A Primary Central Source Determines Perturbation-Evoked N1 Amplitudes in Younger Adults

Abstract

The balance perturbation-evoked N1 potential is a reliable cortical response during reactive balance control that is correlated to a variety of cognitive and motor functions. Although the supplementary motor area (SMA) has been identified as the primary source of the N1, it is less understood whether other brain regions contribute to the N1 recorded at the scalp. We used source localization on electroencephalography (EEG) data from 25 younger adults recorded during backward whole-body perturbations during stance. We identified the sources that contribute to channel-based N1 recordings and quantified their impact on N1 amplitude and latency. In younger adults, N1 amplitudes can be explained by one single source in a central midline cortical region covering the SMA. When reconstructing N1 signals using backprojections with one versus all independent components (IC) identified as brain sources there was no difference in peak amplitudes and a small but significant difference in N1 peak latencies. Parallel brain sources thus deflect the time course of the N1, but not its magnitude. Brain areas associated with ICs contributing to the shift in N1 latency varied between participants. Our results emphasize the dominant influence of central cortical areas on the N1 response, informing hypotheses regarding the nature of the signal and its functional role. Importantly, the extent and location of other cortical structures that influence N1 timing, such as parietal cortex areas and the anterior cingulate cortex, may further elucidate cortical contributions to balance. These markers could be crucial for the early detection of balance problems in clinical populations.

NEW & NOTEWORTHY

We demonstrate that channel-level amplitudes of the balance perturbation-evoked N1 in younger adults primarily reflect neural activity originating from cortical central midline regions, particularly the supplementary motor area (SMA). In contrast, contributions from parallel active brain regions evoked by balance perturbations are indicated by an effect on N1 peak latencies. Our findings imply that the perturbation-evoked N1, unlike other evoked potentials, is not a mixture of multiple neural sources in younger adults.

Graphical Abstract

INTRODUCTION

Electroencephalography (EEG) data offer precise temporal measures of cortical activity during balance control, with the disadvantage of poor spatial resolution. The signals measured at the electrode level reflect a combination of volume conducted brain sources, as well as non-brain activity (e.g., from muscles or external artifacts) (1,2). Alterations in cortical engagement patterns underlying reactive balance control may foreshadow balance impairments associated with aging or cortical pathologies [see (3,4) for overview articles]; therefore, a detailed analysis of the relative contributions of cortical structures to well-studied perturbation-evoked scalp potentials is important. The most prominent perturbation-evoked scalp potential is the N1, which is a large negative deflection recorded over central EEG electrodes 100 to 200 milliseconds after externally elicited perturbations to the body, e.g. support surface translations during stance [see (5,6) for overview articles]. Although the N1 has previously been primarily localized to the supplementary motor area [SMA, (7,8)], the extent to which other brain sources contribute to the N1 remains unclear.

Standing balance perturbations are usually salient events that move the whole body, triggering rapid balance-correcting actions to an abrupt external event. Although perturbations can cause multisensory signals, including vestibular and visual information, proprioception may play a more dominant role (9). However, the N1 is most likely not dependent on one modality, as it has been detected when participants have their eyes closed (10), and in a patient with bilateral labyrinthine loss (11) or during seated conditions (12,13). The N1 is further a highly reliable signal that is typically detected in every participant and on a single-trial basis with prominent amplitudes (8,10,14) that can scale with perturbation intensity (14–16). The characteristics of the exogenous physical perturbations alone are, however, insufficient to explain the observed variability in N1 characteristics. Typically, the amplitude is largest on the first trial of an experiment (8), smaller for predictable than unpredictable perturbation (17), and attenuate with repetitions (8,14,18), suggesting a modulatory influence of top-down processes.

