Task-dependent changes in the cortical control of postural dual tasks: effects of secondary cognitive and motor tasks

Abstract

This study investigated whether the motor cortical control underlying dual task performance depends on the type and difficulty of the dual task. Forty-four adults stood on a balance platform with or without performing a secondary motor (i.e., maintain a ball in the middle of a tray) or cognitive task (i.e., 2-back test) task. Balance task difficulty was manipulated by changing the resistance on the balance platform. Transcranial magnetic stimuli (TMS) were applied to elicit motor-evoked potentials to assess corticospinal excitability and short-interval intracortical inhibition (SICI) from the tibialis anterior. Results indicated that TMS-evoked measures differed based on the type of secondary task. When the motor task was involved, individuals responded with an increase in corticospinal excitability (32.2±5.4%; p<0.001) but no change in SICI compared to the balance-only task. In contrast, when the secondary cognitive task was added, both corticospinal excitability and SICI were unaltered compared to when balancing on its own. These task-specific changes occurred in tandem with differential changes in balance performance. Compared to the balance-only task, performance worsened by 34.2-41.9% (p<0.001) when dual tasking with the secondary motor task but was either unaffected (hard balance task) or improved by 9.1±4.0% (easy balance task; p=0.028) when dual tasking with the cognitive task. Increasing balance task difficulty resulted in 38.7±6.9% larger MEPs (p<0.001) and 21.9±9.4% reduced SICI (p<0.001) but were unaffected by dual-tasking. Together, these findings provide insight into how individuals adopt different motor control strategies, leading to differences in performance, depending on the secondary task.

Introduction

It is common in daily life to encounter situations when an individual must perform two tasks at the same time, otherwise known as dual tasking. The challenge with dual tasking is that if the cognitive or attentional requirements of the two tasks exceed the neural resources that are available, or if the motor commands to perform these tasks can only be processed one at a time, then performance on one or both tasks will be affected. However, the manner in which dual task performance is affected depends on several factors such as the individual’s abilities, the tasks being performed and the difficulty of the tasks. As such, when the dual task requires an individual to maintain balance at the same time as performing a secondary cognitive or visuo-motor task, a range of changes to balance performance have been observed. For example, compared to when a balance task is performed on its own, balance stability has been found to either decrease, not change, or even improve if either the balance or cognitive task is simple (1–4) but stability can disproportionately worsen if the dual task involves a more difficult balance task (2). Larger detriments in dual task performance are also observed when the secondary task involves a motor compared to a cognitive interference task (5). This differential dual task effect might be due to the secondary motor and cognitive tasks activating different areas of the brain for information processing (5). Consequently, this would cause differing amounts of neural interference, with more interference occurring when the motor cortical areas are used for both the balance and secondary tasks (5).

While much research has examined how dual tasking affects task performance, little is known of how the central nervous system, particularly at the motor cortical level, compensates for the increased demands associated with performing two tasks at the same time. Understanding these mechanisms under a dual task situation could be beneficial as previous studies have shown that a flexible modulation of cortical processes may be important for balance control (6–11). Using transcranial magnetic stimulation (TMS), increases in cortical and corticospinal excitability are commonly observed when an individual is freely standing compared to standing with support (6). This shift towards a more cortical control becomes amplified when the balance requirements are made more challenging by either altering the support surface or by removing sensory information. In these cases, individuals typically demonstrate an increase in corticospinal excitability along with a decrease in intracortical inhibition (7–11). Importantly, these neurophysiological measures appear to be functionally relevant since individuals with poor balance control (e.g., older adults) who demonstrate greater instability in response to more challenging balance tasks also exhibit greater reductions in intracortical inhibition during these more difficult tasks (7).

Examining these motor cortical measures under a dual task environment could provide additional insight into how individuals adapt to dual task scenarios and why dual tasking often leads to changes in task performance. In a meta-analysis of 11 studies, Corp et al. (12) found that dual tasking results in a significant decrease in the TMS-evoked silent period, a measure of gamma-aminobutyric acid B receptor (GABA_B)-related inhibition, as well as a smaller, albeit non-significant, increase in corticospinal excitability. Since this review, additional studies have reported on how involvement of the motor cortical areas is altered with dual tasking and have found a variety of neural adaptations with dual task performance. Depending on the age group examined and the tasks being performed, both corticospinal excitability and the duration of the silent period have been found to either not change, decrease or increase when performing a dual compared to a single task (13–15).

This wide range of effects of dual tasking on the motor cortical measures may be attributed to the variety of tasks and/or the difficulty level of the dual tasks being investigated (12, 16). Moreover, only one study has measured TMS-induced responses from the lower limbs (17) and none have incorporated a balance task within their dual task paradigm. Thus, it is currently unknown whether and how these previous findings might extend from single to dual tasks involving balance control. Furthermore, no study has applied a TMS-based paired-pulse paradigm to assess short-interval intracortical inhibition (SICI), a well-recognized parameter to assess gamma-aminobutyric acid A receptor (GABA_A) activity (18). The advantage of measuring SICI instead of the previously reported silent period is that SICI is believed to be a purely cortical phenomenon (19, 20). This contrasts with the silent period, which is influenced by both spinal and cortical mechanisms [for a detailed review, see (21)].

