Different EEG brain activity in right and left handers during visually induced self-motion perception

A total of 25 right handed (14 females, age: 27.68 years SD: 4.02 years) and 25 left handed (18 females, age: 24.83 years SD: 4.24 years) healthy adults participated in the experiment. All participants had normal or corrected-to-normal vision and reported no prior history of vestibular symptoms or neurological disorders. Participants completed the 10-item Edinburgh Handedness Inventory to determine their handedness (right handers: 8% >  + 40, 16% >  + 60, 16% >  + 70, 8% >  + 80, 52% >  + 90; left handers: 8% > − 40, 4% > − 50, 24% > − 60, 20% > − 70, 20% > − 80, 24% > − 90). The experiment was conducted in accordance with the guidelines of the institutional ethics committee and the Declaration of Helsinki. All participants gave their informed written consent prior to their participation and received compensation (€10/h).

The stimuli comprised two movies: a coherent and an incoherent rotating pattern of dots. The stimuli were generated in MATLAB (The MathWorks Inc., Natick, MA, USA) using the Psychophysics Toolbox extensions. The stimuli consisted of 1000 randomly spaced white dots on a black background. A central green dot provided a fixation point. The dots rotated in the roll plane in either a clockwise (CW) or counter-clockwise (CCW) direction at 30°/s. In the coherent condition the dots rotated in a smooth circular formation. In contrast, the dots in the incoherent condition had a random sinusoidal movement in both the X and Y direction added to the overall circular trajectory (i.e. the phase and amplitude of the additional sinusoidal movement was randomized separately for each dot). This resulted in each individual dot appearing to follow a random trajectory, but with the global pattern maintaining either a CW or CCW direction with a mean velocity of 30°/s. The stimuli were projected onto a custom-built dome (diameter: 75 cm) apparatus with the rotation axis passing through the dome centre. The distance between the apex of the dome and the participant’s nasion was 31 cm, with the visual field subtending a visual angle of 100°. The experiment was conducted in a dark room with the visual stimulus covering the participant’s entire field of view, ensuring that no horizontal or vertical cues were observable.

Prior to the beginning of the experiment, the height of the dome apparatus was adjusted such that participant’s line of sight was in-line with the centre of the dome. As illustrated in Fig. 1a each trial began with the presentation of stationary dots for a jittered period (range 3–5 s), followed by dot rotation (20 s) and then a return to stationary dots (10 s). Participants were instructed maintain fixation on the central green dot for the duration of the trial and to avoid following the moving dots with their eyes. During the dot rotation, participants reported vection onset and vection offset by means of a button push on a gaming controller. Separate buttons were denoted for perceived CW and perceived CCW vection onset/offset, with participants holding the controller with both hands and making responses with both index and middle fingers. At the end of each trial, participants were asked to verbally report on the strength of their vection experience on a scale of 0 (‘no vection’) to 10 (‘I felt I was really moving’). This response was recorded by the experimenter before the next trial begun. Participants were seated with their head on a chin rest during each trial. The experiment comprised a total of 100 trials: 50 coherent trials and 50 incoherent trials, each with 25 trials in CW and CCW directions. Trials were presented in a randomized order in blocks of 10 trials. Participants could take a self-timed break at the end of each block and between trials if necessary. Each session began with a short practice block (12 trials in a randomized order: 6 in each condition, with 3 in each direction) during which the participants could become familiar with the experimental procedure and self-calibrate their use of the vection strength scale. Following this practice, the EEG was prepared and the main experiment was conducted. No participant reported motion sickness and all participants completed the experiment.

The EEG was recorded using a 64 active electrode system (EASYCAP and BrainProducts, GmbH, Germany). The electrodes were placed according to the international 10–10 system with the reference electrode placed at FCz. Horizontal and vertical eye movements were recorded using bipolar electrode montages. The data were collected at a sampling rate of 1000 Hz, with no additional online filters. The active electrode impedances were kept below 5 kΩ throughout the recording. The EEG recording was synchronized with the visual motion stimulation and response controller using triggers sent via parallel port to the EEG recording. The EEG data were acquired using BrainVision Recorder software (BrainProducts, GmbH, Germany) and subsequently analysed using EEGLAB [35], custom scripts in MATLAB and FOOOF [36] in Python. The data preprocessing was performed in the following order. The data were re-referenced to the common average reference. Independent component analysis (ICA) was then used to decompose the data using the runICA function in EEGLAB. Components reflecting ocular noise, such as blinks or muscle artifacts were removed and the EEG signals were then reconstructed.