The involvement of the SMA is an indication that the N1 may reflect both bottom-up and top-down neural processes due to its connections to prefrontal and cingulate areas (19) and direct projections to the spinal cord (20,21). Connections between the SMA and the basal ganglia have further put this brain region in the focus of research centered on the motor deficits in Parkinson’s disease (22). Functionally, the SMA has been associated with initiation of internally driven movements or changes in movement plans [see (21) for an overview article], e.g. when an external cue signals that an already initiated movement has to be altered. However, previous studies that have localized the main source of the N1 to Brodman area 6 or specifically to the SMA (7,8) lack a detailed assessment of how much other brain areas influence the N1 potential recorded at channel level. Consequently, it is not known whether the N1 reflects a mixture of multiple parallel active neural sources or a signal that can be primarily attributed to the SMA, complicating the functional interpretation.

Here, we reanalyzed data from a sample of younger adults and used independent component analysis (ICA) to describe brain sources contributing to the perturbation-evoked N1, recorded at channel level. We hypothesize the large negative deflection measured at central scalp locations following perturbations is a signal predominately reflecting SMA activity. We predicted that the time course of activation at the central midline channel with the most prominent N1 would be highly correlated with the time course of activation of the backprojected N1 using the primary contributing independent component (IC) and that this IC would be predominately located in brain regions covering the SMA. We further explored the effects of concurrent activity of other identified sources on the amplitude and timing of the N1 response.

MATERIALS AND METHODS

Participants

Data from 25 healthy younger adults (12 female, mean age: 24 years, SD: 3.9 years) were reanalyzed for the present report. Initial results from this sample have been reported in Mirdamadi et al. 2024 (18) (n=15) and Mirdamadi et al. 2025 (10) (n=10). Written informed consent was given by all participants and the study protocol was approved by the Emory University Institutional Review Board.

Experimental task and setup

For the current analyses, we extracted data from a whole-body motion-perception task in which participants were asked to judge whether two consecutive backward perturbations were in the same or different directions. We only considered the first perturbation of each perturbation pair for our analysis, which was always a straight backward perturbation. These perturbations had a consistent characteristic (7.5cm displacement, 15cm/s velocity, 0.1m/s² peak acceleration), but were delivered at unpredictable time intervals to minimize predictability and adaptation.

EEG recordings and preprocessing

EEG data were recorded continuously with a 64-channel EEG system (ActiCHAMP, Brain Products GmbH, Gilching, Germany) at a sampling rate of 1000Hz. All electrodes were active amplified and arranged according to the extended international 10%-system (23). A ground electrode was positioned at location AFz and the reference channel at Fz. The raw EEG data sets were preprocessed using custom Matlab scripts and the EEGLAB(24)-based BeMoBIL pipeline for mobile brain/body imaging data (25). The timing of the perturbation was determined by the onset of platform acceleration, which was extracted from an accelerometer attached to the platform. The accelerometry data were recorded synchronously with the EEG data through an onboard auxiliary channel of the EEG amplifier.

Basic preprocessing of the EEG data included the assignment of standard coordinates to each electrode label and the removal of line noise (26,27). Noisy channels were identified from filtered data sets using the clean_artifacts function of the EEGLAB clean_rawdata plugin with a [0.25 0.75] transition band high pass filter, a channel correlation criterion of 0.7 as a minimal correlation coefficient and a maximal time window of 300ms that had to be exceeded to flag a channel as bad. Since this function uses a random data sample, which can produce unstable results, the channel rejection procedure was repeated twenty times. Only channels that were flagged in more than a third of all iterations were rejected from the originally minimal filtered sets. Subsequently the rejected channels were reconstructed using spherical interpolation and all channels were re-referenced to a common average, and the data from the original reference channel Fz was reconstructed.

For source localization, preprocessed data sets were rank-reduced by the number of interpolated channels, high-pass filtered with a cutoff frequency of 0.75 Hz (28) and epoched to a time window of 2s before to 2s after each perturbation. Subsequently each data set was decomposed using adaptive mixture independent component analysis (AMICA) (29,30). 10 iterations of the AMICA build-in functionality that removes data samples with a log-likelihood lower than 3 SD were used to improve data quality (30). IC dipole locations were fitted using the EEGLAB DIPFIT toolbox and a standard boundary element method (BEM) head model (31).

