The effect of auditory cues on heading direction during stepping-in-place in healthy adults with experimentally induced vestibular asymmetry Authors: Assan Mary Cedras, Timothy Everett Hullar, Philippe Gourdou, Cyril Duclos, Karim Jamal, Daniel Paromov, François Champoux, Maxime Maheu Published: 2026 ABSTRACT AbstractVisual, vestibular, and proprioceptive cues are known to influence the orientation of our body in space. Auditory cues also influence spatial orientation, as recent studies shown they affect static postural balance. However, the influence of auditory stimulation on dynamic postural balance tasks, such as heading direction during vestibular perturbation, remains understudied. To examine the influence of auditory cues on heading direction during a stepping-in-place task with and without galvanic vestibular stimulation (GVS). Twenty young healthy participants performed a stepping-in-place task, blindfolded with the head tilted forward to produce a GVS-evoked yaw rotation signal, with instructions to maintain the starting body position/orientation. The task was repeated under two auditory conditions (no sound or with sound) and three vestibular conditions (NoGVS, anode left, and anode right). Participants were divided equally according to instructions regarding the sound source position. Pelvis angular deviation relative to the sound source was measured using motion capture. The angular deviation error at the end of the trial and its variability with and without sound were calculated and analysed. Without GVS, auditory cues failed to reduce the angular deviation error of the pelvis rotation and its variability. Additionally, left and right GVS significantly increased pelvis angular deviation toward the anodal side compared to NoGVS. However, GVS-induced angular deviation did not differ with and without sound. Even though vestibular cues have been shown to significantly contribute to orientation in the absence of vision, auditory cues could not help maintaining heading direction during a stepping-in-place task, both under normal conditions and following acute vestibular perturbation.Keywords: Dynamic postural control, Audio-vestibular interaction, Galvanic vestibular stimulation (GVS) Introduction Precise orientation of our body in space is crucial for several daily tasks such as maintaining balance (orientation of body relative to gravity) and navigation (maintaining heading direction). The contribution of proprioception, vestibular, and visual cues to both tasks has been well studied (Mergner et al. 2003; Bove et al. 2002; Fitzpatrick et al. 2006; Peterka 2018). However, the contribution of other sensory cues, such as auditory cues to balance and navigation, remains controversial (for recent reviews: Carpenter and Campos 2020; Zarei et al. 2022). One reason for this controversy could be due to the lack of control evaluation regarding the integrity of the vestibular system, as most data suggest the influence of sounds on postural control is greatest in people with underlying vestibular deficits and postural instability (Stevens et al. 2016; Maheu et al. 2019; Lubetzky et al. 2025). An additional possible explanation for the variability observed in previous results is the effect of prior knowledge relative to the availability and characteristics of the sound source. For example, it has been suggested that gait synchronization with auditory cues only occurs when participants are instructed to synchronize their gait to the auditory stimuli before the task. In contrast, gait synchronization does not occur spontaneously when participants are not informed to synchronize their steps (Mendonça et al., 2014). This suggests a possible cognitive influence on the benefit of auditory cues. As prior information on auditory cues was not controlled in previous studies investigating the compensatory role of auditory cues on static postural control and navigation, it remains unclear if it could modify behavioral response. Many previous studies have focused on standing balance, but as we are mostly navigating in our environment in everyday life, a better understanding of how auditory and vestibular cues contribute to improved spatial body orientation during ambulation is critical to optimize substitution following vestibular impairment. Interestingly, only a few studies investigated the influence of auditory cues during ambulation tasks such as the Fukuda (stepping-in-place) task (Zhong et al., 2013; Munnings et al. 2015; Seiwerth et al. 2018; Karim et al. 2018; Paromov et al.,. 2025). These supported an influence of spatially fixed auditory cues on heading direction in normal healthy individuals, i.e., less deviation of head (Zhong et al., 2013; Seiwerth et al. 2018) and body (Karim et al. 2018) relative to starting position in the presence of auditory cues compared to the conditions without auditory cues. Yet, these studies focused solely on participants with nominally intact vestibular function. Presumably, patients with vestibular losses might be expected to benefit much more from the auditory cues based on previous studies on patients while standing (Hegeman et al., 2005; Dozza et al. 2007; Stevens 2016; Maheu et al. 2019). Unfortunately, some patients with vestibular loss also present concomitant sensorineural hearing loss, thus affecting their ability to localize