There is very little consensus regarding the mechanisms underlying postural control. Whereas some theories suggest that posture is controlled at lower levels (i.e., brain stem and spinal cord), other theories have proposed that upright stance is controlled using higher centers, including the motor cortex. In the current investigation, we used corticomuscular coherence (CMC) to investigate the relationship between cortical and shank muscle activity during conditions of unrestricted and restricted postural sway. Participants were instructed to stand as still as possible in an apparatus that allowed the center of mass to move freely (“Unlocked”) or to be stabilized (“Locked”) without subject awareness. EEG (Cz) and electromyography (soleus and lateral/medial gastrocnemii) were collected and used to estimate CMC over the Unlocked and Locked periods. Confirming our previous results, increases in center of pressure (COP) displacements were observed in 9 of 12 participants in the Locked compared with Unlocked condition. Across these 9 participants, CMC was low or absent in both the Unlocked and Locked conditions. The results from the current study suggest that this increase is not associated with an increase in the relationship between cortical and shank muscle activities. Rather, it may be that increases in COP displacement with locking are mediated by subcortical structures as a means of increasing sway to provide the central nervous system with a critical level of sensory information.
- exploratory behavior
- postural sway
despite extensive research, there is still little consensus on the mechanisms underlying human postural control (Gatev et al. 1999; Loram et al. 2001; Morasso and Schieppati 1999; Nashner 1976; van der Kooij et al. 2005; Winter et al. 1998). For example, some theories assume that stance is controlled, either completely or in part, using a negative feedback system with delays (Loram et al. 2005; Maurer and Peterka 2005; Peterka 2000), whereas others assume sway is controlled by passive mechanisms, whereby the central nervous system is able to set the tone of calf musculature to increase the stiffness to a level that is sufficient to support the load of the body (Winter et al. 1998). More sophisticated control schemes assume that postural sway is controlled using predictive modulation of muscle activity (Gatev et al. 1999; Loram et al. 2001), anticipatory action to overcome sensory conduction delays (Morasso and Schieppati 1999), internal models of the body (Loram et al. 2005; Morasso et al. 1999), or as a means of exploring the environment (Riccio 1993; Riley et al. 1997; Riley and Clark 2003; Stoffregen and Riccio 1988; van Emmerik and van Wegen 2002). Arguments in support of the latter theory have drawn on recent evidence that center of pressure (COP) displacements observed during sway persist, and even increase, when sway movements of the center of mass (COM) are experimentally restricted without participant awareness (Carpenter et al. 2010; Murnaghan et al. 2011, 2013).
The shift toward more sophisticated models of postural control has emphasized greater complexity and levels of control. In addition to theoretical approaches, there is a growing body of empirical evidence from novel measurement techniques, animal models, and patient populations in support of the notion that higher centers may be involved in various aspects of postural control. Using PET scans, the cerebellum (particularly the vermis) has been shown to be activated more when stability is challenged (Ouchi et al. 1999). In addition, recordings from cortical neurons in quadrupeds have shown that neural activity is modulated when a new postural configuration is adopted (i.e., standing on a tilted surface; Karayannidou et al. 2009). Woollacott and Shumway-Cook (2002) have also reviewed a body of work that has shown that by manipulating cognitive load or attention there are decreases in postural stability. Furthermore, those with dementia or other cognitive deficits show decreases in stability and an increased risk of falls (Elble and Leffler 2000; Taylor et al. 2012).
One way to investigate whether changes in COP displacements are cortically generated under both normal and COM restricted conditions is to use a measure called corticomuscular coherence (CMC). CMC provides an estimate of the linear relationship between cortical (measured using EEG) and muscle activities (measured using surface electromyography, EMG) and has a spectral maximum between 15 and 30 Hz (Gerloff et al. 2006; Mima and Hallett 1999). Studies investigating CMC during quiet standing are limited, and although significant magnitudes of coherence have been found while performing a voluntary contraction, coherence was not observed when participants performed a variety of postural tasks (Masakado et al. 2008).