The obtained AMICA weights and dipole models were extracted and projected to the initial basic preprocessed continuous data sets with a high-pass filter with a pass-band edge of 0.2 Hz. In a final data cleaning step, ICs were classified as representing brain activity or as artifact ICs based on the following steps. First, all ICs were rejected if a) the residual variance exceeded 15%, b) the IClabel toolbox (32) (default classifier) classified the IC as representing eye activity, line or channel noise as the category with the highest probability c) if the frequency spectrum had a positive slope (≥ 0.31) which can be indicative of muscle activity (33), or d) if at least one electrode exhibits a z-score > 5, based on the IC topography weights calculated and visualized with the TESA plugin (34). All resulting ICs were visually checked based on their dipole location and corresponding IC scalp maps, power spectrum and the cross-frequency power–power coupling using the powpowcat toolbox (35). On average, 10 ICs (SD=2.6) were identified as brain ICs and backprojected to the channel level.

EEG data analysis

The resulting data sets, containing the channel level data as well as the AMICA weights, were lowpass-filtered with a pass-band edge of 30 Hz. Subsequently, each set was epoched into trials (−1 to 2 s relative to the perturbation onset) and baseline corrected to the pre-perturbation time window of −200 to −50 ms. Trials exceeding a threshold of ±100μV were rejected. Trials for rejection were only detected in five participants (two trials each in two participants; one trial each in three participants), resulting in an average of 61 trials (SD = 2.2) per participant included in the analysis. The cleaned and epoched data sets were used to characterize the reconstructed channel N1 potential for the primary IC contributing to the N1 as well as for the backprojection of all brain ICs. The analysis pipeline described in the following is depicted for two exemplary participants in Fig. 1.

Figure 1.

Analysis pipeline for two exemplary participants: top, A: for a typical participant and bottom, B: for one participant with the highest discrepancy between primary N1 and all brain independent component (IC)-backprojections. A1 and B1: time-activation course for midline electrodes and the N1-peak extraction window in gray; A2 and B2: the individual 50% data extraction window for the N1 brain IC-backprojection to the channel with largest negativity within the N1-time window and the topography for that time window; A3 and B3: the backprojection of each brain IC to the individual channel identified in step 1, i.e. the contribution of each IC to the channel and the IC scalp map and location of the primary IC; A4 and B4: the total backprojection of all brain ICs (red) and the primary N1 IC (blue) to the individual N1-channel, with the N1 time window in gray; A5 and B5: the correlation between all data points within the 50%-window between the traces of the back-projection of all brain ICs and for the primary N1-IC; A6 and B6: the area difference between the curves of the backprojection of all brain ICs in red and for the primary N1 IC in blue in the 50% window.

For the electrode selection, we plotted the time-activation course at the three midline electrodes (FCz, Cz and CPz) from the backprojection of all brain ICs to the channel level and determined the electrode with the highest negativity in the N1 time window between 80 to 200ms after perturbation onset (Fig. 1.A/B.1). We chose the most commonly reported N1 location, Cz, plus the two neighboring midline electrodes, FCz and CPz as our area of interest to account for the inherent spatial imprecision of EEG recordings. Each individual’s N1 time window for analysis was specified as the full width of the negative deflection at 50% of the peak amplitude, referred to as the 50%window (Fig. 1.A/B.2). For each participant, the IC with the highest negativity within this time window was selected as the primary N1 IC with the highest contribution to the potential measured at the N1-channel (Fig. 1.A/B.3).

Peak amplitudes and latency measures were extracted for the backprojection of all brain ICs and for the backprojection of the primary N1 IC (Fig. 1.A/B.4). The MNI coordinates of each individual primary N1 IC were associated with a brain region using the BioImage Suite Web application (https://bioimagesuiteweb.github.io/webapp/mni2tal.html). Approximate brain regions of the second highest contributing ICs are reported as group means. The association of the activation time series of the N1 for both backprojections - the primary N1 IC backprojection and the all brain IC backprojection - within the 50%-window were analyzed using Pearson correlation (Fig. 1.A/B.5). Differences between the curves were further quantified as the area between the curves of the N1-channel for the all brain IC backprojection and the primary N1 IC-backprojection. We integrated the absolute difference between both curves within the 50%-window using the matlab trapz.m function. The area difference is reported as a percentage of the area under curve of the N1 for the all brain IC backprojection (Fig. 1.A/B.6). The mean values and standard deviations of the correlations and area measures are reported to describe the participant-level data.