sound, which is essential for the beneficial effect of sound localization on heading direction (Karim et al. 2018). Galvanic vestibular stimulation (GVS) is a technique that allows to modulate vestibular afferents without significantly modulating the peripheral auditory function (Ueberfuhr et al. 2017; Kwan 2019). Moreover, to adequately explore the fundamental mechanisms underlying audio-vestibular interaction, GVS offers the opportunity of conducting an experimental rather than an observational study. It also allows for the control of possible confounding variables such as adaptation mechanisms or other concurrent deficits associated with vestibular asymmetry in clinical population. These reasons make GVS an interesting method to selectively modulate vestibular cues. In the context of sensorimotor tasks, variability has been shown to reveal subtle effects of sensory manipulations that may not be apparent in average performance measures. Indeed, accuracy and precision are dissociable concepts such that movements can be unbiased on average (accurate) while remaining highly variable across trials (imprecise) (Schmidt et al. 1979; Harris and Wolpert 1998). Importantly, trial-to-trial variability has been shown to reveal effects of sensory manipulations that are not apparent in mean performance measures (Newell et al. 1993; van Beers et al. 2004). For example, the Fukuda stepping task is well known for its intra-individual variability even in healthy adults, demonstrating a near 0-degree deviation angle (Hemm et al. 2023). One can therefore hypothesize that auditory cues, by providing a stable external spatial reference, could not only influence the orientation error (accuracy) but also reduce trial-to-trial variability (improve precision), indicating more consistent motor output. The main objective of the present study was to examine the influence of auditory cues on heading direction when vestibular cues are intact or perturbed. The first specific objective was to determine the influence of auditory cues when vestibular cues are not perturbed. The hypothesis was that the presence of auditory cues on heading direction would reduce variability in the angular deviation as observed in previous studies on the Fukuda stepping-in-place task. A second specific objective was to determine whether auditory cues and prior knowledge of the exact position of the sound source in the instructions given to participants could compensate for the angular deviation induced by GVS. The hypothesis was that the reduction in the angular deviation would be significantly greater in the presence of auditory cues and when the sound source position was specified in the instructions. Methods Participants A total of 20 normal healthy adult participants were recruited. Participants did not report any history of dizziness, vertigo, or imbalance and reported subjectively good hearing (Table 1). All participants were classified as right-handed except two participants in group 2, who were considered mixed handed using the Edinburgh Handedness Inventory (Oldfield 1971). The preferred foot was the right for every participant, except for two who preferred the left. The study was approved by the ethics committee of our institution (CRIR-1473-0220). All participants provided written informed consent prior to testing. Table 1Demographic data of each group (with and without prior knowledge). No significant differences were observed for any of the demographic parametersGroupnAge (yrs; ±std)F/MWeight (kg; ±std)Height (cm; ±std)Handedness score (score; ±std)Group 1: Without prior knowledge1023.2(±2.9)6/466.26(±15.4)168.1(±8.0)95(±12.0)Group 2: With prior knowledge1023.6(±2.2)6/469.39(±6.8)172.6(±8.4)72.5(±40.7) Demographic data of each group (with and without prior knowledge). No significant differences were observed for any of the demographic parameters Protocol Stepping task Each subject performed a stepping-in-place task (50 steps, arms at the sides), blindfolded with the head tilted forward (chin to chest) on a large vinyl tile placed on a wooden subfloor room with instructions to maintain the starting position/orientation (Fitzpatrick et al. 2006; Paquet et al. 2016; Belluscio et al., 2018; Seiwerth et al. 2018). The GVS induced a compensatory rotation (yaw) to the anodal side while walking with the chin tucked to the chest which is best suited for a heading direction task compared to a compensatory tilt, which would be more suited for static balance tasks, if head was kept straight (Fitzpatrick et al. 2006). Participants had to step in place with a metal tube (height: 31 cm; diameter: 11.5 cm) between the feet, the tube being affixed vertically on an MDF board (124.5 cm X 150 cm) on the floor, (Fig. 2A). This tube provided somatosensory cues preventing A/P and M/L displacements but did not provide significant rotation cues (Paquet and Maheu 2023). Schematic representation of subjects’ position during task relative to the sound source (A) and representation of angle measurement relative to the sound source (B) In the beginning, participants were familiarized with the task and then trained to step in place to the beat of a metronome at a stepping cadence set at 94 steps/minute. Participants were asked to maintain that cadence throughout conditions to reduce variability in stepping cadence and to provide a comfortable