Balance and postural equilibrium are achieved through continuous integration and processing of sensory signals related to the orientation of our bodies and limbs in space (Massion, 1994, 1998). Indeed, cortical activity is likely important in maintaining upright postural equilibrium following whole-body perturbations (Dietz et al., 1985; Quant et al., 2004b; Little and Woollacott, 2015). However, little is known as to the precise cortical structures that are active during reactive balance control.

Appropriate reactive motor responses to translational movement are essential in maintaining upright balance. Short and longer latency muscle responses from the spinal cord and brain stem provide means of stabilizing the body immediately following a postural disturbance (Dietz et al., 1985; Ackermann et al., 1986; Macpherson and Inglis, 1993). Cortical potentials arising from a postural perturbation are thought to be involved in error detection for postural control; they would thus be used to update the central nervous system regarding the need to modify responses to upcoming perturbations, in order to impact subsequent control of upright equilibrium (Adkin et al., 2006). Conventional electroencephalography (EEG) allows for the characterization of event-related potentials (ERPs) that can be recorded during whole-body postural perturbations (Dietz et al., 1985; Ackermann et al., 1986; Staines et al., 2002; Quant et al., 2004b; Maki and McIlroy, 2007). Typical ERP components (i.e., P1, N1 and P2) that appear following a rapid surface translation task are thought to represent kinesthetic feedback from the peripheral limbs (Dietz et al., 1985; Ackermann et al., 1986; Staines et al., 2002; Quant et al., 2004b; Maki and McIlroy, 2007). Moreover, the N1 latency in particular is useful in characterizing cortical processing of sensory dynamics due to its consistent appearance compared with the other components (Quant et al., 2004b; Adkin et al., 2006; Little and Woollacott, 2015). Most previous studies investigating ERPs in response to whole-body postural perturbations selected specific electrode locations at the scalp and did not attempt to locate the cortical sources related to maintaining reactive postural control.

Furthermore, dual-task paradigms require simultaneous engagement with a cognitive and motor task (Lajoie et al., 1993; Vander Velde and Woollacott, 2008). Previous studies have demonstrated how various cortical areas likely assist with postural equilibrium due to decrements in performance in either cognitive or postural tasks and an attenuated N1 ERP amplitude (Quant et al., 2004a; Adkin et al., 2006; Little and Woollacott, 2015), ostensibly owing to a finite capacity of the cortex to perform parallel processing (Wickens, 1983). A limited number of investigations have examined cortical activity time-locked to whole-body surface translations when participants performed a cognitively demanding task (Quant et al., 2004a; Adkin et al., 2006; Little and Woollacott, 2015). For example, Adkin et al. (2006), utilized a visual tracking task during whole-body surface translations and found decrements in both tracking performance and N1 amplitude beneath the frontocentral electrode (i.e., Fz and Cz) locations compared with a reactive standing task alone. Further, Little and Woollacott (2015) corroborated these findings by using a visual working memory (VWM) task during whole-body surface translations and reported a decrease in the N1 amplitude beneath electrodes positioned over the motor and somatosensory cortical areas during a dual-task paradigm compared with a single task. These aforementioned reports provide evidence that attentional and postural systems are in competition within the cortex to maintain upright balance, but do not provide insight regarding the cortical sources contributing to the N1 ERP nor the re-allocation of resources responsible for its attenuation during a dual task.

Methodological improvements allow investigation into the electrocortical dynamics during whole-body motion that likely includes head movements, such as perturbations, walking and even running (Makeig et al., 2009; Slobounov et al., 2009; Gramann et al., 2010; Gwin et al., 2010; Sipp et al., 2013; De Sanctis et al., 2014; Seeber et al., 2015). High-density EEG in conjunction with independent component analysis (ICA) and source localization methodology can be used to estimate electrocortical areas that are synchronized to related events (Gwin et al., 2010; Wagner et al., 2012, 2014) and to discriminate these from non-cortical artifacts such as eye and muscle activity (Gramann et al., 2010; Gwin et al., 2010). Recently, investigators have attempted to determine the location of cortical activity related to upright postural control (Slobounov et al., 2009; Marlin et al., 2014). For example, one study (Slobounov et al., 2009) illustrated cortical activity within the anterior cingulate cortex (ACC), the limbic area, and at the junction of the precuneus and occipital lobe when human subjects stood on one leg with their eyes closed. The authors suggested these cortical structures may assist with predicting future postural instability and were activated due to the unstable nature of the task. Because these results are inconsistent with previous findings related to upright bipedal posture (Marlin et al., 2014), additional studies are required to locate the cortical sources related to reactive balance control.

