In the present study, we assessed measures of balance control while participants were standing on a force plate in order to develop an event-related approach sensitive to detect interference between balance control and control of cognitive conflict in the range of sub-second time periods. Specifically, we had participants perform a Simon task, which requires resolution of conflict during selection of left vs. right manual button-press responses. We showed that spatial response conflict in the cognitive Simon task may have affected balance control in terms of a greater reduction in mediolateral body sway variability within a short period preceding the execution of the instructed manual reactions. In contrast, there was neither such cognitive-motor interference associated with target encoding in the Simon task nor with or after response execution. These observations may indicate that process-specific effects in a cognitive task related to response selection carry over to the balance domain.
Analysis of balance control using body sway parameters
While we hypothesized that resolution of response selection conflict would interfere with the control of body sway based on a wealth of previous literature on such interactions assessed by aggregating across much longer time periods, it was important to show that our effects would be related to the occurrence of specifically defined cognitive processes. Moreover, while we could predict that the cognitive process of response selection is critical, based on neurocognitive findings (see Cespón et al.,
2020, for a review), there was no prior knowledge of which aspect of body sway control would be susceptible and sensitive to any cognitive interference. Therefore, we analysed the average moment in each direction as an expression of balance-related state, and we analysed the variability of the moments as an indication of the balance control effort imposed on body sway.
Both body sway parameters (AV and SD moment) expressed a processing stage-dependent modulation across the time course of an entire trial. In the 150 ms after target onset, that is in the stimulus processing stage, the average force moment deviated less from zero in both directions of sway in comparison to the response selection and response execution stages. As Fig.
3 illustrates for an individual participant, this is to be expected due to the centring of the force moments relative to the average determined across a trial’s pre-target fixation period. From presentation of the target onwards, the force moment will increase its distance from the zeroed state until it settles at the time of the response. However, the congruency condition of a particular trial did not have any effect on this dynamical indicator. These observations merely validate the sensitivity of the methodological approach.
More relevant to the research questions posed in this study is that the variability within time bins demonstrated a reduction by about 8% from the period after target presentation towards execution of the response. This indicates that response execution (manual trigger press) in itself did not cause any physical movement artefacts. While target presentation resulted in greater local variability than response execution, we found that cognitive conflict during spatial left vs. right responses reduced variability of the mediolateral moment in the 150 ms time bin before response onset in incongruent trials further, that is, when response selection conflict was resolved and/or inaccurate response tendencies were suppressed. In this situation a mediolateral variability reduction of approximately 14% occurred. Hence, using a new event-related methodology to force plate data, we found a specific balance correlate for cognitive spatial response conflict that did not generalize to anterior–posterior movements or to time bins for which such effects were not predicted beforehand (i.e., with target encoding or only after response execution).
Note that our finding of reduced sway variability in ML moment in incongruent trials relative to congruent trials is specific in many respects. (i) Different from previous studies that examined the control of body sway in traditional measures with low temporal resolution, we investigated body sway variability on a very short timescale, (ii) the congruency effect is only visible when the moment time series data were aligned by the onset of the manual reactions, (iii) the influence of trial congruency was limited to the mediolateral direction of sway during the selection phase of the correct manual response, the plane in which the directional conflict exists, and (iv) only the temporal bin integrating the 150-ms period before the manual response was susceptible to the congruency condition of the target. In contrast, the target-aligned 150 ms time bin following target presentation showed no effect of congruency condition. Together, the data strongly suggest that in the bin 150 ms prior to response onset, processes related to response selection dominate activity instead of processes related to target processing.
We designed our analytical approach after event-related neurophysiological studies of the Simon task (see Cespón et al.,
2020, for a review). The relatively high sample rate of 1 kHz that we employed to acquire body sway responses as well as the alignment of the traces of the ground reaction force moments by either the onset of the visual target or the onset of the manual response enabled us to look for spatiotemporal commonalities in the time series data. In contrast to Simon task effects expressed by altered event-related potentials, however, we did not find a specific Simon effect on the average moment before or after the response. A reason may be that the fluctuation of the moment does not seem to possess a characteristic or stereotypical waveform that could allow time series superposition for the detection of specific spatiotemporal features as exemplified in Fig.
3. Note though that this impression does not preclude the possibility that, in the future, more sensitive data analytical procedures could potentially identify fluctuations in the average moment that distinguish between conditions with and without conflict during selection of the manual reaction.
The reduction in the mediolateral variability of the moment of the ground reaction force in incongruent trials may be an expression of a lowered probability for a direction-specific balance adjustment to occur during the cognitive selection of a manual response if there is a response conflict to be resolved. As the time point of presentation of a target stimulus was quasi-jittered in relation to the time point at which a balance correction had to be performed by the randomization of a trial’s congruency condition and by the variability of a participant’s past response latencies, we assume that interference between the requirement to adjust balance and a conflict in response selection did not emerge in every single incongruent trial but in the statistical majority of those trials. Nevertheless, future studies might choose to add deliberate jitter to the inter-trial-intervals to further decouple the regularity of stimulus presentation from regularities in balance control.
It is possible that cross-talk between the cognitive task and balance control was caused either by body movements associated with the manual reactions or by target-oriented saccades or gaze shifts. The preparation or execution of either actions might have had a direct impact on body sway. However, these effects ought to have affected trials with congruent and incongruent targets to the same degree. Also, target position and any associated saccades and bodily orienting responses were uncorrelated with the congruency variable as target location was not the response-relevant feature.