cadence to reduce fatigue across multiple trials. For the familiarization trials, before intervention, they performed the task three times without any vestibular or auditory stimulation. Auditory cues A sound source (Micro Soundlink, Bose, Massachusetts, USA) was located at ear level, two meters in front of the participant’s starting position. The sound source emitted a continuous click train (equal energy in the frequency region between 20 Hz and 6 kHz, with a decrease of energy of 5 dB for every successive increase of 3 kHz) calibrated at 60 dB SPL and was turned on/off before trial depending on auditory condition. This sound was designed to help participants identify the sound’s direction (Shayman et al. 2024). Vestibular cues (stimulation) GVS stimulus was a 1 mA direct current stimulation applied to the mastoid processes using 3cm2 carbon-rubber electrodes (Fitzpatrick et al. 2006). The GVS stimulator (NeuroConn, Germany) was fixed in a backpack on the participant. For GVS conditions, the GVS was started by the experimenter before the start of the stepping-in-place task and was turned off immediately after the completion of the task. Experimental conditions and procedure Three blocks were performed, each comprising six experimental conditions that were presented in a pseudo-random order (Fig. 1). The six experimental conditions were a combination of two auditory conditions with sound (sound source turned on) or without sound (earplugs and the sound source turned off) and three vestibular conditions (NoGVS; GVS anode to the right mastoid; GVS anode to the left mastoid). After each trial, participants remained blindfolded and were rotated back to the starting point following a random path to avoid learning effect. Representation of an experimental block. Each block comprised six conditions resulting from the combination of three GVS conditions (NoGVS, Anode Left, Anode Right) and two auditory conditions (NoAudio, Audio). The six conditions were presented in a randomized order within each block for each participant As an additional control, to prevent possible influence of instructions, participants were equally divided, randomly, into two instruction groups (Group 1 -without prior knowledge; Group 2-with prior knowledge). Both groups received the same instructions regarding the task, but the instructions given to both groups diverged regarding the sound source. Group 1 received no specific instructions regarding the sounds, apart from being informed that some trials might include auditory stimuli. On the contrary, Group 2 was informed about the sound source (position in space and spatial stability) and asked to maintain their starting position, which is facing the fixed sound source. Instructions for both groups were read out loud to the participants at the beginning of the study, ensuring that each participant received the same instructions. No additional instructions other than the pre-determined script were given to participants. Data collection Angular deviation of the pelvis in the axial plane relative to the sound source was measured during each trial. To do so, twelve infrared markers were placed on the pelvis (4 triangular rigid bodies with 3 non-collinear markers each) and six on the head (2 triangular rigid bodies with 3 non-collinear markers each). Their 3D position was recorded using 4 Certus cameras (Northern Digital Inc, Waterloo, Canada) at a sampling frequency of 60 Hz. In addition, anatomical landmarks on the pelvis and head were identified using a rigid body probe to establish their correct orientation (Fig. 2B). Additional markers (n = 5) were placed on each foot to record stepping-frequency, number of steps (50), and to divide recordings into stride cycles (25). Data Analysis To calibrate the initial position relative to the exact laboratory references, the initial angular position of the first step cycle was subtracted from each data point so that the initial angular position at the beginning of each trial was zero (Frissen et al., 2011). The median angular deviation of the last step cycle was calculated across three trials and used for analysis. To test the hypothesis that variability would be reduced when auditory cues are available, in the condition without GVS, inter-trial variability of each individual was calculated separately for each group and for each auditory condition as follows: where represents the angular deviation in the trial j for the participant i, and n are respectively the mean and the total number of the trials done by the participant, respectively. All raw data were processed in Matlab (R2020a) using custom scripts. Statistical analysis Repeated-measures ANOVA was used with Group (Group1, Group2) as a between-subject factor and Audio (Audio, No-Audio) and GVS (NoGVS, GVS-L, GVS-R) as within-subject factors. The model was fitted using MATLAB’s repeated measures. Normality of the data was assessed using Kolmogorov–Smirnov (Lilliefors) tests applied to difference scores (Audio − NoAudio) within each GVS condition. To overcome normality assumptions, permutation testing was applied to all ANOVA effects (Ludbrook et al., 1998). Within-subject effects (Audio, GVS, and their interaction) were assessed using parametric repeated-measures ANOVA. Group labels were randomly permuted across participants (5000 permutations), the