Statistical analysis

For each participant, the peak amplitudes and corresponding latencies were extracted for statistical analysis using the N1-channel that was reconstructed using backprojections of a) all brain ICs and b) only the primary N1 IC. Differences in amplitudes and latencies were assessed with paired t-tests.

Pearson’s correlation coefficients were used to assess associations between N1 latencies and amplitudes of both backprojections. To further explore the potential association of the significant differences in peak latencies, post-hoc correlations were calculated with the differences in amplitudes and with the differences in area under the curve in the 50%window (expressed as the percentage change from all brain ICs to the primary N1 IC backprojection).

RESULTS

Descriptive subject level results

This section summarizes the descriptive subject level results for the whole sample, which are depicted in Fig. 1 for two exemplary participants: for one participant with results representing the group mean and for the participants with the highest area difference between the two backprojections. All single subject primary N1 ICs and event-related potential traces are shown in Fig. 2.1 and Fig. 2.2. The grand average event-related potentials are shown in Fig. 2.3.

Figure 2.

MNI coordinates for the individual primary N1 independent component (IC) (1); single subject event-related potential traces with all IC-backprojections in red and primary N1 IC-backprojections in blue (x-axis: ms, y-axis: μV) (2); group-average event-related potentials with the shaded region showing the standard error of the mean, the all brain IC-backprojection is shown in red and primary N1 IC-backprojection in blue (3); dots indicate the single subject differences (4) between the all brain IC and the primary N1 IC-backprojection in amplitude (μV) in purple on the left and in green for latencies (ms) on the right, boxplots depict the distribution of the data, with the box representing the interquartile range (25th–75th percentiles) and the central line indicating the median. Whiskers extend to the 2nd and 98th percentiles (4).

Channel Cz was the channel with the highest negativity in the N1 time window for 19 participants; for 5 participants it was channel CPz and for one participant it was channel FCz (see Fig. 1.A/B.1). For the individually determined N1-channel, the average time window spanning the full width of the N1 at 50% of the peak amplitude was 64.32ms (SD = 15.09) and ranged from 39ms to 103ms (see Fig. 1.A/B.2).

The within-participant correlations between the curves of the primary N1 IC and all brain ICs backprojection within the 50%window ranged from 0.24 to 1.00, with a mean of r = 0.90 (SD = 0.16), cf. Fig 1.A/B.5. The average area between both curves was 10.5% (SD = 10.19) of the total area under the N1 of the all brain ICs backprojection, with a range of 2.1% to 50.1% (cf. Fig. 1.A/B.6).

The primary N1 IC was typically localized within Brodmann Area 6, a brain region covering the medially located SMA as well as lateral premotor areas (Table 1 and Fig. 2.1). For 14 participants, the primary N1 IC was located within the central part of Brodmann Area 6 and thus in the part that has been associated with the SMA. For 9 participants, the primary N1 IC was located outside of defined brain areas, but near the central part of Brodmann Area 6 (e.g. within the central sulcus). For one participant, the primary N1 IC was located on the border of Brodmann Area 6 and Brodmann Area 24. The latter comprises the anterior cingulate cortex. For one participant, the primary N1 IC was located within Brodmann Area 4 that covers the primary motor cortex area.

Table 1.

MNI coordinates and associated brain areas for the individual primary N1 ICs as well as the mean of all ICs as the cluster centroid. Parentheses indicate that the coordinates are outside defined areas, e.g. in the central sulcus. The nearest area is indicated within the brackets. BA 4, Brodmann Area 4/Primary Motor Cortex; BA 6, Brodmann Area 6/SMA and premotor areas; BA 24, Brodmann Area 24/anterior cingulate cortex; IC, independent component. MNI (x, y, z): Montreal Neurological Institute (MNI) template x, y and coordinates.

