Exploring Embodied Cognition and Brain Dynamics Under Multi-Tasks Target Detection in Immerse Projector-Based Augmented Reality (IPAR) Scenarios

Embodied cognition explores the intricate interaction between the brain, body, and the surrounding environment. The advancement of mobile devices, such as immersive interactive computing and wireless electroencephalogram (EEG) devices, has presented new challenges and opportunities for studying embodied cognition. To address how mobile technology within immersive hybrid settings affects embodied cognition, we propose a target detection multitask incorporating mixed body movement interference and an environmental distraction light signal. We aim to investigate human embodied cognition in immersive projector-based augmented reality (IPAR) scenarios using wireless EEG technology. We recruited and engaged fifteen participants in four multitasking conditions: standing without distraction (SND), walking without distraction (WND), standing with distraction (SD), and walking with distraction (WD). We pre-processed the EEG data using Independent Component Analysis (ICA) to isolate brain sources and K-means clustering to categorize Independent Components (ICs). Following that, we conducted time-frequency and correlation analyses to identify neural dynamics changes associated with multitasking. Our findings reveal a decline in behavioral performance during multitasking activities. We also observed decreases in alpha and beta power in the frontal and motor cortex during standing target search tasks, decreases in theta power, and increases in alpha power in the occipital lobe during multitasking. We also noted perturbations in theta band power during distraction tasks. Notably, physical movement induced more significant fluctuations in the frontal and motor cortex than distractions from social environment light signals. Particularly in scenarios involving walking and multitasking, there was a noticeable reduction in beta suppression. Our study underscores the importance of brain-body collaboration in multitasking scenarios, where the simultaneous engagement of the body and brain in complex tasks highlights the dynamic nature of cognitive processes within the framework of embodied cognition. Furthermore, integrating immersive augmented reality technology into embodied cognition research enhances our understanding of the interplay between the body, environment, and cognitive functions, with profound implications for advancing human-computer interaction and elucidating cognitive dynamics in multitasking.


Exploring Embodied Cognition and Brain
Dynamics Under Multi-Tasks Target Detection in Immerse Projector-Based Augmented Reality (IPAR) Scenarios Congying He , Graduate Student Member, IEEE, Yu-Yi Chen, Chun-Ren Phang, I-Ping Chen , Shey-Cherng Tzou , Tzyy-Ping Jung , Fellow, IEEE, and Li-Wei Ko , Member, IEEE challenges and opportunities for studying embodied cognition.To address how mobile technology within immersive hybrid settings affects embodied cognition, we propose a target detection multitask incorporating mixed body movement interference and an environmental distraction light signal.We aim to investigate human embodied cognition in immersive projector-based augmented reality (IPAR) scenarios using wireless EEG technology.We recruited and engaged fifteen participants in four multitasking conditions: standing without distraction (SND), walking without distraction (WND), standing with distraction (SD), and walking with distraction (WD).We pre-processed the EEG data using Independent Component Analysis (ICA) to isolate brain sources and K-means clustering to categorize Independent Components (ICs).Following that, we conducted time-frequency and correlation analyses to identify neural dynamics changes associated with multitasking.Our findings reveal a decline in behavioral performance during multitasking activities.We also observed decreases in alpha and beta power in the frontal and motor cortex during standing target search tasks, decreases in theta power, and increases in alpha power in the occipital lobe during multitasking.We also noted perturbations in theta band power during distraction tasks.Notably, physical movement induced more significant fluctuations in the frontal and motor cortex than distractions from social environment light signals.Particularly in scenarios involving walking and multitasking, there was a noticeable reduction in beta suppression.Our study underscores the importance of brain-body collaboration in multitasking scenarios, where the simultaneous engagement of the body and brain in complex tasks highlights the dynamic nature of cognitive processes within the framework of embodied cognition.Furthermore, integrating immersive augmented reality technology into embodied cognition research enhances our understanding of the interplay between the body, environment, and cognitive functions, with profound implications for advancing human-computer interaction and elucidating cognitive dynamics in multitasking.