In order to compare the neural activity in the coherent and incoherent conditions, the data from motion onset to motion offset (i.e. 20 s of stimulus rotation data) was extracted for each participant. The CW and CCW trials were collapsed, resulting in 50 coherent and 50 incoherent trials per participant. Next, the 20 s of data in each trial was segmented in non-overlapping 1 s segments. In order to reduce the potential contribution of motor response artifacts or residual eye artifacts, any segment containing a button–push or channel signal exceeding 100 μV was rejected. The remaining segments were then multiplied with a Hanning window and spectra were computed using an FFT approach. A grand average was calculated for each participant, reflecting their respective spectral amplitude at each electrode position. Each participant’s individual peak in the theta (4–7 Hz), alpha (7–14 Hz) and beta (20–30 Hz) bands were then identified using the FOOOF algorithm [36], with the peak being defined as the largest fitted peak, within the respective band, above the 1/f component of the spectrum. This resulted in an individual peak, i.e. in each frequency band, at each electrode in both the coherent and incoherent conditions for each participant.

In order to study the temporal dynamics of the alpha band activity at vection onset and during ongoing vection, a time–frequency (TF) decomposition was conducted on the coherent condition data. To be included in this analysis a trial had to (a) have at least 3 s between motion onset and vection onset and (b) have at least 5 s of continual ongoing vection after vection onset. For all trials meeting this criteria an 8 s window of data was extracted, ranging from 3 s before vection onset to 5 s after vection onset (i.e.  − 3 s to + 5 s, with time 0 = vection onset). A Morlet wavelet transformation was used to calculate TF maps, linearly ranging from 3 to 40 Hz, for each participant across all segments. The wavelet had a width of 7 cycles. The TF maps were normalized for each participant with respect to a pre-vection interval (− 2.5 s to − 1.5 s relative to vection onset) using a decibel conversion. Next, regions of interest (ROIs) were defined separately for left and right handers based on the results of the FFT analysis. The left hander ROI consisted of electrodes CP1, P3, CP3, P1, P5, PO7 and PO3, while the right hander ROI consisted of electrodes Pz, CP2, P1, P2 and CPz. The spectral power in the alpha band was then averaged within the respective ROIs. The baseline window was defined as − 1.5 s to − 0.5 s (Fig. 1b) and the time points in this window were averaged, resulting in one value per participant. The vection window was defined as − 0.5 s to + 4 s (Fig. 1b), with each participant having one value per time point within this range.

The exact low-resolution brain electromagnetic tomography algorithm (eLORETA), as developed and implemented by Pascual-Marqui [37, 38] and freely available through the LORETA webpage (https://​www.​uzh.​ch/​keyinst/​loreta.​htm), was used to estimate the location of EEG sources. The employed LORETA implementation uses a realistic head model [39] and restricts estimated solutions to cortical grey matter which is modelled by 6239 voxels of 5 mm resolution. The LORETA analysis included coherent condition trials in which vection was present and incoherent trials in which vection was absent. For coherent trials, each participant’s mean time course within the vection window (i.e. − 0.5 to + 4 s relative to vection onset) was extracted and exported to LORETA. In the incoherent trials there was no vection onset and thus no clear window of data to extract. In this case, each participant’s mean time course in a comparable window (i.e. 0.5 to + 4 s) was extracted and exported to LORETA, using their respective mean vection onset latency from the coherent condition as a marker (i.e. time = 0). A transformation matrix based on the coordinates of the electrode positions was created and applied to the coherent and incoherent data. To test for significant effects, paired samples t tests were conducted between the LORETA transformed coherent and incoherent conditions at each time point, for both left and right handers.

In the present study, left and right handers demonstrated remarkably similar vection characteristics with almost identical scores on measures of vection presence, onset latency, duration and subjective strength, for both coherent and incoherent conditions. To some extent this finding contradicts the recent observation that right handers display quicker perceptual transitions from world- to self-motion (i.e. vection onset latency), compared to left handers [42]. One possible explanation for the discrepancy between these findings is differences in the employed visual motion stimuli and the vection they induced. In the previous study, the optokinetic stimuli consisted of a drum marked with vertical black and white lines, which rotated around the participant in the earth-vertical axis, inducing circular vection. In contrast, the visual motion stimuli used in the present study consisted of white dots on a black background, rotating along the roll axis, inducing the perception of roll vection. Notably, the onset latencies in the present study were considerably shorter than those reported in the previous study [42], which could be indicative of a more intense vection experience resulting in less subjective variation on behavioural measures across participants. Although literature on the topic is sparse, there is some evidence showing differences in onset latency [1] and subjective intensity [4] across different types of vection, i.e. along different planes/axes. Indeed, different types of motion stimuli have also been found to show different patterns of neural activation both in the absence [43] and presence [4] of vection. Such findings make it difficult to compare and interpret the relevance of differences in onset latency findings between this and the previous study. A small negative correlation, across both left and right handers, revealed that vection strength decreased very slightly over the course of the experiment. Crucially, this small decrease in perceived vection strength was comparable across the coherent and incoherent conditions, suggesting that both conditions were comparably affected by slight vection habituation over time. In summary, the behavioural data indicate that left and right handers experience roll vection in a similar manner.