ID	MNI (x,y,z)	Area
1	5, 1, 55	BA 6
2	−1, −3, 66	BA 6
3	−4, −7, 60	BA 6
4	0, −7, 71	(BA 6)
5	−2, −17, 60	BA 6
6	1, 2, 61	BA 6
7	−2, −13, 75	(BA 6)
8	−7, −15, 51	BA 6
9	3, −7, 63	(BA 6)
10	−5, −20, 74	BA 6
11	−2, −13, 63	(BA 6)
12	−1, −11, 60	BA 6
13	2, −8, 57	BA6
14	1, −19, 72	(BA 6)
15	1, −12, 52	(BA 24/ BA 6)
16	−1, −15, 61	BA 6
17	6, −29, 69	BA 4
18	−2, −15, 69	BA 6
19	2, 11, 74	(BA 6)
20	−2, −5, 64	BA 6
21	3, 11, 74	(BA 6)
22	1, 6, 642	(BA 6)
23	−2, −8, 65	BA 6
24	5, 2, 65	BA 6
25	0, −3, 63	(BA 6)
mean	0, −8, 64	(BA 6)

For twelve participants the IC with the second highest negativity within the 50% time window was located in Brodmann Area 24 and 32, i.e brain regions including the anterior cingulate cortex. For three participants, the second IC was located in Brodmann Area 23 and 31; brain regions that comprise the posterior cingulate cortex. For three participants, the second IC was located in Brodmann Area 6, a region that spans the SMA and premotor area. Other brain areas for the second IC included Brodmann Area 8 (frontal eye field) for two participants, Brodmann Area 5 (somatosensory association cortex) for two participants, Brodmann Area 18 (visual association area V2) for one participant, the caudate for one participant and Brodmann Area 7 (precuneus/ superior parietal lobule) for one participant. On average, the second IC peaked 24.2ms (SD: 22.4ms) after the first IC.

Group Level results

The mean N1 peak amplitude of the all brain IC backprojection was −32μV, (SD = 13) and −32μV, (SD = 14) for the primary N1 IC backprojection (t(24)=−0.308, p = .761), with a mean difference of −0.11μV, (SD = 1.82) (Fig. 2.3).

The mean N1 peak latency of the all brain IC backprojection was 157ms, (SD = 11) and 152ms, (SD = 11) for the primary N1 IC backprojection (t(24)= 3.37, p = .003), with a mean difference of 4.60ms, (SD = 6.83) (Fig. 2.4). The effect for latencies remained significant after correction for one participant with a latency difference > 3SD away from the mean. To assess whether the observed 5 ms latency difference was robust to low-pass filtering and IC cleaning, we conducted a post-hoc sensitivity analysis comparing the backprojection of the primary N1 IC to the backprojection of all ICs — including those labeled as non-brain components — onto data filtered only with a 0.2 Hz high-pass filter. We found the same pattern of results, with no significant difference in N1 peak amplitudes, t(24)=−1.17, p = .252, and a significant difference in peak latencies, t(24)=−2.48, p = .020. The mean difference in N1 peak amplitudes was −0.54 μV (SD = 2.29), and the mean difference in peak latencies was 3.76 ms (SD = 7.57).

The correlation between the all brain IC backprojection and the N1 IC backprojection was r(23)= 0.99, p < .001 for the peak amplitudes and r(23)= 0.81, p < .001 for the corresponding latencies (see Fig. 3). The differences in latencies between both backprojections was not correlated with the differences in peak amplitudes (r(23)=−0.37, p = .070), but with the normalized area difference (r(23)= 0.65, p < .001, see Fig. 3). However, the correlation with the area difference was driven by one participant with a value that is more than 3 SD away from the mean. Following exclusion of this participant, the correlation was no longer significant (r(22)= 0.19 p = .380).

Figure 3.