ANOVA was recalculated for each permutation to generate an F-value distribution from which p-value was computed. Permutation testing was restricted to the between-subject factor (Group), such that group labels were reassigned while preserving the repeated-measures structure. Post hoc analyses were performed for significant main or interaction effects, using independent samples or paired permutation tests as appropriate, with Holm correction for multiple comparisons. Within-subject effects were further explored using paired permutation post hoc tests. Results Mean angular deviation (accuracy) Repeated-measures ANOVA revealed a significant main effect of GVS, indicating that mean angular deviation at the end of the gait cycle differed across vestibular stimulation conditions [F(2,36) = 111.24, p < 0.001]. A significant Group × GVS interaction was also observed [F(2,36) = 4.89, p = 0.013], demonstrating that the effect of GVS on angular deviation differed between groups. In addition, the main effect of Group was marginally significant [F(1,18) = 4.45, p = 0.049], and this effect was confirmed using permutation testing (pperm = 0.047). Normality tests did not indicate significant deviations from normality (all p ≥ 0.18), supporting the use of parametric repeated-measures ANOVA. Permutation ANOVA revealed a significant main effect of GVS, indicating that mean angular deviation at the end of the gait cycle differed across vestibular stimulation conditions [F(2,34) = 117.89, pperm = 0.012]. A significant Group × GVS interaction was also observed [F(2,34) = 6.40, pperm = 0.014], demonstrating that the effect of GVS on angular deviation differed between groups. In addition, a significant Group × Audio × GVS interaction was found [F(2,34) = 4.05, pperm = 0.047]. The main effects of Audio [F(1,17) = 0.99, pperm = 0.889] and Group [F(1,17) = 3.99, pperm = 0.060], as well as the Group × Audio interaction [F(1,17) = 0.47, pperm = 0.509], were not significant. Post-hoc permutation tests (Holm-corrected) showed that, within each group, mean angular deviation differed significantly between all pairs of GVS conditions (all pₚₑrₘ ≤ 0.010), confirming a strong effect of vestibular stimulation. However, post-hoc comparisons between groups at individual GVS conditions did not reveal robust differences after correction. In addition, no reliable Audio vs. No-Audio effects were observed within individual GVS conditions. No significant differences between groups were observed in the effect of Audio (Audio − NoAudio) at any GVS condition. ». Inter-trial variability (precision) Permutation ANOVA conducted on the inter-trial standard deviation of angular deviation revealed no significant main effects or interactions involving Audio or GVS (all p > 0.47). Repeated-measures ANOVA similarly indicated no significant main effects of Audio [F(1,18) = 3.20, p = 0.09], GVS [F(2,36) = 0.76, p = 0.48], or interactions involving these factors (all p ≥ 0.28). The main effect of Group showed a non-significant trend [F(1,17) = 4.24, p = 0.055], indicating a tendency toward group differences in response variability that did not reach statistical significance under permutation testing. This effect was significant in the parametric ANOVA [F(1,18) = 5.79, p = 0.027] and was confirmed by permutation testing (pperm = 0.026), indicating a difference in variability between groups. Normality tests (Kolmogorov–Smirnov) applied to the Audio effect within each GVS condition did not indicate significant deviations from normality (all p ≥ 0.066), although some values approached the significance threshold, supporting the use of parametric repeated-measures ANOVA. Post hoc permutation analyses revealed no significant differences between groups in the Audio effect (Audio − NoAudio) at any GVS condition (all pperm = 0.70), and no significant within-group differences across conditions Mean values and corresponding 95% bootstrap confidence intervals for both accuracy and precision measures are shown in Fig. 3 and 4. The time series data was also presented in Fig. 5. Representation of the performance accuracy of the participants using the angular deviation for group 1 during the auditory conditions (NoAudio; Audio) and the GVS conditions (GVSL; NoGVS; GVSR) (A). Representation of the performance accuracy of the participants using the angular deviation for group 2 during the auditory conditions (NoAudio; Audio) and the GVS conditions (GVSL; NoGVS; GVSR) (B). Representation of the performance accuracy of the participants using the angular deviation during the three GVS conditions when averaging across the two auditory conditions (C). Mean angular deviation differed significantly between all pairs of GVS conditions, regardless of auditory cues, suggesting a strong impact of vestibular perturbation on the task (pperm ≤ 0.010).Individual subject data are shown as jittered points to illustrate inter-individual variability Representation of the precision performance of the participants on the variance of the angular deviation for group 1 during the GVS conditions (GVSL; NoGVS; GVSR) and the auditory conditions (NoAudio; Audio (A). Representation of the precision performance of the participants on the variance of the angular deviation for group 2 during the GVS conditions (GVSL; NoGVS; GVSR) and the auditory conditions (NoAudio; Audio) (B). Representation of the precision performance of the participants on the variance of the angular deviation during the three GVS conditions when averaging across the two auditory conditions. No significant differences were obtained between conditions and groups (C). Individual subject data are shown as jittered points to illustrate inter-individual variability Time series of the angular deviation of each participant (combined groups) for the auditory conditions separately (NoAudio; Audio) during the GVS conditions (GVSL; NoGVS; GVSR). Each color represents a participant. Solid lines represent group 1 and dashed lines represent group 2 Discussion This study investigated the influence of auditory cues with or without GVS during a stepping in place task. The results showed that auditory cues were not sufficient to reduce accuracy or precision during the stepping in place task, both in the presence and in the absence of GVS. Our results regarding the benefits of auditory cues without peripheral vestibular system perturbation do not support previous work in the literature (Zhong et al., 2013; Seiwerth et al. 2018; Karim et al. 2018). In those studies, authors reported less absolute angular deviation during a Fukuda stepping task when auditory cues were present as opposed to the absence of auditory cues (Zhong et al. (2013): 5.6º vs. 23.6º; Seiwerth et al. (2018): 14.2º vs. 28.3º, Karim et al. (2018): 3.8° vs. 31.6º). However, a major consideration is the fact that, unlike previous studies, we analyzed raw angular values, which preserved the directional information of the angular deviation. Moreover, this difference between previous results and the present study could also be explained by the fact that in our methodology, participants had to maintain the head in a nose-down pitch position as opposed to a head straight position. In fact, the ability to localize a sound source in this position is significantly reduced. Karim et al. (2018) demonstrated that the ability to adequately use the spatial landmarks represented by sound sources is based on the minimal audible movement angle (MAMA) performance, which is known to be two to four times worse in the elevation plane as compared to horizontal plane (Wettschureck 1973; Morrongiello and Rocca 1987). This head position could also account for the tendency observed in Fig. 3, where participants in group 1 seem to perform better as opposed to when there is a greater focus on sound source in group 2 during the NoGVS condition. Indeed, by explicitly directing participants’ attention toward the sound source, those in group 2 may have relied more strongly on auditory cues even in situations where these cues were less reliable, rather than vestibular and proprioceptive estimates of heading direction, thereby potentially leading to systematic biases. In contrast, participants in group 1 may have relied more consistently on vestibular and proprioceptive cues, which could account for their relatively better performance in the NoGVS condition. Since GVS requires placing the head in a nose-down pitch position to effectively perturb heading direction (as opposed to postural control balance) in the yaw plane, future studies will have to assess different sound source position that will not modify elevation to keep sound source in the azimuth plane, such as positioning it to the side or aligned with participant’s forward-facing (i.e. spatially fixed sound under headphones). Also, our study is the only one restraining participants’ movement in M/L and A/P directions. It is improbable given its shape that the tube itself provided any valuable somatosensory cues relative to rotation. In support, Paquet and Maheu (2023) revealed no difference in rotation angle when comparing free to move Fukuda as to the method described here (tube). However, limiting A/P and M/L displacement may have limited auditory cues useful to detect MAMA (Owen Brimijoin and Akeroyd 2014). The present study adds to the existing literature by being the first study to control simultaneously for vestibular and auditory cues during a stepping in place task. The results showed not only that vestibular cues are of great importance to maintain body orientation, but that auditory cues could not compensate while vestibular signals are acutely perturbed during the stepping-in-place task. One of the hypotheses to support the absence of auditory compensation during perturbed vestibular cues is that updating auditory spatial position requires the integration of multiple congruent sensory cues. As proposed by Genzel et al. (2016), auditory spatial updating highly depends on the vestibular signal, but also on proprioceptive/efferent signal. In their study, they demonstrated that participants most accurately updated auditory spatial representation while head and trunk were passively rotated but overestimated auditory position when head-on-trunk was actively moved (proprioception and vestibular cues) and underestimated auditory position when proprioception/efferent were isolated (cancellation trials). In our study, when GVS was present, a passive vestibular signal produced an illusion of rotation in the yaw plane. However, no cues from proprioception nor from efferent copy could be used to update spatial auditory representation as the head was pitched forward. Additionally, the asymmetry