods and results are not described in adequate detail. The following comments may help address these issues:

1. It is unclear whether accuracy was emphasized to participants. What instructions were participants provided about the speed and accuracy of their movements? This may affect how quickly and accurately they respond to the auditory stimuli.

2. Was a recall or recognition test used to assess whether participants were able to determine that there was a sequence in the stimuli? It seems so, as some results are described, but the procedure and questions asked should be described in the methods.

3. What were the results of the paired t-tests in Experiment 1 and the ANOVA in Experiment 2? The methods describe these analyses, but they are not reported in the results.

4. What is the purpose of the Pearson correlations? What part of the research question were they trying to answer?

5. As older adults may make more anticipatory postural adjustments, how were they accounted for with respect to the postural adjustments all participants made to get ready to take a step? Was there any difference in these augments between the older and young adults?

6. The authors statement that, “Surprisingly, despite decreased functional reserves and reduced locomotor stability that occurs with aging (i.e. chronological age), it had no obvious influence on the acquisition of the stepping sequence in terms of any latencies, movement parameters or error rates” (lines 594-596) is a bit confusion. It seems from Figure 4 that older adults were overall slower across all blocks and made more errors. Perhaps the ANOVA results would clarify this.

7. The conclusion seems a bit overstated and broad. For example, how can this task be used as a clinical tool? Without assessing the task on a clinical population, it is unclear which parameters would be valuable for assessing a clinical population, especially since the authors did not find any significant differences between the older and young adults.

Minor

8. This is a minor stylistic suggestion. There are many abbreviations used in the manuscript that disrupt the flow of the paper. Some abbreviations are defined, but never referred to again (e.g., AR in line 157) that can be taken out.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

28 Jan 2022

Dear Dr. King and Reviewers,

we kindly appreciate your time for editing and reviewing our submitted manuscript. We have considered your feedback and suggestions and have addressed them as best as possible. In the following overview, we have listed your single recommendations and our responses and corresponding changes made to the manuscript.

Editor comments

1. “While I agree with Reviewer 2 that the intro is well-written, it is my opinion that it would be beneficial to considerably reduce the introduction in length. For example, the SRTT paradigm is a common protocol and can be succinctly and clearly introduced in a sentence or two. Similarly, it would be advantageous to only introduce those concepts that are directly relevant to the current paper (e.g., different sequence representations, chunking, etc. were not assessed here and thus do no warrant such an elaborate introduction).”

Response: the introduction was reduced in length by removal of those paragraphs relating to chunking and sequence representations including a short section referring to research conducted in monkeys.

2. “A similar comment can be applied to the abstract. Not necessary here to mention the number of blocks or the specific block numbers used in the computation of learning.”

Response: the abstract was slightly shortened by the removal of methodological information but also received an edited conclusion.

3. “It is unclear as to how results from Experiment 1 were used for the power computation for Experiment 2. Experiment 1 did not assess age differences; and, ultimately, data from the available literature was used for the power comp (Howard & Howard). Regardless, the two groups then exceeded this desired sample size (27 and 22 in the two groups as opposed to 16 each). So there appears to be no data-driven justification for the precise sample in Experiment 2.”

Response: the comment in the methods section of experiment 2 regarding power calculations for sample size estimation based on aging effects reported in the literature has be revised and reformulated. We still have to refer to the available literature but restrict our expectations to large effects only. This has also been considered in the general discussion section.

4. “The following information (applicable to both experiments) is better fit for the data processing / reduction section: “.. the first trial of each block was excluded from analysis. In addition, for all outcome parameters except error proportions, error trials and the trials following an error trial were excluded to avoid the influence of post-error slowing on latencies”

Response: these lines have been moved into methods sections of both experiments respectively.

5. “If I understood correctly, the measure T-PWS was not valid for mats 3 and 4. Is this correct? If yes, this no longer seems like a useful dependent measure given that the large number of trials that were discarded.”

Response: all mentioning of force plate recordings and derived preparatory weight shift outcome parameters has been removed from the manuscript.

6. “For experiment 2, it would be advantageous to indicate whether the sequence specific learning effects (middle columns; Table 3) are significant WITHIN each age group. The last column collapses across the two groups and shows significant learning effects, but this appears to benefit from the large n (collapsed across young and old) and effects that are disproportionately larger in young (although not sig. different from older). In brief, it does not appear that the older adults exhibit significant learning per se on many metrics. This information would be of interest. Note that this does not take away from the main between-group comparisons, but would be worth specifying nonetheless.”

Response: separate t-test for within-group significance of learning effects have been calculated and added to Table 3.