Therefore, the purpose of this study was to examine whether motor cortical control, as measured through corticospinal excitability and SICI, is altered when balancing while performing a concurrent secondary task. Various dual task conditions were assessed to parse whether this cortical control is dependent on a) the type of the secondary task (i.e., cognitive versus motor). and b) the difficulty of the primary balance task. Based on the previously mentioned TMS-studies showing increased corticospinal excitability and reduced SICI during more challenging postural tasks as well as the work of Corp et al. (12), it was hypothesized that compared to maintaining balance alone, dual tasking would lead to an increase in corticospinal excitability and a decrease in intracortical inhibition. Furthermore, it was hypothesized that a secondary motor task would challenge postural control to a greater extent than a secondary cognitive task and this would be reflected by a greater change in cortical measures.

Materials and Methods

Participants

Forty-four healthy young adults (23 males, 21 females), with a mean±one standard deviation (SD) age of 26.1±3.6 y, height of 175.7±8.8 cm and mass of 69.9±12.2 kg participated in this study. None of the participants reported any recent musculoskeletal, neurologic or orthopedic injuries that could affect their balance or mobility. All participants provided written informed consent prior to their participation in the study. All procedures were approved by the local research ethics boards and were in accordance with the Declaration of Helsinki.

Experimental set-up

Areas of skin over the tibialis anterior (TA) of the right leg were shaved and cleansed with alcohol before surface electrodes (Trigno with Quattro Sensors, Delsys, Natick, MA, USA) was placed overtop this muscle. The electromyographic (EMG) signal was amplified 1000 times and sampled at 4 kHz using commercially available hardware (Digital I/O system, National Instruments, Austin, TX, USA) and custom written Python-based software. EMG signals were filtered with a 20-450 Hz bandpass Butterworth filter prior to any further analyses.

A 95 mm focal butterfly-shaped TMS coil (D-B80) connected to a MagPro X100 stimulator (MagVenture, Farum, Denmark) was then placed over the area of the motor cortex representing the right TA. Although the plantar flexor muscles generate the torque necessary to maintain an upright posture, this study focused on the TA due to its sensory role in standing balance control (22), its susceptibility to age-related declines in balance (7), and its more pronounced and task-specific modulation of intracortical inhibition (23). The TA also has a lower threshold to TMS and therefore, does not require as high a stimulus intensity to elicit a motor-evoked potential (MEP) (23, 24). The coil was oriented to evoke posterior-anterior current flow in the interhemispheric fissure and each magnetic pulse was of a biphasic waveform (9, 25, 26). Although monophasic TMS may lead to stronger inhibitory effects, Wessel et al. (27) reported comparable effects between monophasic and biphasic waveforms on SICI. The optimal stimulation location was determined by moving the coil in ~1 cm intervals in the antero-posterior and medio-lateral directions until the largest MEP could be elicited in the TA with minimal stimulator intensity. Once the ideal location was determined, the coil was fixed to the participant’s head using a custom-built helmet to ensure that the coil position relative to the skull was consistent throughout the experiment (Fig. 1).

Fig. 1. Photo of the experimental setup.

The TMS coil is held in place by a helmet worn on the participant’s head. Participants held the wooden tray in their hands for all experimental conditions; during conditions with the secondary motor task, participants were required to maintain a small ball in the middle of this tray.

With the coil secured and the participant standing on the balance board at the “easy” setting (see below), active motor threshold (aMT) of the TA was determined by adjusting the TMS intensity until MEPs of at least 100 μV and that were discernible from the background EMG activity were observed in three of five consecutive trials (28). With the use of these criteria, the TA aMT across all participants occurred at 56.4±10.6 % of maximal stimulator output.

Experimental conditions

Participants performed the balance task alone, or the balance task with a secondary task together. As shown in Fig. 1, the balance task required participants to stand with their feet shoulder width apart on a custom-made balance platform that could rotate ±19° about a fixed axis (i.e., ankle plantar and dorsiflexion) (29). Participants were instructed to maintain the platform as horizontal and stable as possible throughout the duration of the trial. The difficulty of the balance task was adjusted by attaching springs with different spring constants (0.277 N/mm to 2.077 N/mm) to the underside of the platform, making it more or less stable. The “easy” setting involved the attachment of two 2.077 N/mm springs for all participants. The “hard” setting involved two springs between 0.277 and 1.119 N/mm, depending on each participant’s balance ability.

While on the balance platform, participants performed two types of secondary tasks. The 2-back test was used as the secondary cognitive task. This task required participants to watch a sequence of interchanging numbers, with each number displayed on a screen for 1 s, and verbally indicate whenever the current number matched the one shown from two steps earlier in the sequence. In contrast, the secondary motor task required participants to maintain a 10 mm diameter ball in the middle of a handheld circular wooden tray (44 cm diameter; Skala, IKEA). To ensure consistency across conditions, participants held the wooden tray, but without the ball, for all conditions that did not involve the secondary motor task.