In our study, both left and right handers exhibited decreases in alpha power whilst viewing a vection–compatible visual motion stimulus, in contrast to a control stimulus. Determining whether changes in activation patterns reflect visual motion stimulation in general, differences between experimental and control conditions and/or vection itself is experimentally challenging. For example, studies which contrast vection-inducing visual motion stimuli with static control stimuli, e.g. [4, 44], likely include activation changes that reflect the difference between stationary and moving stimuli, irrespective of whether or not vection was present. Also, contrasts between coherent and incoherent/random motion stimuli may result in activation changes which are attributable to physical stimulus differences rather than vection per se [2, 45]. As such, it is apt to examine the extent to which the decrease in alpha power, observed in the present study, can be attributed to vection.

Firstly, the behavioural results clearly demonstrate that on the group level, vection was almost always present in the coherent visual motion condition (i.e. on 96% and 94% of trials for left and right handers, respectively) but rarely, if ever, present in the incoherent visual motion condition (i.e. on 4% and 8% of trials for left and right handers, respectively). This demonstrates that the observed decrease in alpha power is associated with the presence of vection. Secondly, the incoherent visual control stimulus was designed such its physical stimulus properties matched those of the coherent stimulus as closely as possible. Although the incoherent stimulus contained additional local sinusoidal motion, both stimuli presented the same average global velocity. Physical stimulus differences would have likely produced differences in occipital activations between the two conditions, reflecting early visual processing. The absence of such occipital activations in our results suggests that our findings are not attributable to differences between the employed stimuli. Furthermore, this also indicates that the observed differences contain limited, if any, activity related to the processing of visual motion stimulation rather than vection per se. Lastly, subsequent time–frequency analysis of vection-present coherent trials, showed changes in alpha power which were time locked to vection onset, for both left and right handers. This provides further evidence that the changes in alpha power observed in this study are related to vection, rather than general visual motion stimulation.

A key finding in this study is that the decrease in alpha power that results from exposure to a vection compatible stimulus is observed at different topographical regions for left and right handers. Left handers exhibited a lateralized response to the vection stimulus over a left centro-parietal region. In contrast, right handers showed a bilateral response in a midline centro-parietal area (Fig. 3). Multiple comparisons can be drawn between these findings and the existing literature. First, although non-significant, the source localization results suggest that the vection-compatible stimulus, which results in different activity patterns for left and right handers, and the incoherent stimulus are maximally different at regions within the vestibular network. Although further research is required to substantiate this finding, it is congruent with the established idea that visually-induced vection relies on visual–vestibular interaction [2]. Secondly, in contrast to previous fMRI and PET studies [2‐5], which demonstrate vection-related activation of visual cortex and concurrent deactivation of PIVC, the present study found that vection is associated with activity changes at centro-parietal regions. This discrepancy could be explained by a number of factors including our use of a novel control stimulus which controlled for low-level visual properties and average global motion, and also to differences in the type of activity and spatial resolution measured by EEG, fMRI and PET. Lastly, the observation that left and right handers show vection-related alpha band changes at different topographical regions is congruent with fMRI and PET studies demonstrating handedness-dependent vestibular thalamo-cortical dominance [17‐28]. Notably, while such studies show vestibular dominance in the right hemisphere in right handers and in the left hemisphere for left handers, the present study found a bilateral activation for right handers and a left-lateralized activation for left handers.

Given the topographical differences in vection-related activity for left and right handers, subsequent time–frequency analyses were conducted separately for both groups in order to maximize our ability to examine the neural correlates of vection within each group. Despite the topographical differences in where vection-related activity was observed, both left and right handers exhibited a similar pattern of alpha band changes over the course of vection. Both groups showed decreased alpha power around vection onset (significant only for left handers) and increased alpha power during ongoing vection (Fig. 4). These findings not only compliment those of a recent EEG study examining the neural correlates of vection in right-handed participants [31], but also extend the findings to left handers. Notably, left handers appear to show increased alpha power during vection much earlier than right handers in our study. However, as both groups demonstrate the same trend of increasing alpha power during ongoing vection and that the measures of alpha power come from different topographical regions, we are hesitant to label this temporal discrepancy as an effect of handedness. The observation that left and right handers show similar alpha patterns during the course of vection matches well with our observation of comparable behavioural characteristics between the two groups and suggests that vection and its respective processes are consistent across handedness, despite those same processes being observed at different topographical regions in a handedness-dependent manner.