7. “I leave this up to the authors but I am not convinced the methods of the two experiments cannot be presented within a single methods section followed by results of the two experiments.”

Response: in an earlier draft of the manuscript before submission the methods sections in Experiments 1 and 2 were unified. However, the authors decided to split up the methods sections and prefer to keep this separation.

Additional requirements

8. “We note that the grant information you provided in the ‘Funding Information’ and ‘Financial Disclosure’ sections do not match. When you resubmit, please ensure that you provide the correct grant numbers for the awards you received for your study in the ‘Funding Information’ section.”

Response: the grant information in funding information and financial disclosure statements have been matched.

9. “We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.”

Response: the funding information has been removed from the acknowledgements section.

Reviewer 1 comments

10. “The locomotor SRTT has three different step types (forward step, backward step, foot on same spot), two of which (forward, backward) could have followed a step by the same foot, or by the opposite foot. It seems that these five conditions (forward/same foot, forward/different foot, backward/same foot, backward/different foot, same spot) present different postural and balance concerns and difficulty, which could result in different challenges to older adults. The finger movement SRTT literature demonstrates aging effects increase for higher-order sequence information and other forms of increased complexity (e.g., Howard et al., 2004, Psychology and Aging; Bennett et al., 2007, Journals of Gerontology - Series B); it may be then that in a locomotor SRTT, aging effects become apparent for the more difficult/complex movements. Did step type have any impact on sequence learning, or interact with age effects on sequence learning?”

Response: we preformed this suggested analysis and included it in the methods, results and discussion sections of Experiment 2 and the general discussion. Although, stepping complexity effects were found, we observed no interactions with age group.

11. “Line 105: The quick mention of generalization feels out of place in the paragraph on chunking; I would recommend separating these concepts, and perhaps expanding on the generalization point.”

Response: the paragraphs referring to chunking, generalization and representations have been removed from the introduction.

12. “Line 257: What is meant by “Each participant received an individualized set of three randomized sequences”? I presume that this means the sequences used had the same archetypal format across participants (i.e., A-C-D-B-A-C), but that the mapping of [1,2,3,4] to [A,B,C,D] is what was randomized three times for each participant. If this is instead referring to a loop within the random condition, that should be clarified, and potential predictability driven by that loop would need to be addressed in all analyses. Could you please clarify this point?”

Response: this statement has been rephrased to express that each participant received completely randomized blocks 1, 2, and 9. At the start of the experiment the random sequence of elements were generated anew for each participant.

13. “Line 309-310: This sentence is unclear; I believe it is missing a phrase (“In addition, for response latencies but not ???, error trials and…”)”

Response: this sentence has been rephrased to clarify its meaning.

14. “Table 1: It would be good to include the random condition means (Block 9) in this Table for direct comparison with the sequence condition means. Also, assumedly EP means error proportion, but that is not specified- and if the units are % it cannot be a proportion.”

Response: the random condition means move been included in the Tables 1 and 3. Also, “error proportions” have been rephrased as “percent error” in the entire manuscript.

15. “Line 381: It is good that participants with severe hearing impairments were excluded, but what about non-severe aging-related hearing loss? Given that the cues were auditory, even moderate presbycusis could cause additional cue processing time for older adults that would affect responses in a manner unrelated to movement planning. Were audiometric thresholds obtained for older adults, or was hearing assessed in any other objective manner?”

Response: the description of the recruitment procedure for the older adults in Experiment 2

has been reformulated. Although participants were contacted via a hospital database, any audiometric thresholds were not assessed. Nevertheless, all participants were able to understand the auditory target cues acoustically.

16. “Figure 3: What do the red and blue traces in the CoP position time series plots differentiate? I am assuming ML and AP, but it should be labelled. Also, perhaps this is just an aspect of the coordinate system used for measurements, but if this Figure illustrates a forward step by the left foot, why are left foot displacement and velocity negative?”

Response: the figure has been edited to that a forward foot displacement results in a flipped, positively increasing trajectory. The original figure showed the raw data in the kinematic coordinate frame, which had negative in the forward stepping direction.

17. “Line 471: Though the Chi-squared test did not reach significance, the descriptive statistics suggest an aging effect here. Moreover, for older adults the full range of number-of-elements-reported is given, whereas only the lower bound is mentioned for young adults. Did older and young adults differ on number of sequence elements correctly reported?”

Response: the average number of elements in a correct sequence was calculated and tested between both age groups. No differences were found. This was included in the results section.

18. “Line 496: After stating the lack of age x sequence condition interaction, it would be helpful to report the lower bound of the observed p-value (e.g., p > 0.2). This comment applies to all such statements of null effects intended to be meaningful in the Results sections.”