Once participants felt comfortable with the balance, secondary motor and secondary cognitive tasks, participants completed a total of six experimental conditions, consisting of three tasks (i.e., balance only, balance with secondary motor, balance with secondary cognitive) across two balance difficulty levels (i.e., easy and hard). Two sets of trials were performed for each of the six conditions. The first set was considered the “performance trials” and the purpose of these trials was to measure isolated balance, motor and/or cognitive task performance. TMS was not applied during these performance trials to avoid performance measures being affected by the anticipation of receiving TMS or the involuntary muscle twitches due to the stimuli. Each of the performance trials was 30 s in duration and two performance trials were completed for each of the six conditions. Participants completed all performance trials in 8-10 min. Once all 12 performance trials (two trials per condition) were completed, participants completed the second set of trials. This set was considered the “TMS trials” and the purpose of these trials was to collect the TMS-related measures. Each TMS trial was approximately 2 min in duration and two TMS trials were completed for each of the six conditions. Participants generally completed all TMS trials within 30-45 min. The three tasks were presented in a random order between subjects, but the difficulty level was presented in a blocked order within each subject. Therefore, although all of the conditions for one balance difficulty level were performed together before switching to the other difficulty level, they were presented in a random order between subjects. The order of conditions was the same between the performance and TMS trials within a participant. Adequate rest was provided after each trial.

Throughout each of the TMS trials described above, participants received eight single (test) and eight paired (conditioned) pulse magnetic stimuli. For the single pulse stimuli, the stimulus intensity was set at 120% of the predetermined aMT. Each paired pulse stimulus involved an initial conditioning stimulus of 70% aMT, which was followed 2.5 ms later by a suprathreshold test stimulus of 120% aMT (28). Since two trials were completed for each experimental condition, participants experienced a total of 16 test and 16 conditioned stimuli per condition. All stimuli were presented in a random order, with a minimum of 4 s between stimuli. Further, each stimulus was only triggered when the board crossed 1º below the horizontal position at less than 40 º/s while the ankles were moving into dorsiflexion. The TMS pulses were only applied at this specific position and velocity to minimize the variability in background EMG activity, muscle length and muscle velocity between trials and conditions.

Data analysis

Task performance

The angular position of the balance platform was tracked with a goniometer (MP20, Megatron Elektronik, Putzbrunn, Germany) placed at the rotating axis. Data was collected at a sampling rate of 1 kHz and later low-pass filtered at 10 Hz. Balance performance was then quantified in two ways. First, the filtered position signal was differentiated and rectified to determine the average sway velocity. Second, the filtered sway position was downsampled to 100 Hz before calculating sample entropy (SampEn) (3). SampEn is the negative natural logarithm of the conditional probability that two time-series data sequences of the same length remain similar when another data point is added (30). Thus, SampEn assesses the regularity or predictability of a signal, with greater SampEn values indicating a more irregular and unpredictable signal while lower SampEn values indicating a higher degree of regularity or predictability (30). In the context of balance control, researchers have argued that changes in SampEn reflect the amount of attention directed to balance control, with greater SampEn values (i.e., more unpredictable

signals) signifying a shift of attention away from balance and thus, a more automatic control of balance (3, 31). In contrast, a decrease in SampEn is believed to reflect a shift towards more attention being directed toward controlling balance and thus, a less automatic control of balance (3, 31). All SampEn calculations on the filtered angular position signal were performed using publicly available MATLAB scripts (MathWorks Inc., MA, USA) using the formula presented in Rickman and Moorman (30): $\text{SampleEn}(\text{m},\text{r},\text{N})=-\text{log}\left(\frac{A(r)}{B(r)}\right)$, where m is the length of the sequences to be compared, r is the tolerance value for accepting matches, N is length of the data sequence, and A(r) and B(r) are the total number of concordant sequences according to the level of tolerance r in the appropriate dimensional space (i.e., m for B(r) and m + 1 for A(r)). Parameter values were set to m=2 and r=0.15 x the standard deviation of the data set.

Performance on the secondary cognitive task was quantified by the number of errors made on the 2-back task during each 30 s trial. In contrast, motor task performance was determined by tracking the movement of the ball using a video camera, recording at 100 Hz, placed overhead and above the participant. The positions of the ball and the midpoint of handheld tray were digitized at a sampling rate of 10 Hz, using Open Source software (Tracker, version 6.1.1, https://physlets.org/tracker). The position data of the ball was then differentiated to quantify the average velocity of the ball throughout each trial. Further, position error was determined as the average distance between the ball and tray midpoint (target) throughout each trial.

For all performance measures, the average from the two performance trials was calculated within each condition and used for further statistical analysis.