The purpose of this study was to estimate the location of the cortical sources related to a whole-body surface translation during a single- and dual-task paradigm. A second aim was to determine whether a secondary cognitive task (i.e., diverted attention) attenuated the N1 ERP amplitude corresponding to the postural perturbation. We hypothesized that electrocortical activity related to the whole-body surface translations would be located within the sensory and (pre)motor cortices and the dual task would elicit additional electrocortical sources within the frontoparietal attentional networks, similar to previous work using functional magnetic resonance imaging (fMRI; Deprez et al., 2013; Johannsen et al., 2013). We also expected that during a dual-task compared with a single-task paradigm, the N1 ERP amplitude would decrease within the cortical areas related to the reactive postural task.

The measure projection analysis clustering algorithm was used to provide estimates of distinct cortical sources related to the single and dual-task conditions for all subjects (Figure 3). Anatomical and Brodmann area (BA) probability values related to each domain are listed within Table 1 and calculated using the Laboratory of Neuroimaging project probabilistic (Department of Neurology, University of California, Los Angeles, CA, USA) and BA (Lancaster et al., 1997; Shattuck et al., 2008) atlases. Listed functional associations with each BA are based on Brodmann’s interactive atlas (Bigdely-Shamlo et al., 2013). The visual representations of cortical domains with larger volumes demonstrate increased physical distances for all dipoles and the opposite for those with more condensed volumes. For descriptions of BAs and anatomical locations distinct domains see Table 1. As a note, the methodologies used here estimated source localization within subcortical areas (e.g., caudate) related to the postural perturbation, but owing to technological concerns and limitations of EEG (Olbrich, 2015), we refrain from reporting and discussing these results within this manuscript.

The primary aim of the current study was to estimate the location of cortical areas related to the onset of whole-body surface translation with or without performing an executive function task (i.e., VWM task). Second, we aimed to address whether the N1 ERP amplitude was attenuated for the dual compared with single task within the same cortical regions. For the reactive balance task alone (single task), the domains identified related to the whole-body perturbation included cortical representation from primary, pre-, and supplementary motor, primary somatosensory and spatial processing and error detecting areas (Table 1). Moreover, for the dual task, cortical sources shifted to include visuospatial attention, but less error detection areas (Table 1). In agreement with our hypothesis, when N1 ERP amplitude from the four most representative domains was compared, there was a significant depreciation for the dual than single task (Figure 7). Our novel results provide convincing evidence regarding the cortical sources and attenuation of the N1 ERP in relation to reactive postural control and extend previous findings on upright posture (Dietz et al., 1985; Staines et al., 2002; Quant et al., 2004a; Adkin et al., 2006; Little and Woollacott, 2015; Mierau et al., 2015) and dual-task processing using fMRI (Deprez et al., 2013; Johannsen et al., 2013).

For the single-task condition, domains 1, 2 and 4 were estimated to include the somatosensory and pre-, primary and supplementary motor cortices (Table 1), which corroborate previous findings (Dietz et al., 1985; Quant et al., 2004b; Adkin et al., 2006; Marlin et al., 2014; Little and Woollacott, 2015; Mierau et al., 2015). Furthermore, domains 3 and 4 of the single task included cortical sources representing error detection and processing areas localized to the ACC and frontal cortex (Table 1). Previously, some studies (Slobounov et al., 2009; Mierau et al., 2015) reported the involvement of the ACC during a single-legged balance task, but another study (Marlin et al., 2014), implementing a lean and release reactive balance paradigm, did not. The disparate findings may be related to differences in experimental paradigms or difficulty of the respective reactive postural control tasks. Adkin et al. (2006) speculated that following unexpected perturbations, error-detecting centers are activated to assess overall postural state before and after translational movement to ensure that the appropriate motor responses are selected to achieve balance equilibrium. Thus, based on our results and that of others (Dietz et al., 1985; Quant et al., 2004b; Adkin et al., 2006; Slobounov et al., 2009; Marlin et al., 2014; Little and Woollacott, 2015; Mierau et al., 2015), the motor and somatosensory areas and ACC are likely important cortical sources related to reactive balance control.

Even though, many of the estimated cortical sources were similar for the single and dual tasks, notable differences between conditions were observed, particularly within the motor and error detection regions. For example, three cortical domains (1, 2 and 4) were identified for the single task in which motor areas were most likely associated with the corresponding domain (probability = 0.89, 0.28 and 0.16, respectively); whereas only two domains (2 and 5) were estimated for the dual task (probability values of 0.87 and 0.11, respectively; Table 1). Error detection and processing areas were represented by two domains (3 and 4) for the single task (Table 1); however, the dual task only had one source (domain 1), which exhibited a low probability that an error detection area (ACC) was most likely associated with the respective domain (Table 1). These results provide direct evidence in support of the idea of a limitation in the availability and re-allocation of cognitive resources while processing tasks simultaneously (Wickens, 1983; Little and Woollacott, 2015). During the present protocol, participants were fixated on a VWM task for the dual-task protocol and prior to the perturbation, their cognitive resources were likely distributed to posterior parietal and occipital cortex areas (Vogel and Machizawa, 2004; Xu and Chun, 2006; McCollough et al., 2007). Activating these particular areas prior to the perturbation onset would potentially withdraw cortical resources from the motor, somatosensory and error detection areas, limiting overall availability to efficiently process activity related to the postural perturbation. Our localization results in combination with significant decrements in the behavioral outcome measures (i.e., VWM capacity and center of pressure peak trajectory) for the dual compared with single task (Little and Woollacott, 2015) indicate a re-distribution and sharing of attentional resources.