Correlations between the peak amplitudes (in μV) of the all brain ICs backprojection and the primary N1 IC backprojection (upper left); between the peak latencies (in ms) of the all brain ICs backprojection and the primary N1 IC backprojection (upper right); between the differences in peak latencies (in ms on the x-axis) and differences in peak amplitude (y-axis in μV, lower left); between the latency differences (in ms on the x-axis) and area differences (y-axis, lower right). Each color and dot represents an individual participant.

DISCUSSION

We found that the balance-perturbation evoked N1 responses in younger adults reflect almost exclusively activity from a single source located in central brain areas corresponding to the SMA, whereas differences in the peak timing of the N1 indicate parallel activity from a larger cortical network active during reactive balance control. Although a dominant N1 source was identified for each participant, the source of the secondary parallel activity varied between participants. These differences in cortical engagement during whole-body perturbation could be crucial for our understanding of cortical involvement pattern underlying voluntary balance control, particularly for characterizing balance impairments.

N1 amplitudes can be explained by one source in or near the SMA

Our results demonstrate that the balance perturbation-evoked N1 in younger adults recorded at central scalp locations originates from a single source located in or near the SMA. Importantly, we did not find evidence to suggest that other sources influence its magnitude. Using a relatively simple approach, we found one single IC for each participant that explains most of the N1 amplitude. We were able to show that the average N1 peak from the backprojection of one single IC per participant is nearly identical in shape and highly correlated with the N1 extracted from the backprojection of the complete brain IC solution. The sources of the identified ICs that contribute primarily to the channel level response, as measured by the reconstruction of all brain ICs, form a tight cluster around central brain regions associated with the SMA with relatively little spread. Although EEG data can only provide rough estimates of brain source locations, the large N1 amplitude signals presumably led to better signal-to-noise ratios and a relatively stable source localization. In our study, this resulted in a tight group-level IC cluster with a centroid near the central sulcus close to Brodmann area 6. Based on the medial location of the cluster and the MNI coordinates of the individual ICs, we argue that the origin is within the medial SMA part and not in the lateral premotor areas of Brodmann area 6. Despite the inherent limitations in spatial resolution of the EEG data and the absence of co-registration with the individual MRI data in the present study, we observe prevalent sources within the approximate SMA region, which replicates prior results (7,8). Therefore, we propose, that the whole-body balance N1 can provide a direct readout of evoked activity of brain structures covering the SMA during balance perturbations. Previous studies have demonstrated that the balance perturbation N1 can be reliably assessed using a simple configuration consisting of six trials, three channels, and only basic preprocessing, such as high and low pass filtering (10). Our present findings add that the balance perturbation-evoked N1 in younger adults is nearly exclusively generated in one central brain region. Given the negligible influence from other sources on the N1 magnitude, it can be assumed that even with setups that contain fewer electrodes and minimal preprocessing, activity from a brain region comprising the SMA will be predominantly captured.

Parallel activity might be masked by the dominant central N1

The high N1 amplitudes found in the present study, with a group mean around −30mV and up to −69mV for one participant, replicate previous findings (5), and may mask parallel activity at smaller scales. The very large N1 magnitudes are a distinctive attribute of balance-perturbation evoked responses, as event-related potentials typically reflect propagated multi-source activity at lower magnitudes that often only manifests in multi-trial averages. For example, in studies using low to very intense auditory stimuli, average amplitudes have been reported to range from −1 to −15μV [see (36) for an overview article]. The observed differences in magnitude of the balance N1 compared with those evoked by other sensory stimuli are challenging to attribute to factors other than the nature of the stimulus itself, including potential differences in hardware or preprocessing methodologies (e.g., the EEG reference scheme). Whole-body perturbations are triggered by large proprioceptive or cutaneous bursts of activity and trigger brainstem-mediated involuntary balance-correcting responses that counteract the net direction of body motion. Our prior work shows that the N1 may lie along a parallel transcortical pathway that evokes later cortically-mediated muscle activity (37,38) that may include strategy changes such as stepping responses (15,39). Unlike auditory stimuli, balance perturbations have inherent ecological relevance, as failing to respond appropriately can lead to a fall. The relatively large N1 response may thus indicate the potential negative consequences of the stimuli, but may also mask the activation of parallel cortical network activity (40) that manifests at lower amplitudes and with maxima at different scalp locations.