Motor evoked potentials

The peak-to-peak amplitude of each MEP from the TA was measured and the average amplitude in response to all single (test) and paired (conditioned) stimuli for each condition were determined. SICI was calculated as the percentage of inhibition from the test MEP amplitude, using the formula [1-(conditioned MEP/test MEP)] x 100%, such that larger SICI values indicate greater intracortical inhibition. The background EMG (bEMG) activity prior to each stimulus was determined as the root mean square of the filtered EMG signal during the 125 ms preceding each TMS pulse.

Statistical analysis

Data were first checked for normality, skewness and kurtosis; those that did not meet the criteria were log transformed prior to further analyses. Balance performance measures were analysed using separate two-way repeated measures analysis of variance (ANOVA), with secondary task (balance only, balance plus motor, balance plus cognitive) and balance difficulty level (easy, hard) as the repeating factors. Since the primary focus of this study was to examine changes in motor cortical control from single to dual tasking, post-hoc analyses parsing out significant main and interaction effects of task were conducted by calculating what is often referred to as the dual task cost or ratio (14). This involves normalizing the dual task data as a percentage or change from the corresponding single task (i.e., balance only) condition, and quantifies how much of the change in performance was due to the addition of the secondary task. The dual task cost was calculated as (dual task score minus the corresponding single task score) divided by single task score to reflect the percent change. Since a dual task cost of 0 indicates that there was no difference between single and dual tasking, dual task cost data were statistically analyzed using one-sample t-tests to assess whether the average dual task cost differed from 0.

Since the secondary motor and cognitive tasks were only performed for two of the experimental conditions, performance on these two tasks were analysed using a paired t-test and a Wilcoxon Signed Ranks test, respectively.

Similar to the balance performance measures, the bEMG data were also analyzed using a two-way repeated measures ANOVA, with secondary task (balance only, balance plus motor, balance plus cognitive) and balance difficulty level (easy, hard) as the repeating factors. This analysis revealed significant differences in bEMG between secondary task and balance difficulty conditions. Since changes in BEMG could potentially affect the size of the MEP and SICI responses (32–34), MEP and SICI data were analyzed using linear mixed-effects models (35) with secondary task, balance difficulty and their interaction as fixed effects. bEMG was included as a continuous coverate and mean-centered prior to analysis. Random intercepts for each participant were included to account for the repeated-measures design. In this way, the main effects and interactions of task and difficulty could be determined while controlling for any changes in bEMG between conditions. Lastly, dual task cost for MEP was also calculated in the same manner as the balance performance measures. In contrast, the dual task cost for SICI was calculated based on the absolute difference in SICI (i.e., dual task SICI minus corresponding single task SICI) rather than a percent change since there were several cases with near zero inhibition during the balance-only task that resulted in an abnormally large percentage change. Dual task costs were then analyzed using separate linear mixed-effects models, with the dual task cost as the outcome, mean-centered average bEMG as a covariate, and random intercepts for participant.

Significance for all tests was set at p≤0.05. Partial eta squared $\left({\eta }_p^2\right)$, Cohen’s d or Wilcoxon Effect Size (r) were used to report effect sizes. Statistical analyses were performed using SPSS (version 28.0.1.1; Chicago, IL, USA) or R (version 4.5.1; Vienna, Austria). Unless noted otherwise, all reported values are the means ± one standard error (SE).

Results

Task performance

When sway velocity was considered, a secondary task x balance difficulty interaction effect was observed (Fig. 2A). While the dual task cost for sway velocity was not different between the easy (12.5±4.5%) and hard (19.5±3.0%) difficulty levels when collapsed across the two secondary tasks (t₄₃=-1.42; p=0.162; Cohen’s d=0.22), post-hoc analyses indicated that there was a positive dual task cost (i.e., increased sway velocity or a reduced performance) when dual tasking with the secondary motor task (Fig. 2B). Specifically, performing the secondary motor task resulted in a 34.2±5.8% and 41.9±4.3% increase in sway velocity during the easy (t₄₃=5.91; p<0.001; Cohen’s d=0.89) and hard (t₄₃=9.71; p<0.001; Cohen’s d=1.46) balance difficulty levels, respectively, relative to the balance-only task. In contrast, when the secondary task comprised of the cognitive task, there was a negative dual task cost (-9.1±4.0%), indicating a decrease in sway velocity, during the easy balance task (t₄₃=-2.27; p=0.028; Cohen’s d=0.34) but no dual task cost (-3.0±2.5%), indicating no change in sway velocity, when the secondary cognitive task was performed with the hard balance task (t₄₃=-1.19; p=0.239; Cohen’s d=0.18) (Fig. 2B).

Fig. 2.

Mean ± 1 SE balance performance as quantified by (A) sway velocity (A) and (C) sample entropy. In panels B and D, the two balance performance measures are represented in terms of dual task (DT) cost. The white, grey and black bars and circles represent the balance only, balance with secondary motor task and balance with secondary cognitive task conditions, respectively. Asterisks (*) indicate a significant dual task cost (p<0.05).