I. INTRODUCTION
E MBODIED cognition is a theoretical concept of how people think, act and in general make sense of the world.[1], [2].This framework posits that bodily states and modality-specific systems for perception and action constitute the basis of information processing.It's thought that embodiment plays a role in numerous aspects and outcomes of mental phenomena [1], [3].In conventional cognitive model experiments, the body's role is limited to reacting solely to sensory information within symbolic mental representations regulated by logical and computational rules.Nevertheless, embodied cognition posits that the body's physical engagement was not just a vehicle for logical and computational processes; it actively contributed to and co-produced cognitive processes [1].The complexity and adaptability of embodied cognition make it challenging to elucidate through a singular experiment.However, its essence can be observed through experiments examining the relationship between cognition and physical behavior [1], [3].Embodied cognition emerges from interactions between brain, body and social environment [4].
Electroencephalography (EEG) is more widely used in neurocognitive research, brain dynamics, and real-world applications because of its non-invasiveness, portability, compact design, and high temporal resolution [5], [6].Among them, the wireless EEG system is more suitable for use in fields related to daily life research because of its unfettered operation, short preparation time, easy portability, and fewer restrictions on usage scenarios [7].In the past decade, research and applications on wireless EEG have increased year by year [8].The necessity for wireless EEG systems becomes paramount, especially in studies involving monitoring daily neural activity, such as regular social interactions or body movements [7], [9].At the same time, with the rise of such new fields as immersive augmented reality [10], mixed reality, computing, visible interaction [11], [12], and wearable EEG devices [8], we are witnessing an all-time trend toward integrating physical form and cognitive process.
Previous studies have shown that individuals exhibit distinct brain dynamics in response to various factors, such as social distractions or body motion effects [13], [14], [15].Widely accepted good indicators of cognitive distraction in EEG measurements include changes in theta (4-7 Hz), alpha (8-13 Hz), and beta (14-30 Hz) activities [16].Lin et al. [13] created five scenarios varying in stimulus onset asynchrony to explore how deviations and equations might distract or interfere with each other during driving.Their findings indicate that increases in theta power in the frontal area correlate with driver distraction, reflecting the magnitude of distraction in real-life situations.Additionally, they observed suppressions in alpha and beta power in the motor area.Almahasneh et al. [17] used EEG to examine the effects of different cognitive tasks (math and decision-making problems) on drivers' cognitive states.They revealed significant influences of the distractor task on the theta and beta bands in both the right and left frontal lobes.Moreover, extensive research has explored the influence of body motion on EEG.These studies have shown that EEG spectral power in the mu (10-12 Hz) and beta band decreases over motor areas [18] during isolated foot movements [19] and walking on the treadmill [20] compared with non-movement conditions.Seeber et al. [14] studied the amplitude differences between walking and upright standing conditions.They found that active walking suppressed upper mu and beta oscillations compared to upright standing.
However, how social environmental distraction and physical movement distraction interact to influence embodied cognition deserves further exploration.Therefore, given the embodied cognition theory and drawing on the development of immersive interactive devices in new mobile fields, it is a new challenge to explore how the brain-body-social environment interacts through mobile devices in an immersive hybrid environment.This study investigates embodied cognition interaction affected by body motion interference and social environment distraction under multitasking conditions in immersive mixed scenarios via wearable EEG.In this current study, (1) We designed a novel multi-motion-distraction task experiment, combined with the treadmill and a new light signal distraction device, formed four multi-task conditions, such as: stand no distraction condition (SND), walk no distraction condition (WND), stand distraction condition (SD), and walk distraction condition (WD), to explore embodied cognition dynamics and behavior of performing target detection trials in immersive projector-based augmented reality (IPAR) scenarios; (2) To compare the behavioral performance of the four multi-task conditions; (3) To investigate the brain neural activities and the cortical location based on time-frequency analysis, advanced specific band power, and correlation analysis related to the four multi-task conditions.

II. METHOD A. Participants
This study enrolled fifteen participants (26.4±2.9 years).None of the individuals had a history of neurological or psychiatric conditions, and they were all right-handed and had normal or corrected vision.The experiment was performed under the regulations and national laws of the Institutional Review Board (IRB) of National Yang Ming Chiao Tung University, Hsinchu, Taiwan, and was approved by the Research Ethics Committee of National Yang Ming Chiao Tung University, Hsinchu, Taiwan, under the protocol code NCTU-REC-108-085E.Before participating in the experiment, all participants received a detailed description and signed written informed consent.

B. Embodied Cognitive Perspective Experiment
The embodied cognition perspective experiment emerges from interactions between the brain, body, and the physicalsocial environment, as shown in Figure 1.It's composed of a self-made randomly changed red-green light signal for social environment distraction, a treadmill (BLADEZ, Ares S30, USA) for body walking, the projector-based (Optoma, EC300HT, Taiwan) augmented reality (IPAR) scenario for showing immersive target detection tasks, and a wearable EEG system for collecting neurophysiological data.The following are the details: the randomly changed red-green light signal, which includes two lights, two ultrasonic sensors with the control circuit, and the ultrasonic sensors were evenly distributed on the left and right sides of the treadmill.Subjects needed to always stand/walk on the side of the green light signal whenever it lit up or changed color.If the ultrasonic sensor on the side of the red signal detected a subject, it emitted a sound to influence their behavior.Furthermore, we utilized a treadmill to induce body motion, and subjects performed IPAR target detection tasks on the treadmill, whether standing or walking.Subsequently, we combined body motion and the distraction of sudden light signal changes to create four different conditions: stand no distraction condition (SND), walk no distraction condition (WND), stand distraction condition (SD), and walk distraction condition (WD).This allowed us to explore changes in brain dynamics and behavioral responses in individuals immersed in performing tasks across various mixed-task scenarios.