Additional sources shift the timing of the N1

Our data support the previously reported finding that the channel-level, perturbation-evoked N1 in younger adults primarily arises from event-related activity in one central brain region (7,8) whereas additional sources influence N1 peak latencies. Although the practical implications of an average peak difference of less than 5ms might be questionable, this shift indicates that parallel sources are active during balance perturbations. Activity from these sources may not be large enough to deflect the N1 magnitude, but they are potentially within common event-related potential amplitude ranges and can significantly affect the N1 timing. Especially since the functional role of the N1 is still under debate (5,41), contributions of additional sources—active in different time windows and with different scalp maxima — may also contribute to reactive balance responses. Although a detailed analysis of these sources and different processing steps is beyond the scope of this short report, we examined the remaining ICs with the second-highest contribution to the N1. However, we were unable to identify a single, spatially well-defined activity cluster further supporting the primary finding that a single SMA source is the predominant contributor to the scalp N1 in younger adults. Sources with pronounced event-related activity within the N1 time range were distributed from frontal to parietal regions. Several participants in our sample exhibited significant negativities in two sources within the N1 time-window (for an example see Fig. 1.B3). The participant with the lowest correlation between both curves and the largest amplitude and latency deviations revealed one large negative deflection around 150ms after perturbation onset in or near the SMA with the typical N1 timing and location and one smaller negativity around 30ms later in the anterior cingulate cortex. These findings may indicate parallel activity or reciprocal connections between the two structures during reactive balance control. Anterior cingulate cortex activity is consistent with prior work demonstrating increased task-related connectivity between the SMA and the anterior cingulate cortex during lateral pull-perturbations in standing and walking (19). As the anterior cingulate cortex is associated with conflict monitoring and error processing, one might speculate, that this structure is only triggered in participants with balance challenges and trials that require additional cortical involvement beyond regions around the SMA, and an initial brain stem modulated balance response. However, it is important to note that the analyzed data for this report were part of a complex perception task. Participants were asked to determine whether the direction of a subsequent perturbation was in a different direction than the straight-backward perturbations analyzed here. Therefore, it is necessary to determine whether the reported secondary sources are associated with the balance response or potential secondary task demands.

What insights can the source contribution to the N1 reveal about its functional role?

Individual source profiles of parallel activities in combination with the behavioral measure of the balance response may be key for our understanding of the functional role of the perturbation-evoked N1. It is intriguing to argue that the N1 is a top-down modulated response evoked by the large proprioceptive input of a whole-body perturbation. However, based on the functionalities contributed to the SMA (42–45), the N1 could also be interpreted as a motor preparation process for a voluntary response or even as a ramped-up readiness-potential that builds up in parallel to the first brainstem modulated motor response leading up to the onset or inhibition of additional, slower cortical-mediated motor responses. Following this line of reasoning, the N1 could be a sensorimotor threshold correlate (43,46) for the time-sensitive motor preparation for a sudden change of the environment. Importantly, this would indicate that the N1 is not necessarily related to a motor output and thus might not directly relate to behavioral measures as the perceptual and central processing could also result in the inhibition of a cortical modulated response. Examining populations with balance problems may provide further insight into this debate. It has been hypothesized that changes in activation patterns, such as more widespread cortical activation in older adults compared with younger adults performing the same task, are compensatory for age-related declines (47,48). Similarly, shifts in the contributions of different sources to the balance-evoked response may clarify their functional relevance. The composition of sources contributing to the N1 might be a candidate marker for the identification of early balance problems in aging and clinical populations, as the timely involvement of additional cortical structures could reflect a switch from automatic to balance responses requiring cortical compensation.

DATA AVAILABILITY

Data that support the findings of this study are available from the corresponding author [L.H.T.] upon reasonable request.