SampEn in the sway position signal was influenced by a secondary task x balance difficulty interaction effect (F_2,86=4.71; p=0.011; ${\eta }_p^2=0.10$) (Fig. 2C). While SampEn values were generally greater, indicating a more irregular signal, during the easy compared to the hard balance task, SampEn was also affected by the type of secondary task. Post-hoc analyses examining the dual task cost revealed that when the balance task was at its easy difficulty level, there was a positive dual task cost when dual tasking with the motor task (16.4±5.1%; t₄₃=3.23; p=0.002; Cohen’s d=0.49) and with the cognitive task (21.5±6.5%; t₄₃=3.29; p=0.002; Cohen’s d=0.50) (Fig. 2D). This indicates that SampEn was greater, reflecting a more irregular sway position signal, during both dual task conditions compared to the balance-only condition. In contrast, when the balance task was set to its hard difficulty level, there was only a significant dual task cost (13.8±4.0%) when dual tasking with the cognitive task (t₄₃=3.49; p=0.001; Cohen’s d=0.53) (Fig. 2D). At this difficulty level, SampEn was not different between the balance-only and the dual task with secondary motor task (t₄₃=-0.22; p=0.830; Cohen’s d=0.03).

Performance on the secondary motor task was affected by the difficulty of the balance task. Despite the motor task itself not changing, motor task performance worsened, as reflected by an increase in velocity of the ball and an increase in the distance of the ball from the midpoint by 121.4±12.2% (t₄₃=12.75; p<0.001; Cohen’s d=1.34) and 63.1±9.3% (t₄₃=11.57; p<0.001; Cohen’s d=1.23), respectively, when the balance task was set to the hard compared to the easy balance difficulty level (Fig. 3). In contrast to the worsening of motor task performance with increasing balance task difficulty, cognitive task performance did not differ between the easy and hard balance difficulty levels. Over the 30 s trial, participants made 0.64±0.10 and 0.47±0.08 errors when performing the secondary cognitive task during the easy and hard balance difficulty level, respectively (Z=1.50; p=0.133; r=0.27).

Fig. 3.

Mean ± 1 SE motor task performance as quantified by (A) ball velocity and (B) distance of the ball from the target (midpoint). The white and black bars represent the easy balance with secondary motor task and the hard balance with secondary motor task conditions, respectively. Asterisks (*) indicate statistical significance (p<0.05).

Background EMG activity

The TA bEMG activity was influenced by a secondary task x balance difficulty interaction effect (F_2,86=4.71; p=0.011; ${\eta }_p^2=0.10$) (Fig. 4). Post-hoc analyses indicated that there was a positive dual task cost for the TA bEMG (i.e., increase in bEMG activity) when dual tasking with the motor task, but the dual task cost was negative (i.e., decrease in bEMG activity) when dual tasking with the cognitive task (Fig. 4). Specifically, performing the secondary motor task resulted in a 35.5±9.6% and 57.8±11.1% increase in TA bEMG during the easy (t₄₃=3.62; p<0.001; Cohen’s d=0.55) and hard (t₄₃=5.58; p<0.001; Cohen’s d=0.84) balance difficulty levels, respectively, relative to the balance-only task. When the secondary task involved the cognitive task, the TA bEMG was 11.1±5.1% and 22.1±4.0% smaller during the easy (t₄₃=-3.06; p=0.004; Cohen’s d=0.46) and hard (t₄₃=-5.23; p<0.001; Cohen’s d=0.79) balance difficulty levels, respectively, relative to the balance-only task.

Fig. 4.

Left, mean ± 1 SE tibialis anterior (TA) background EMG activity. Right, the background EMG measure is represented in terms of dual task (DT) cost. The white, grey and black bars and circles represent the balance only, balance with secondary motor task and balance with secondary cognitive task conditions, respectively. Asterisks (*) indicate a significant dual task cost (p<0.05).

TMS measures

Corticospinal excitability of the TA was influenced by main effects of secondary task (F_2,214.29=16.82; p<0.001; ${\eta }_p^2=0.14$) and balance difficulty (F_1,216.30=10.82; p<0.001; ${\eta }_p^2=0.048$) but not a secondary task x balance difficulty interaction effect (F_2,213.82=1.53; p=0.218; ${\eta }_p^2=0.01$) when analyzed with bEMG as a covariate (Fig. 5A). When post-hoc analyses were conducted to examine the main effect of task, it was found that regardless of balance task difficulty, dual tasking with the motor task resulted in a significant dual task cost for corticospinal excitability (32.2±5.4%; t_42.26=6.00; p<0.001; ${\eta }_p^2=0.46$) (Fig. 5B). In contrast, the dual task cost for corticospinal excitability when dual tasking with the cognitive task (-1.2±3.3%) was not significant (t_41.76=-0.36; p=0.722; ${\eta }_p^2=0.003$) (Fig. 5B). Lastly, when the main effect of difficulty was examined, 38.7±6.9% larger MEPs, indicating increased corticospinal excitability, were observed when balancing on the hard compared to the easy balance difficulty level. When only the dual tasks were considered, the dual task costs for MEP amplitudes were not different between the easy (20.7±6.6%) and hard (9.8±3.2%) difficulty levels (t₈₅=1.07; p=0.288; ${\eta }_p^2=0.013$) (Fig. 5B).