As such, the objective of the current study was to use CMC to investigate whether there is cortical drive to musculature of the lower leg when participants were standing freely (“Unlocked”) and when the COM was stabilized (“Locked”) in the sagittal plane. We hypothesized that there would be little to no coherence when participants were unlocked and freely moving. Based on previous studies, we further hypothesized that, in the majority of participants, there would be an increase in COP displacements when the COM was stabilized during quiet stance. Since CMC may provide an indication of cortical drive to the muscles, it was also hypothesized that, in those participants whose COP displacements increased with locking, significant increases in coherence at frequencies in the range of 15–30 Hz would also be observed.
MATERIALS AND METHODS
Twelve healthy young adults (8 females; mean ± SD for age = 23.8 ± 3.9 yr; height = 173.3 ± 7.0 cm; weight = 67.8 ± 11.3 kg) participated in the study. Each participant provided informed, written consent, and the experimental protocol was approved by the Behavioural Research Ethics Board at the University of British Columbia. All participants were completely naïve to the goals of the experiment and the intended effect of the apparatus on postural sway.
The apparatus used in the study was the same as that described in Carpenter et al. (2010) and Murnaghan et al. (2011). During all experimental trials, participants were firmly braced with their backs against a rigid board with adjustable straps tightened firmly around the head, shoulders, chest, waist, hips/upper thighs, and upper shank to prevent movement at any joint except the ankle. The board was 1.66 m high (including head rest) × 0.61 m wide and had a total mass of 12.5 kg. The board was attached to a closed-loop pulley system that allowed “normal postural sway” at the ankle joint unless the experimenter applied a brake that discretely locked the board (and thus COM) in place in the sagittal plane. To eliminate any chance that the participants could receive auditory cues indicating that they were being locked, all participants wore earplugs that reduced any noise within the testing area. Participants also wore blinders designed to occlude both horizontal and vertical peripheral vision while maintaining full visual input anteriorly. In all conditions, participants stood with their arms crossed and feet shoulder-width apart on a force plate (#K00407; Bertec).
The study required measurable levels of CMC between EEG and EMG activities during voluntary muscle contraction. This cannot be obtained in all participants and is reported to occur in approximately 50–75% of individuals (Masakado et al. 2008; Perez et al. 2012). As such, participants were initially screened to ensure that they showed significant CMC with at least one muscle of the lower leg [lateral and medial gastrocnemii (LGAS and MGAS) and soleus (SOL)] during a 2-min seated low-level contraction (∼30% maximum voluntary contraction) and that this magnitude of coherence was significantly larger than that recorded during a 2-min seated relaxed trial (Fig. 1, A and B). Twelve participants showed measurable levels of CMC. Each of the twelve participants were fitted into the apparatus and asked to stand as still as possible in one trial of ∼6-min standing duration (Fig. 2). The 1st 30 s of the trial was used to allow the transient component of sway to stabilize (Carroll and Freedman 1993), and the following 30 s was used to calculate the mean COP position to be used as the threshold for locking. Following the initial 60 s, participants stood freely in the unlocked condition for a minimum of 135 s (Unlocked), and then the board was locked without participant knowledge when the COP was within 2 SD of the calculated mean COP for a minimum of 135 s (Locked). One-hundred-thirty-five-second periods were used to ensure 2 min of time could be analyzed in each of the Unlocked and Locked conditions.
To characterize the oscillatory interactions between motor cortical areas and motor neuron pools of muscles using coherence analysis, we collected synchronous measures of EMG and EEG activities. Surface EMG activity was sampled bilaterally from LGAS, MGAS, and SOL muscles using bipolar Ag/AgCl surface electrodes placed ∼2 cm apart on each muscle belly. EMG data were sampled at 2,048 Hz, band-pass filtered between 30 and 500 Hz, and full-wave rectified to capture the temporal pattern of grouped motor unit firing regardless of its shape, which can vary with the relative position of active and reference electrodes (Halliday et al. 1995; Mima and Hallett 1999). EEG was collected using a 32-channel electrode configuration based on the International 10-20 standard (CAP-ANTWG32; Advanced Neuro Technology) with the vertex (Cz) positioned halfway between the nasion and inion and the reference electrode positioned anteriorly to the vertex at AFz. EEG data were sampled at 2,048 Hz and band-pass filtered between 1 and 500 Hz. We focused primarily on recordings over the leg area of the sensorimotor cortex (Cz) where coherence between leg muscle EMG and the cortex is strongest (Salenius et al. 1997). During all EEG recordings, participants were instructed to limit any blinking, jaw clenching, or facial expressions that could introduce artifacts into the EEG signal. EEG and rectified EMG data from each trial were then segmented into 120-s periods of Unlocked and Locked data. The 120-s measurement periods began 135 s before when participants were initially locked and 135 s before when the lock was removed, respectively.