induced by GVS passively modified the body’s orientation relative to the sound source. Participants were unaware of their movement with respect to the sound source. More surprisingly, all participants, including those informed that the sound source was spatially fixed, reported perceiving the sound as moving. The absence of a significant effect of group (instructions) may be because participants in group 2 relied more on their perception that the sound source was moving than on the information provided in the instructions. This observation echoes a recent finding from our research team (Cedras et al. 2025). In that study, participants had to perform a sound source localization task under earphones during GVS. The results showed that participants exhibited a greater error ratio in localizing the sound source when GVS was applied as opposed to no vestibular stimulation. Therefore, it would be interesting for future study to evaluate whether this observation can be generalized to more naturalistic navigation tasks, such as walking. One last hypothesis could be linked to the interactions and connectivity between brain regions that process both auditory and vestibular signals. In fact, Bucher et al. (1998) were interested in the cortical activation following GVS in neurotypical population. Among the different areas that are activated, the results showed that some areas that are involved in auditory processing, such as the bilateral cortical activation in Brodmann area 22 (BA 22) located in the superior temporal gyrus, and Heschl’s gyrus was also activated. Therefore, vestibular perturbation appears capable of altering the attribution of motion between the environment and self by enabling the brain to maintain the stable characteristic of the spatial auditory stimulus. Despite the non-significant differences between instruction groups and anodal position, there seems to be a tendency that group 2 is less influenced by the anode right condition compared to the anode left. One possible hypothesis could be related to the different cortical influence relative to hemispheric vestibular dominance that may have led to different cognitive biases relative to the knowledge of the spatially fixed characteristic of the sound source (Bense et al., 2001). Indeed, it has been demonstrated that anode right condition induces a bilateral activation of the vestibular network, which may have amplified the weighting of external spatial cues, including auditory landmarks (Eickhoff et al. 2006). Therefore, this tendency of a differential anodal influence on the use of auditory landmarks should be further explored by future studies. Moreover, several limitations should be acknowledged. First, the small number of participants that composed the two groups could reduce the statistical power of the study. Even though a significant effect of GVS was found following appropriate statistical analysis, this limitation is particularly relevant given the substantial inter-individual variability commonly reported in spatial orientation and auditory cueing tasks. Secondly, although we implemented several measures to ensure normal hearing and vestibular functions in our cohort, such as the recruitment of young healthy participants, questionnaires regarding the absence of auditory and vestibular symptoms, neither normal hearing nor normal vestibular function in the groups could be confirmed, as no objective vestibular and hearing assessments were performed. Finally, the findings should be cautiously extrapolated to the population with vestibular loss. As mentioned earlier, GVS, which is composed of two electrodes placed on the mastoid, stimulates the vestibular system bilaterally. The cathode electrode activates the afferent from all five sensory end organs on one side, and the anodal electrode inhibits the afferent on the opposite side, thereby bypassing the mechano-transduction of both canal and otolith organs, which contrasts with natural head movements (Dlugaiczyk et al. 2019; Lopez et al., 2024). Consequently, it would be interesting to evaluate the contribution of auditory cues during the stepping task in vestibular patients who present no concomitant hearing loss. Conclusion The present study demonstrates that a spatially fixed sound source may be insufficient to reduce either the angular heading direction error or its variability during passively induced vestibular asymmetry in healthy participants while stepping in place, likely because the vestibular perturbation may have altered the perceptual stability of the auditory reference. Future studies should assess the impact of vestibular perturbation on spatial auditory processing and compare the difference between active and passive movements. Acknowledgements The authors gratefully acknowledge Robert J. Peterka for his valuable insights into the study’s methodology and analytical approach. Author contributions Designed research: MM, TEH Performed research: AMC, PG, MM; Analyzed data: All authors.; Wrote the paper: All authors. Funding Natural Sciences and Engineering Research Council of Canada, Canada- RGPIN-2022-04402. Data availability No datasets were generated or analysed during the current study. Declarations Conflict of interest The authors have no relevant conflicts of interest to disclose. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.