Fig. 5.

The mean ± 1 SE (A) MEP amplitude and (C) short-interval intracortical inhibition (SICI). In panels B and D, the two measures are represented in terms of dual task (DT) cost. The white, grey and black bars and circles represent the balance only, balance with secondary motor task and balance with secondary cognitive task conditions, respectively. Asterisks (*) indicate a significant dual task cost (p<0.05).

SICI was influenced by a main effect of balance difficulty (F_1,231.40=15.30; p<0.001; ${\eta }_p^2=0.06$), but not a main effect of secondary task (F_2,219.54=1.31; p=0.273; ${\eta }_p^2=0.01$) or a secondary task x balance difficulty interaction effect (F_2,216.28=0.53; p=0.590; ${\eta }_p^2=0.01$) when analyzed with bEMG as a covariate (Fig. 5C). The main effect of difficulty was due to a 21.9±9.4% reduction in SICI during the hard compared to the easy balance difficulty level. When calculating the difference from dual-tasking to balance-only, no differences in dual task cost were observed between the easy (0.4±2.7%) and hard (3.4±2.8%) balance difficulty levels (t₈₅=-0.88; p=0.383; ${\eta }_p^2=0.009$) (Fig. 5D).

Exploratory analyses were performed to investigate if the neurophysiological measures were functionally relevant by correlating dual task costs for MEPs and SICI with balance performance. Dual task cost for sway velocity was found to be correlated with the dual task cost for corticospinal excitability only when the balance task was at its easy difficulty setting and only when performing the secondary motor task. Specifically, when dual tasking resulted in a larger increase in sway velocity compared to the balance-only task, this was associated with a larger concomitant increase in MEP amplitude when the dual task involved the secondary motor task (r=0.328, p=0.030) (Fig. 6). No other relationships between balance performance and either MEPs or SICI were found to be significant.

Fig. 6.

A significant relationship was found between the dual task (DT) cost for sway velocity and single pulse MEP (i.e., corticospinal excitability) during the easy balance difficulty level with the secondary motor task.

Discussion

Few studies have examined how the motor cortical control is altered under dual task conditions. Not only have these studies reported variable effects, none has incorporated a balance task as part of their dual task paradigm. Therefore, the purpose of the current study was to examine whether and how these previous findings extend to dual tasks involving balance control.

The effect of secondary task type on dual task performance

The effects of dual tasking on balance performance are known to depend on various factors such as an individual’s cognitive and balance abilities as well as the difficulty and type of dual tasks being performed. The results of this study provide further evidence that dual tasking leads to variable effects, as the two secondary tasks examined in this study yielded opposing consequences to balance performance. When the dual task involved the secondary motor task, positive dual task costs were observed whereby balance performance worsened (i.e., increase in sway velocity) during the dual compared to the single task (balance only) condition. In contrast, when the secondary task consisted of the cognitive n-back test, the dual task cost was either zero or negative, indicating no change or an improvement in balance performance, from single to dual tasking depending on the difficulty of the balance task. An improvement or even a lack of change in balance performance when dual tasking might be considered unexpected given the greater amount of resources that would be needed to perform both the balance and cognitive tasks at the same time, but other studies have also observed a reduction in sway amplitude or variability when a cognitive task is performed while standing (1, 3, 4). As will be discussed next, this effect might be the result of task-dependent differences in the cortical control and/or automatic balance processes underlying dual task performance.

The effect of secondary task type on the cortical control of dual tasking

A novel finding of the current study is that similar to the differing effects of secondary task type on balance performance, the underlying motor cortical control was also dependent on the type of secondary task being performed. Specifically, when individuals performed the secondary motor task with the balance task together, this resulted in an increase in corticospinal excitability but without a concomitant change in SICI. This contrasts with what happened when dual task conditions involved the secondary cognitive task, where both corticospinal excitability and SICI were unaltered compared to when balancing on its own.