Coherence analysis was based on the theoretical framework described by Halliday et al. (1995) and Rosenberg et al. (1989) and calculated by modifying publicly available MATLAB scripts (NeuroSpec 2.0; Dakin et al. 2010; Halliday et al. 1995; Rosenberg et al. 1989). The first part of the analysis was performed on individual records from each participant using frequency domain measures of correlation between EEG and rectified EMG in both the Unlocked and Locked conditions with EEG as the reference signal and EMG as the output signal. The second part of the analysis required further concatenation across all participants who showed an increase in COP displacements with COM stabilization to create a single pooled data array for each condition. Because the amplitudes of data records being pooled may vary across participants, the records from each participant were normalized by dividing by the SD before estimating pooled parameters as suggested by Baker (2000) and Halliday and Rosenberg (2000). In both the individual and pooled analyses, data from each condition were analyzed using segments of 1,024 points (0.5 s), giving a frequency resolution of 2 Hz. Coherence estimates are unitless, bounded between 0 and 1, and enable the identification of correlated frequencies between the two signals, where 1 indicates a perfect relationship, and 0 indicates independence. Any coherence was deemed significant at a particular frequency when it surpassed the 95% confidence interval, which was calculated according to the number of segments (Halliday et al. 1995). Differences in coherence between conditions were then calculated using a difference of coherence test (Amjad et al. 1989).
Ground reaction forces and moments were sampled at 1,000 Hz and low-pass filtered offline using a 5-Hz dual-pass Butterworth filter before calculating COP in the anterior-posterior (AP) and medial-lateral (ML) directions. From these signals, the root mean square (COP RMS) was calculated. Kinematics were sampled at 500 Hz for the duration of each trial using an OptoTrak three-dimensional optical motion analysis system (Northern Digital, Waterloo, Ontario, Canada) with infrared light-emitting diodes (IRED) placed on the cable, ankle joint, and back of the board at a level that approximated the height of the COM. Kinematic data were filtered at 5 Hz with a dual-pass Butterworth filter and used to calculate linear displacements of the COM in the sagittal plane. From this signal, the RMS of AP COM displacements (COM RMS) were calculated. Finally, the magnitude of EMG from LGAS, MGAS, and SOL was calculated as an integrated area of the processed and rectified EMG averaged across the left and right sides. All dependent variables were calculated over the same 120-s Unlocked and Locked periods.
In addition to the statistical tests described above, which were used to compare estimates of coherence, we tested whether there were significant differences in all other dependent variables across Unlocked and Locked conditions using dependent t-tests. Significance was assumed at an α-level of 0.05 and was corrected for multiple comparisons using a Bonferroni correction.
In the current study, the apparatus was effective in minimizing displacements of the COM without participant awareness. Across all 12 participants, COM RMS was significantly reduced in the Locked (0.55 mm) compared with Unlocked (5.92 mm) condition (P < 0.001). The result of stabilizing the COM on COP displacements was similar to that reported in previous studies (Carpenter et al. 2010; Murnaghan et al. 2011). Increases in COP displacements were observed primarily in the AP direction with minimal and nonsignificant differences in the ML direction (P = 0.16). In the AP direction, 9 of 12 subjects showed increases in COP displacements from Unlocked to Locked, and, on average, COP displacements increased from 4.18 to 13.2 mm (P = 0.02; Fig. 3). Increases in AP COP displacements from Unlocked to Locked were associated with a significant increase in the amplitude of MGAS EMG (P = 0.02) and LGAS (P = 0.05) with a similar trend observed in SOL EMG (P = 0.08).