In order to explain the differences in response-locked sway control variability between congruent and incongruent trials, one would have to assume differences in the characteristics of these motion and oculomotor byproducts. Future studies ought to record pressure forces at the trigger button, hand and body kinematics and eye movements to control for these possible confounds. An argument against possible influences of response button presses on body sway is the observation that variability of sway is numerically greater during the target perceptual stage than the response stage. A manual pressure-related byproduct should evoke increased variability on the latter stage.
From the dual-task literature it is well known that two capacity-limited processes interfere with each other (i.e., produce a processing bottleneck) when required on a very short time scale of only dozens of milliseconds, and this interference has been termed “psychological refractory period” (PRP) effect (for reviews see Koch et al.,
2018; Pashler,
1994). In the present experimental conditions with presumably high response conflict in incongruent trials, we believe that conflict-induced delay of response selection in the Simon task can propagate to regularly triggered, small balance corrective adjustments. Overlap between capacity-limited “central” cognitive processes, required for resolution of response conflict and the effortful selection of a manual response, and processes involved in the triggering of a balance correction may have created a transient “micro-bottleneck”. More specifically, these coincidental micro-bottlenecks could occur because balance corrections that would have been triggered close in time to the independently triggered response selection process in the Simon task may have been altered or omitted because of an (i) absence, (ii) a partial reduction or (iii) a delay in an intermittent balance control signal. For example, Loram and colleagues (2011) assumed that feedback control of body balance is a serial, ballistic process in which the balance state is observed continuously but adjustments occur in an intermittent, predictive open-loop fashion (Gawthrop et al.,
2011; Loram et al.,
2011). While peripheral mechanisms for balance control are supposed to have a high processing bandwidth, the bandwidth of central balance control is considered relatively low. For example, relatively long feedback time delays of latencies longer than 150 ms indicate low bandwidth but more flexible control of the direction and amplitude of body sway by the involvement of intentional control mechanisms in a context-specific manner (Loram et al.,
2009). Thus, as our Simon task may have created conflict in spatial terms along the mediolateral (body) axis, the equivalence with a congruency effect on mediolateral balance control is a plausible finding. Possibly the contextual demands of the Simon task implicitly imposed direction-specific constraints on the balance control of ML sway. For example, the Simon task demands may have interfered more selectively with neuromuscular control of ML sway via the hips compared to AP sway controlled by the ankles (Winter et al.,
1996).
Processes of cognitive conflict resolution may play a direct role in balance control such as for the resolution of intersensory conflict (Redfern et al.,
2009). Recently, Redfern and colleagues (
2018) observed relationships between body sway in conditions with differing demands on intersensory conflict resolution and diverse cognitive functions, such as decision speed, control of cognitive conflict and abilities of visuospatial processing and memory. Performance in a cognitive conflict task correlated with sway especially when intersensory conflict was induced by sway-referenced visual feedback and a fixed support base (Redfern et al.,
2018). The authors concluded that visual cognitive conflict resolution shares processes with sensory integration and intersensory conflict resolution when balance control relies predominantly on somatosensory afferences and a relative down-weighting of vision.
An important distinction between theoretical approaches that assume an intermittent balance control scheme is the question whether any balance corrections are performed at regular temporal intervals or in an event-dependent fashion, by which a balance adjustment is triggered when the estimated balance state, either measured or predicted, transgresses a threshold criterion. The decision at which point to activate and deactivate specific muscles could be made by a high-level monitoring process that considers the rate of change in the stability component in a phase space representation (Tanabe et al.,
2017). Thus, it seems justified to conclude that balance control incorporates high- and low-level processes and may be intermittent. While low-level and more automatic processes may be regulating the maintenance of postural stiffness and damping at a higher temporal frequency, high-level processes monitoring and predicting states of balance stability may show periods of “cognitive neglect” lasting 150 ms and longer when engaged in cognitive response conflict resolution.
We did not use secondary-task methodology to analyze cognitive-motor interference but instead an event-related approach examining interference at the level of specific processes with high temporal resolution. Previous observations using the traditional secondary-task methodology, where measures of cognitive and balance performance were contrasted across blocked conditions, reported sway reduction as the result of an adopted balance control strategy (Brown et al.,
2002). The assumption that the balance control system strategically increases postural stiffness, however, does not apply to the methodology used in our present study. Here congruent and incongruent trials alternated randomly in short intervals and which, therefore, rendered rapid switching between balance control strategies following target presentation implausible. Instead, it is more reasonable to assume that during the entire experiment, an individual’s mode of balance control rested in a single default state of sensorimotor organization.
A limitation of the present study is its exploratory nature so that a replication in the same or in similar experimental paradigms would seem desirable. In fact, the present paper suggests that cognitive-balance interference can be uncovered using an event-related approach. Still, the observed selectivity of the reported findings might be a spurious result, effectively leading to a false-positive erroneous conclusion. The question at this point is, however, what can be gained from the assumption that cognitive control does interfere with the control of balance on the global level of task blocks, as seen in studies that pursued the traditional multitasking approach, but NOT on the level of individual cognitive trials? The assumption that conflict resolution has no effect on balance control at a local level or only when it occurs as a uniform effect across all trial phases, both directions of sway and in all complementary performance parameters, is of a very strong order. This assumption would in its nature disregard many aspects of our current understanding about the selectivity of cognitive control processes as well as the known sensitivity of balance control to the context of a specific task. Nevertheless, the experimental methodology for investigating the interaction between balance and cognitive control ought to be further extended and improved in order to detect the omission or delay of single balance adjustments.