Several reasons might explain why individuals adopted different motor cortical strategies depending on the dual tasks being performed. One possibility is that the two secondary tasks activate different areas of the brain for information processing and consequently, differently affect task performance. This would support the findings of Beurskens et al. (5), who found that the differential dual task costs associated with a secondary motor compared to a cognitive task were in line with distinct electroencephalographic activation patterns, where greater activity in the beta frequencies arising from the frontal areas (FP_z and F_z) were observed when walking and performing a motor compared to a cognitive task. For the current study, the n-back task that comprised the secondary cognitive task has previously been shown to elicit activation of various brain regions such as the prefrontal cortex, supplementary motor area, premotor cortex and posterior parietal cortex (36). Thus, when the cognitive task is performed while maintaining balance, one might expect less overlap and competition for the same neural resources. This may have enabled individuals to maintain or downregulate the involvement of the motor cortex and rely instead on sub-cortical or more automatic processes for balance control (37, 38). In fact, such a possibility is supported by our SampEn results, where an increase in SampEn, indicating a more irregular and unpredictable sway position signal, was observed when dual tasking with the secondary cognitive task. Previous studies applying SampEn measures on balance control have also inferred that when SampEn increases during a balance task, this may reflect less attention toward balance and a more automatic control of balance (3, 31). Since relying on automatic processes to control balance would decrease the amount of interference that an individual experiences, this might also explain why a lesser dual task cost (i.e., an increase in dual task performance) was observed when dual tasking with the secondary cognitive task. In contrast, when the chosen secondary motor task was involved, this likely required activation of the primary and secondary motor areas as well as the coordination of different sources of visual information to keep the ball in the middle of the tray (5). Since these same cortical areas are also needed for maintaining balance, it is assumed that this would cause a greater interference effect. As a result, the observed increase in corticospinal excitability when dual tasking with the secondary motor task might reflect an overall facilitation of the motor cortex to control both tasks at the same time. However, since balance performance worsened when the secondary motor task was involved, the increase in corticospinal excitability might be considered to be insufficient to compensate for the increased corticomotor demands. In other words, the resources of the primary motor cortex for controlling two motor tasks at the same time are finite and reached their limit, resulting in reductions in motor performance.

In addition to each secondary task activating different areas of the brain, the individual’s focus of attention may have also differed between secondary tasks. When performing the cognitive task, it is possible that the lesser interference between tasks enabled individuals to direct their attentional focus away from the balance task, or that individuals chose to focus on the cognitive task (and away from the balance task) to minimize interference. Either strategy would be beneficial because it can lead to a lesser conscious movement processing for postural control (39), promote greater automaticity (3) and similar to what was observed in this study, improve balance performance. In addition, if individuals chose to adopt an external focus of attention (i.e., concentrating on the external environment, task goal and movement outcome) instead, this will have further fostered automatized motor control processes that rely less on cortical sources and at the same time, improve balance performance (40). An external focus of attention has been associated with an upregulation of intracortical inhibition (e.g., SICI and surround inhibition) to presumably increase the efficiency of movement execution (37, 41).

One might have expected that similar to the secondary cognitive task, performing the secondary motor task when maintaining balance would also lead to a more external focus of attention. This is particularly the case since participants were instructed to prioritize maintaining the ball in the middle of the handheld tray. However, since the amount of stability on the balance task can influence how well one can perform on the secondary motor task, it is possible that individuals chose to prioritize balance control, particularly when the balance task was more difficult. The lack of change in SampEn when dual tasking with the secondary motor task at the hard balance difficulty setting provides some indirect support that no change in attention focus occurred from the balance-only condition. Alternatively, even if individuals attempted to rely on automatized motor control involving subcortical structures, this would also face the same “limited resources” problem. Thus, when simultaneously performing two motor tasks that both rely on similar cortical and subcortical (e.g., cerebellum) brain areas, this might simply overuse the available resources at both sites, and prevent the interference of cortical brain regions important for motor control and motor performance in both tasks. As such, the observed increase in corticospinal excitability when dual tasking with the secondary motor task may have reflected a strategy whereby individuals had to more strongly activate the motor cortex because other resources were exhausted. Despite this greater activation, a loss of performance could not be prevented.

The effect of dual task difficulty on the underlying cortical control and task performance

Regardless of whether individuals were performing a single or dual task, there was an increase in corticospinal excitability and a decrease in intracortical inhibition with increasing balance task difficulty. These results are in line to previous studies that have examined how the cortical control is altered when performing more challenging balance tasks on its own (8, 42, 43). In these studies, it was speculated that the increase in corticospinal excitability and the removal of intracortical inhibition act to not only increase lower limb muscle activity but also to allow for a more cortical control of these muscles to ensure maintenance of balance (8, 43). Interestingly, increasing the difficulty of the balance task did not alter the dual task cost of motor cortical control. The non-significant dual task costs when performing either of the two secondary tasks indicates that any changes in motor cortical control that were observed from the easy to the hard balance-only tasks were not magnified when dual tasking. This finding is similar to what was found by Corp et al. (14), who observed no change in dual task cost for MEP amplitude and SP duration between two dual task difficulty levels. The authors attributed this due to the difference in difficulty between dual task conditions not being large enough to require a change in cortical control (14). It cannot be determined whether this explanation would apply to the results of this study. While the dual task cost for sway velocity did not differ between the easy and hard dual task difficulty levels, performance on the secondary motor task worsened during the hard compared to the easy balance task. This latter result might suggest that there was indeed greater dual task interference when maintaining balance and performing the motor task at the same time. To compensate however, individuals of this study may have mitigated some of the neural effects by focusing on the balance task at the expense of the secondary task. This might also explain why there was no significant dual task cost for SampEn, proposed to indicate the amount of attention placed on the balance task, when performing the secondary motor task while balancing on the hard difficulty level.