Analysis of CMC showed that increases in AP COP displacements with locking were not associated with increases in coherence at frequencies in the range of 15–30 Hz. Specifically, when analyzing the individual records from each of the nine participants whose AP COP displacements increased from Unlocked to Locked, we found that there was no concomitant increase in coherence between oscillatory activities of the EEG and triceps surae EMG. Despite having significant increases in coherence in at least one muscle during the seated voluntary contraction (Fig. 4A), when standing freely in the Unlocked condition, coherence between Cz and triceps surae muscles was reduced or absent (Fig. 4B). Although AP COP displacements increased with locking in this group of participants, we found that the magnitude of coherence did not change in eight of nine participants (Fig. 4B). In the participant whose magnitude of coherence between Unlocked and Locked was significantly different, coherence was actually greater in the Unlocked condition compared with the Locked condition.
To confirm that increases in AP COP displacements were not associated with concomitant increases in CMC and that the absence of any change in CMC between Unlocked and Locked conditions was not the result of insufficient power, data records from each of the nine participants whose AP COP increased from Unlocked to Locked were pooled across each condition. Similar to our findings from the individual data analysis, clear increases in CMC were observed from seated relaxed to seated contracting conditions, illustrating that measurable levels of CMC between EEG and EMG activities were attainable during voluntary contraction (Fig. 5A). When standing freely in the Unlocked condition, measures of CMC were minimal or absent, and there was no significant difference in coherence from Unlocked to Locked (Fig. 5B).
Theories describing the mechanisms underlying postural control are abundant and contribute diverse perspectives to our understanding of how posture is controlled. Whereas some theories suggest that the mechanisms underlying postural control operate at lower levels (i.e., brain stem and spinal cord), other theories have proposed that posture may be controlled using higher centers, including the motor cortex. In the current study, we used CMC to investigate whether COP displacements are generated by means of cortical drive to muscles of the lower leg under normal and COM stabilized conditions.
When participants were standing in the apparatus and freely swaying, little to no coherence between EEG and surface EMG (LGAS, MGAS, or SOL) in the 15- to 30-Hz frequency range was observed. Observations of absent or low levels of coherence during normal standing were consistent with results of previous studies that have been unable to detect similar magnitudes of EEG-EMG coherence during postural tasks (standing relaxed and standing still) as those obtained in the same participants during voluntary contraction (Luu 2010; Masakado et al. 2008). As a whole, these results suggest that postural sway in unrestricted conditions is not actively generated by the leg region of the motor cortex.
Based on previous studies, we also hypothesized that when movements of the COM were stabilized in the sagittal plane, increases in COP displacements would be observed. In accordance with our previous findings (Carpenter et al. 2010; Murnaghan et al. 2011, 2013), when displacements of the COM were minimized by the experimenter without participants' knowledge, AP COP displacements increased in 75% of participants, and these increases were accompanied by increases in triceps surae muscle activity. Given these increases, it seems plausible that under conditions of external stabilization increases in COP displacements could be actively driven and potentially be mediated by the motor cortex. We initially hypothesized that when the COM was stabilized by the experimenter, increases in the magnitude of coherence would be observed. However, contrary to our hypothesis, we found little to no coherence despite increases in triceps surae EMG. These results suggest that increases in AP COP displacements with locking were not the result of increasing oscillatory cortical drive (Mima and Hallett 1999).
Despite the absence of oscillatory 15- to 30-Hz cortical drive to the musculature of the lower leg in both the Unlocked and Locked conditions, we cannot entirely dismiss a cortical origin for AP COP displacements. Significant magnitudes of CMC between EEG and EMG are thought to arise when there is synchronous discharge of a large number of cortical neurons in an oscillatory manner. However, it could be that the motor cortex activates triceps surae motoneurons differently in postural tasks from in voluntary contractions (Masakado et al. 2008), and, in using an estimate of CMC, any cortical drive that is nonoscillatory in nature would go undetected (Petersen et al. 2012). Furthermore, it could be argued that AP COP displacements in unrestricted conditions, as well as increases observed following stabilization, do not originate from the leg sensorimotor area recorded in this study. Although CMC has been reported to be largest between Cz and shank muscle EMG under voluntary conditions (Salenius et al. 1997), it has been documented that there are corticospinal connections from other frontal and parietal areas (premotor and supplementary motor areas as well as area 5; Dum and Strick 1996; Geyer et al. 2000), and these areas have been suggested to contribute to various aspects of postural control (Jacobs and Horak 2007; Maki and McIlroy 2007). In addition, there is debate regarding how lower-limb musculature associates with changes in COP displacements (Gatev et al. 1999; Loram et al. 2004; Winter et al. 1998), questioning whether activation of triceps surae musculature represents the means by which the COP is controlled. Therefore, it is possible that muscles other than those of the triceps surae may contribute to COP displacements observed in Unlocked conditions or to the increases in COP observed following stabilization. In fact, some research has suggested that foot musculature may provide active control of posture independent of any changes in ankle movement (Kelly et al. 2012; Wright et al. 2012). As such, we cannot ignore that there could be oscillatory cortical drive to smaller muscles of the foot that were not recorded in the current protocol. However, it should be noted that we observed changes in COP displacements with locking in the AP direction, whereas the effects of foot muscle activation (abductor hallucis, flexor digitorum brevis, and quadratus plantae) have been reported to correlate with ML but not AP displacements of the COP (Kelly et al. 2012).