In support of limited cognitive resources, analysis of activity over all electrode sites, as represented by the ERP envelopes, revealed attenuation of the N1 ERP amplitude for the dual compared with single task (Figure 7). This study supports and expands upon previous reports only analyzing N1 ERP amplitudes from pre-selected electrodes (Quant et al., 2004b; Adkin et al., 2006; Little and Woollacott, 2015) by providing calculated pvaf values for each domain. The pvaf values illustrate how much variability of the grand average ERP is explained based on source dynamics (domain dipoles) within the brain. Thus, this analysis may provide a better representation of the cortical areas contributing to the grand average compared with measured ERP amplitudes at the electrode sites; a technique traditionally performed in EEG analysis (Quant et al., 2004a; Adkin et al., 2006; Little and Woollacott, 2015).

In contrast to the reduced cortical sources associated with motor and error processing, there was an increase in the localization of areas involved with spatial processing and attention for the dual compared with single task. For the single task, only domain 2 included the probability that a spatial processing area was most likely associated with that source and was localized to the parietal cortex (Table 1); however, for the dual task, four sources (domains 1, 3, 4 and 5) consisting of spatial processing and attention areas (Table 1) were localized within the parietal and posterior cingulate cortices. It is interesting that spatial processing and attention are related to the onset of the perturbation and this may indicate that these areas are related to the simultaneous organization of visual-spatial change detection and the spatial coordination of reactive postural control during a dual-task paradigm. Previously, an fMRI study, aimed at elucidating the neural correlates mediating parallel processing of voluntary ankle motion and simultaneous executive function (Johannsen et al., 2013) reported that the inferior frontal gyrus likely facilitates dual-task processing; whereas another fMRI report found a large distribution of frontoparietal activity during concurrent auditory and visual cognitive tasks (Deprez et al., 2013). Similarly, for the dual-task condition in our study, event-related cortical sources were estimated within the frontal, parietal, cingulate, occipital and temporal cortices, indicating several cortical regions are related with the perturbation when multiple tasks are performed in parallel. As such, cortical sources corresponding to the onset of a whole-body perturbation are diverted in the presence of increased cognitive demand from mostly frontoparietal areas to include other parietal, as well as temporal and occipital sources.

Interestingly, during the single task involving postural recovery alone, cortical sources were predominantly estimated within the left hemisphere; whereas in the dual task it was found predominantly in the right hemisphere (Table 1). Serrien et al. (2006) discuss shifts in the dominance of the two hemispheres during different tasks, and note that the left hemisphere is considered to be dominant for motor behavior in right-handers. However, they also state that lateralization of motor function is indeed flexible and driven by several factors. The authors (Serrien et al., 2006) suggest the right hemisphere specializes in spatial functions, including spatial attention, and propose that the relative involvement of each hemisphere in a task depends on that task’s specific characteristics. In the current study, adding the secondary VWM task to the primary motor task (reactive postural control), could shift activity from the left to the right hemisphere, as the task shifted from primarily motor (left hemisphere) to one which also involved visual-spatial attentional processing (right hemisphere).

In summary, the present study estimated that the pre-, primary and supplementary motor areas—albeit to a lesser extent for the dual task—and somatosensory areas are primary cortical sources related to reactive postural control. For the single task, the cortical sources were mainly located within the frontal and parietal areas; whereas for the dual task, cortical sources included the frontal and parietal lobes and shifted to other locations including the temporal and occipital lobes. There was also a reduction in the mean absolute N1 ERP peak amplitude in the dual compared with single task. Although biased towards females, the current findings indicate that when attention is divided by performing a VWM task, resources within the brain are re-allocated in relation to the onset of a whole-body perturbation. Thus, we emphasize here that reactive postural control involves distinct electrocortical dynamics depending on the situation (i.e., single vs. dual task).

MDB contributed to the drafting, revising, analysis and interpretation of the data, and approval of the work and is accountable for all aspects of the work. PIB contributed to the analysis, interpretation of data, critical revision and approval of the work and is accountable for all aspects of the work. CEL contributed to the acquisition, interpretation of data, critical revision and approval of the work and is accountable for all aspects of the work. MHW and BHD contributed to the conception, design and interpretation of the data of the work. MHW and BHD revised, approved and is accountable for all aspects of the work.

BHD was supported by a New Investigator Grant from the Medical Research Foundation of the Oregon and Health Sciences University Foundation.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.