Response: for all observed non-significant p-values, the F or t statistics, p values and partial eta-squared have been included in the manuscript.

19. “Line 533: What was the reasoning for considering the T-PWS, T-STEP, and T-TARGET measures to be the main movement parameters? If these are the parameters of primary interest, why were multiple other measures included in analyses too?”

Response: an explanation was added to better explain the reasoning for the inclusion of secondary movement parameters.

20. “Line 550: The comma between “participants” and “who” disrupts the meaning of the sentence. There are multiple odd commas like this throughout the manuscript- these should be identified and removed.”

Response: we have probably overgeneralized german punctuation rules concerning relative clauses and inserted superfluous commas. We tried to identify and remove those.

21. “Line 553: Would be good to remind here that the reason T-STEP and T-PWS are not mentioned is that those measures were not possible in Experiment 1. You mention this later (line 580), but it would be useful here.”

Response: a reminder has been added that the T-STEP parameter etc. were not available in Experiment 1.

22. “Line 562: I don’t recall seeing the T-PWS sequence effect being reported as marginal in the Results. I see in Table 3 the p-value for that comparison was 0.17, which is a stretch for including as marginal- the authors do hint here though that the effect was more reliable in young adults alone, which is not reported. Given the lack of age x sequence interaction though, such a breakdown would not be a licensed comparison. This claim should either be better defended, or removed.”

Response: this statement has been removed.

23. “Line 601: The given interpretation that locomotor sequence learning may be less susceptible to aging decline than finger movement sequence learning is certainly reasonable. An alternate interpretation the authors may want to consider is that the aging effects reported in the finger movement literature may instead reflect a relative expertise of young adults for finger sequence movements, rather than an aging-related decline, and that this aging-related difference in expertise is not observed for stepping movements.”

Response: a statement has been added to the discussion which considers a possible specialization of younger adults in finger activities with respect to the interpretation of aging effects in the literature.

Reviewer 2 comments

24. “It is unclear whether accuracy was emphasized to participants. What instructions were participants provided about the speed and accuracy of their movements? This may affect how quickly and accurately they respond to the auditory stimuli.”

Response: in order to clarify the instructions regarding emphasis on speed or accuracy we moved this information to a more prominent position.

25. “Was a recall or recognition test used to assess whether participants were able to determine that there was a sequence in the stimuli? It seems so, as some results are described, but the procedure and questions asked should be described in the methods.”

Response: the description of the procedure for the recall and recognition has been rephrased with additional detail.

26. “What were the results of the paired t-tests in Experiment 1 and the ANOVA in Experiment 2? The methods describe these analyses, but they are not reported in the results.”

Response: we reformulated sentences in the methods and results sections to signpost the results of paired t-tests and ANOVA better in both experiments.

27. “What is the purpose of the Pearson correlations? What part of the research question were they trying to answer?”

Response: we added further explanations of the purpose of the Pearson correlations.

28. “As older adults may make more anticipatory postural adjustments, how were they accounted for with respect to the postural adjustments all participants made to get ready to take a step? Was there any difference in these augments between the older and young adults?”

Response: as all force plate measures have been removed from the manuscript a difference between old adults and young regarding APAs is no longer reasonable in the context of the experiments. It is an interesting question indeed but would need a follow-up study (currently in progress).

29. “The authors statement that, “Surprisingly, despite decreased functional reserves and reduced locomotor stability that occurs with aging (i.e. chronological age), it had no obvious influence on the acquisition of the stepping sequence in terms of any latencies, movement parameters or error rates” (lines 594-596) is a bit confusion. It seems from Figure 4 that older adults were overall slower across all blocks and made more errors. Perhaps the ANOVA results would clarify this.”

Response: in order to improve the clarity of this statement this section has been edited slightly.

30. “The conclusion seems a bit overstated and broad. For example, how can this task be used as a clinical tool? Without assessing the task on a clinical population, it is unclear which parameters would be valuable for assessing a clinical population, especially since the authors did not find any significant differences between the older and young adults.”

Response: a paragraph has been added to the general discussion section to suggest possible clinical applications of the paradigm.

31. “This is a minor stylistic suggestion. There are many abbreviations used in the manuscript that disrupt the flow of the paper. Some abbreviations are defined, but never referred to again (e.g., AR in line 157) that can be taken out.”

Response: we reduced the number of abbreviations.

Submitted filename: Johannsen_Bollheimer_responses_to_comments_b.docx

Click here for additional data file.