It could be argued further that displacements of the COP observed in Unlocked and, to an even greater extent, during Locked conditions could be considered “dynamic” in nature and therefore not observable using estimates of CMC in the 15- to 30-Hz frequency band. In dynamic tasks that contain a dynamic ramp phase of contraction, the magnitude of CMC in the 15- to 30-Hz frequency range has been shown to be reduced compared with when the contraction is maintained (Kilner et al. 2000). Rather than observing a simple reduction of frequencies within the 15- to 30-Hz range during dynamic tasks, some have reported that there is actually a shift toward higher frequencies (Omlor et al. 2007). However, our data do not suggest that this is the case since we did not find significant magnitudes of CMC in any frequencies up to 50 Hz in either the Unlocked or Locked conditions (Figs. 4B and 5B). In addition, although the magnitude of coherence was reported to be larger during phases of walking where the EMG is more stable or “static” in nature, significant magnitudes of coherence were still observed throughout phases of walking, a task that would involve greater modulation of EMG than that observed during our standing postural task (Petersen et al. 2012).
Finally, although our data do not suggest that AP COP displacements, observed while freely standing (Unlocked) or following COM stabilization (Locked), are produced via oscillatory cortical drive to triceps surae musculature, we should consider the possibility that they could originate from other efferent pathways. One line of evidence supporting this hypothesis comes from experiments investigating whisking behavior in rodents. Whisking is a sensorimotor behavior used to gather detailed sensory information about the surrounding environment rapidly, and this behavior has been shown to persist even in decorticated rats (Deschênes et al. 2012; Semba and Komisaruk 1984). Whereas the premotor circuitry has been heavily debated, many have suggested that a large component of the modulation of whisking behavior arises from a brain-stem sensorimotor loop (Deschênes et al. 2012; Nguyen and Kleinfeld 2005). With the many parallels that can be drawn between postural sway and whisking in the rat, it is therefore conceivable that COP displacements could be driven by subcortical structures.
In conclusion, the results of the current study suggest that when participants are standing freely it is unlikely that posture is controlled by means of cortical drive from the leg area of the sensorimotor cortex. When movements of the body are stabilized in the sagittal plane without participants' knowledge, increases in COP and EMG activity were observed. However, our results do not support that this increase is produced via oscillatory cortical input to lower-limb musculature. Rather, this increase may be driven by other efferent pathways, potentially originating within the brain stem or other subcortical structures.
We gratefully acknowledge the funding support provided by the Natural Sciences and Engineering Research Council of Canada to M. G. Carpenter, J. T. Inglis, and C. D. Murnaghan.
No conflicts of interest, financial or otherwise, are declared by the author(s).
C.D.M., R.C., J.T.I., and M.G.C. conception and design of research; C.D.M. and J.W.S. performed experiments; C.D.M. and J.W.S. analyzed data; C.D.M., J.W.S., R.C., J.T.I., and M.G.C. interpreted results of experiments; C.D.M. prepared figures; C.D.M. drafted manuscript; C.D.M., J.W.S., R.C., J.T.I., and M.G.C. edited and revised manuscript; C.D.M., J.W.S., R.C., J.T.I., and M.G.C. approved final version of manuscript.
- Copyright © 2014 the American Physiological Society