Functional relevance of neurophysiological measures with dual task performance

Several studies have reported on the functional relevance of the TMS-elicited measures to dual task performance but there is considerable discrepancy between studies. For example, when inhibition, as reflected by the duration of the TMS-evoked silent period (SP), was considered, one study did not observe a significant relationship between the amount of SP and performance (13), another study reported a poorer dual task performance was associated with a reduction in SP in older but not younger adults (44), with a third study found a decrease in dual task performance was correlated with an increase in SP (15). Lastly, a trend towards a significant relationship has also been observed, such that when the task was easy, an increase in dual task performance cost was associated with a smaller dual task cost for both MEP amplitude and the silent period (14).

In contrast to these previous studies, the current study found only one significant correlation between dual task performance and TMS measures. Specifically, when balancing at the easy difficulty setting, individuals demonstrating a larger increase in sway velocity due to the dual task involving the motor task were associated with a larger increase in MEP amplitude (r=0.328). Although this relationship was weak in strength, it nevertheless hints at a potential functional consequence of altering cortical processes for successful dual task performance when balance control is involved. Specifically, individuals who perform poorly require a stronger facilitation of the motor cortex, via a greater corticospinal excitability, to actively control the lower limb muscles for maintaining balance. Alternatively, it is possible that these results indicate that those who are unable to adequately downregulate the excitability of the motor cortex are more affected by dual tasking.

It is acknowledged that further work is needed to confirm our effects as it is unclear why a relationship was only observed during the easy difficulty level when performing the secondary motor task and why such large variations in the relationship between dual task performance and TMS measures exist between studies. One possibility is that since a range of primary tasks (e.g., pincer force, arm tracking, standing balance) of varying complexity have been examined, the effects are task-specific or difficulty-specific. It is also possible that the results are dependent on the TMS-evoked measure. Second, unlike the current study, previous studies have relied on the SP as a measure of cortical inhibition. While the SP is more easily obtained by applying suprathreshold TMS pulses in the active muscle, the outcome and underlying mechanisms are complex. Not only is the SP influenced by both GABA_B-mediated cortical inhibition and spinal mechanisms, the duration of the SP is highly variable as it can depend on factors such as the amount and type of muscle contraction, as well as instructions on how to maintain the contraction (21). While SICI is also susceptible to both experimental procedures and between-subject variation (45), it is generally accepted that it strictly represents inhibition at the cortical level (19, 28).

Limitations

In this study, TMS was only applied when the ankles were dorsiflexing past the neutral position. Despite this control, changes in the underlying TA background muscle activity were still observed between experimental conditions. This was likely a result of the more difficult conditions causing a greater sway velocity and consequently, individuals needing to increase lower limb muscle activation to keep the ankle angle near its neutral position for body stability. This introduces a potential confound when interpreting the results as it is possible that changes in bEMG activity could have influenced the modulation of the MEP responses. Previous studies have found that force output or voluntary contraction is positively related to MEP measures, where a stronger contraction is associated with an increase in the unconditioned MEP response (32) and a reduction in intracortical inhibition (33, 34, 46).

To address this concern, statistical analyses were conducted with bEMG activity as a covariate such that the reported effects of secondary task and balance difficulty are adjusted for the average level of bEMG. However, these analyses may not fully account for a possible differential effect of bEMG on task and difficulty depending on the level of bEMG (e.g,. low vs. high). For example, the relationship between contraction level and MEP amplitude has previously been observed to be more prominent at higher levels of muscle contraction (33). Therefore, it would be beneficial for future research to consider protocols that better incorporate similar levels of bEMG activity while still implementing different dual task difficulties or task requirements.

Conclusion

In conclusion, both balance performance and the cortical control underlying dual task performance involving a balance task differed depending on the secondary task being performed. When the dual task involved the secondary motor task, an increase in corticospinal excitability but no change in intracortical inhibition was observed while balance performance was impaired. In contrast, when the secondary task comprised of the cognitive task, there was no change in corticospinal excitability or intracortical inhibition with either no change or an improvement in balance performance. Considering the behavioral and neurophysiological findings together, it is speculated that dual tasks involving two tasks that use similar cortical and subcortical brain areas (i.e., both primary and secondary tasks are motor tasks) does not provide a compensatory opportunity and results in performance deficits despite higher activity of the primary motor cortex. In contrast, when the two simultaneous tasks likely differ in the utilization of subcortical areas (i.e., primary and secondary tasks have less overlap), this may allow individuals to shift the control of the primary balance task from predominantly cortical to more subcortical areas such that balance is performed in a more automated manner.

New & Noteworthy.

We examined motor cortical excitability when maintaining balance while performing a secondary motor or cognitive task. When the secondary motor task was involved, balance performance worsened and corticospinal excitability increased. Dual tasks involving the secondary cognitive task resulted in contrasting effects, with either no change or an improvement in balance performance along with no change in cortical control. These findings demonstrate how different motor cortical strategies are adopted depending on the dual task requirements.