17 Mar 2022

24 Mar 2022

Dear Dr. King and Reviewers,

once again, we like to express our gratitude for editing and reviewing our submitted manuscript. We have taken the additional feedback and suggestions seriously and addressed them in the manuscript accordingly. In the following overview, we have listed each recommendation and our response and the changes made to the manuscript.

Reviewer 1 comments

1) The analysis breakdown by step complexity for Experiment 2 is welcome, and fully addresses my concern that more complex movements might reveal aging-related impacts on sequence learning. The non-age-dependent effects of step complexity on kinematic measures reported for Experiment 2 are also informative for future work that may use similar designs. However, it is unclear why the complexity analyses were only performed for Experiment 2; there is no obvious reason why similar analyses could not have also been performed on Experiment 1, which would give a sense of reliability to the complexity effects. This would also make sense conceptually, as Experiment 1 is introduced as a way to validate the approach used in Experiment 2, so it is unclear why an unvalidated additional analysis is introduced for Experiment 2.

Response: we have now performed and included the analysis of the effect of step complexity for Experiment 1. This change required alterations to the methods sections of both Experiments for the sake of consistency. Similarities between both experiments are discussed in the summary section of Experiment 2.

2) A power analysis based on data from Experiment 1 indicated that n = 9 was sufficient for detecting large within-group response latency effects, and previous work used ns between 15–40 to demonstrate age effects. However, these points are then used to justify a sample size of 22 as an a priori target for detecting large age effects across a number of dependent variables; it is not clear how those two data points support that conclusion. It seems that 22 was the minimum group n, and the power justification is more post-hoc than a priori. Wouldn’t it make more sense then to do a post-hoc power calculation instead and state what effect sizes you had power to detect?

Response: we have followed the reviewer’s suggestion and modified the paragraphs in the methods section of Experiment 2 in terms of a discussion of a post-hoc power analysis. In the General discussion section we are also now referring to the power issue.

3) For Experiment 2, why were ANOVAs used only for T-STEP and T-TARGET? Mixed ANOVA seems an appropriate statistical choice for all dependent variables analyzed here.

Response: thank you for pointing this out. This statement is clearly mistaken, it represented an earlier state of the draft, and has been corrected. Of course, mixed ANOVAs were also applied for the remaining dependent variables.

4) Figure 1: The caption here suggests that the reflective markers for the optoelectronic system were worn for both Experiment 1 and Experiment 2, but if I understand correctly the optoelectronic system was only employed for Experiment 2. If that is correct this caption needs to reflect that.

Response: we have extended the caption of this figure to state that the markers were only used for Experiment 2.

5) Line 352, what is meant by “I presentation of the auditory stimuli”?

Response: this was a typo and has been corrected.

6) Line 371, “was defined” seems extraneous.

Response: has been removed.

7) Line 382, “mediolateral” is not in Figure 3 and doesn’t need an acronym defined here.

Response: has been removed.

8) Lines 384 – 408, this first paragraph in the Experiment 2 Statistical Analysis section would benefit from being broken into 2-3 smaller paragraphs.

Response: this section has been broken down into smaller paragraphs.

9) Lines 476 – 477, it is claimed that Table 3 contains the t-test and ANOVA results for all DVs, but Table 3 only contains t-test results.

Response: has been corrected by removing the reference to any ANOVA results

10) Table 3: STEP-DUR t-test results for each group are not significant, but bolded.

Response: has been unbolded.

11) Table 4: Bolding is inconsistent in Table 4 too. Unless alpha in Table 4 was 0.001? If so, that should be specified somewhere.

Response: we have stated in the statistics section now, that a more conservative significance criterion was chosen.

12) Table 5: What do the italics indicate? Also, please define acronyms in Table caption.

Response: the italics only served a purpose during preparation of the manuscript and have been removed. Also, we explained the acronyms in the table.

13) Lines 600 – 606, Although the group differences in explicit knowledge did not reach

significance, the group means were consistent with the extant literature. So, I would not lean too hard on the current null finding as counter-evidence to prior demonstrations of age differences in explicit sequence knowledge during the SRTT.

Response: we have softened our conclusion accordingly and addressed the issue of statistical power here.

Submitted filename: Johannsen_Bollheimer_responses_to_comments_rev2.docx

Click here for additional data file.

28 Mar 2022

Keeping in step with the young: chronometric and kinematic data show intact procedural locomotor sequence learning in older adults

PONE-D-21-22593R2

Dear Dr. Johannsen,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Bradley R. King

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

25 Apr 2022

PONE-D-21-22593R2

Keeping in step with the young: chronometric and kinematic data show intact procedural locomotor sequence learning in older adults

Dear Dr. Johannsen:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Bradley R. King

Academic Editor

PLOS ONE