This study made two key observations about alpha oscillations in relation to vection: (1) a decrease in alpha power is observed during exposure to a vection-compatible stimulus, relative to a matched control and (2) both decreases and increases in alpha power are observed during the course of vection. Beginning with the former, the decrease in alpha power during exposure to a vection compatible stimulus is consistent with desynchronization and a release from inhibition [46], correlating with excited neural structures or activated cortical areas [47]. This would suggest that exposure to a vection compatible stimulus induces increased activity in centro-parietal regions, with the effect being (midline) bilateral for right handers and left-lateralized for left handers. Evidence exploring the relationship between alpha oscillations and activations observed in fMRI and PET imaging suggests that alpha power is negatively correlated with the blood-oxygen-level-dependent (BOLD) signal [48] and also with regional cerebral blood flow (rCBF) in primary and association visual cortex [49]. Although the origins of the alpha oscillations observed in this study are unclear, the inverse relationship between alpha power and cortical activation suggests that the observed decrease in alpha power in the present study maps well with findings of increased activation of parieto-occipital regions in previous vection studies [2, 4]. Such previous studies employed only right handed participants and therefore, it is unclear whether spatial differences between left and right handers similar to those observed in this study, would also present themselves in parieto-occipital activations in PET and fMRI studies.

Although our findings indicate a clear link between alpha oscillations and vection, the exact nature of this relationship requires further investigation. The first possibility is that the decrease in alpha power reflects an increase in cortical activation which is compatible with an inhibitory visual–vestibular interaction. This increased activation could operate by amplifying visual signals over vestibular signals such that the visual dominates or by triggering inhibition of vestibular cortex itself. A second possibility is that decreased alpha power reflects an inhibitory visual–visual interaction, with the increased activation either amplifying visual signals consistent with self-motion over those consistent with object–motion or even triggering inhibition of visual object–motion processing.

A third possibility is that the decreased alpha and increased activation reflect additional attentional processes involved in vection perception. It is possible that the sensation of self-motion and/or the sensory mismatch involved in vection perception might require or induce additional attentional processes which involve increased cortical activation. For example, alpha desynchronization has also been observed during turning movements in virtual reality environments, with stronger decreases occurring when there is incongruency between sensory modalities and thus, increased demands on visuospatial attention [50]. Furthermore, the observed decrease in alpha power in the present study could also reflect top-down attentional control. Vection requires that visual motion is interpreted as resulting from self-motion rather than environmental motion. This erroneous interpretation can be explained by our a priori anticipation that the external world is stable, making it more probable that motion cues are attributable to self-motion [1]. In terms of visual processing, anticipatory alpha band modulations have been implicated in the top-down allocation of selective visuospatial attention [51]. Further, the amplitude of alpha desynchronization has been shown to follow the time course of temporal expectation, suggesting that alpha oscillations have a role in the regulation of cortical excitability as a function of anticipatory visuospatial attention and may act as a mechanism for biasing perception [52]. Indeed, in addition to the decreases in alpha power during exposure to a vection compatible stimulus, the time–frequency analyses also revealed a decrease in alpha power beginning prior to vection onset. In this context, this finding could be interpreted to reflect top-down attentional control and anticipation of vection. This decrease in alpha power around the time of vection onset is also consistent with the literature on bistable perception, in which decreasing alpha power is thought to reflect destabilization of the current percept, with decreases beyond a given threshold resulting in the development of the alternative percept [32, 34].

During ongoing vection, we observed an increase in alpha power, which has been shown to be associated with increased inhibition and reduced activation [46]. On one hand, this increase in alpha power during ongoing vection could be a result of object–motion compatible signals and/or vestibular signals (i.e. indicating a lack of self-motion) exerting a reciprocal inhibition on visual self-motion signals. The temporal characteristics of vection vary widely between participants and in order to ensure sufficient trials for analysis and that there was no contamination from changes in alpha power due to vection or stimulus offset, we selected a very limited time window for our analysis of ongoing vection. Unfortunately, this means that we do not know if alpha power continued to increase beyond our window of investigation and if that decrease would ultimately lead to the breakdown of vection. An alternative explanation for the increased alpha power is that while an initial decrease in alpha and an associated increase in activity, is necessary for one perceptual interpretation (i.e. self-motion) to become dominant, the dominant percept can be sustained without continuing increased activation, thus allowing a shift to a less active baseline state with greater alpha power. We propose that future studies combining EEG and non-invasive brain stimulation techniques (e.g. TMS) or positron emission tomography (PET) are necessary to disentangle the role of alpha power increases and decreases and to establish a causal link between alpha oscillations and vection.