C. Experimental Conditions and Data Acquisition
Figure 2 shows the detailed experimental paradigm for target detection under four multi-tasks in IPAR scenarios.The total duration of the experiment was about 24 minutes, divided into 1 minute of baseline (resting EEG), four different sessions, and four rests.We disrupted the order of four multi-task conditions (SND, WND, SD, and WD) to counterbalance the session order.Each subject completed four multi-task sessions.Figure 2 (b) shows the target detection task presented in the IPAR scene, we used Unity (2018.3.6 version) to design this target detection task [21].The left side is the target detection task without any interference.The right side is when performing the search task.At the same time, the subjects need to pay attention to the distraction light signal on the treadmill.Each session contains 108 trials, a single trial (2.6s) includes the fixation 1(1400ms), cue (150ms), fixation 2 (450ms), the target appears, searching, and response phase (650ms).During the searching stage, nine randomly selected icons were prompted, with one being the target and the remaining eight being distractors.Once searched, the subject pressed the button corresponding to the side of the appeared target.Then, the Unity program will record the reaction times (RT) from the target appearance until the subject presses the button, and the number of correct and incorrect executions.Figure 2 (c) shows an example of one subject wearing the St. EEG TM Vega cap performing the experiment for target detection under four multi-tasks in the IPAR scenario on the treadmill.The study used a 32-channel wireless EEG recording system (St.EEG TM Vega, Taiwan).The sampling rate was 500 Hz.The reference electrodes were A1 and A2, and the ground electrode was FPz.The locations of all 32 recording channels matched the International 10-20 systems.The subjects either stood or walked on the treadmill, and the speed of the treadmill was set at 3.5km/h during walking.

D. Four Multi-Tasks Behavioral Data Statistical Analysis
The behavioral data were collected by Unity (2018.3.6 version) and processed by MATLAB (R2021a).For each participant, we analyzed the response time corresponding to the time of the target appearance.Then, we calculated the average response time for SND, WND, SD, and WD conditions (as in Figure 2c).Furthermore, we computed target detection accuracy by dividing the number of correct key presses by the total number of trials.Finally, we calculated the averages and standard deviations for 15 subjects performing multiple tasks.Outliers were excluded if the response time dispersed outside the mean response time plus three times the standard deviation of each session.We then tested the significance of response accuracy and response time differences between two of these multi-four conditions on 15 subjects using a t-test.That is, significant differences were tested pair-wise for SND and WND conditions, SD and WD conditions, SND and SD conditions, and WND and WD conditions (p < 0.05).

E. Four Multi-Task Conditions EEG Data Analysis
We analyzed the collected EEG data to decode the changes in brain dynamics under four multi-task conditions, as shown in Figure 3.
1) EEG Signal Preprocessing: We conducted all EEG data analysis using the EEGLAB toolbox (version 2021.1) in MATLAB [22].The main steps in preprocessing EEG signals included band-pass filtering, initial Artifact Subspace Reconstruction (ASR), Independent Component Analysis (ICA), EEG source localization, and epoch extraction.First, we applied a 1-50 Hz band-pass finite impulse response (FIR) filter to remove baseline drift and high-frequency environmental noise.The EEG signals are often noisy and contaminated by motion and muscle artifacts because the experiment involves walking.Therefore, the initial steps included the preliminary removal of bad channels, that is a visual inspection of channels by scrolling through the whole dataset to identify and remove bad or atypical channels (e.g., amplitudes exceeding ±100µV) [23].Then, we performed artifact subspace reconstruction (ASR) bad burst correction, setting the "BurstCriterion" parameter at 15 [24], [25], to automatically extract high-amplitude muscle artifacts and reconstruct the period EEG data (WindowCriterion=0.25,BurstRejection=off) [24].Furthermore, ICA [26] (Infomax ICA, implemented using runica) was employed to separate the scalp EEG data into independent components (ICs).Subsequently, we used the ICLabel toolbox [27] to identify the ICs as brain activity, eye movement artifacts, muscle activity, heart activity, or other noise sources.We removed the EEG contaminations by back-projecting only the component accounting for the brain signals, and further validated manually, retaining at least 10 ± 3 components for each subject.After ICA, EEG dipole source localization analysis was performed using Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.the DIPFIT2 routine functionality of the EEGLAB toolbox.A spherical head model co-registered with the Montreal Neurological Institute was used in the DIPFIT software to provide approximate Talairach coordinates for every analogous dipole source [28].Finally, EEG data epochs were extracted based on four multi-task conditions and phases.Each epoch included a baseline, cue to target appearance, and target search phrase.
2) Group Analysis of Clustering ICA Components: Clustering was performed on components from multiple subjects based on spatial distribution and EEG characteristics to investigate the cross-subject stability of ICA decomposition components [13], [29].Using k-means statistical analysis criteria (k = 6) in the EEGLAB clustering toolbox, dipole sources from 15 subjects were semi-automatically selected and clustered to identify the interested Independent Components (ICs) based on scalp maps, dipole source locations, and within-subject consistency.We then selected ICs based on the activities elicited by environmental distractions, body movements, and visual processing.We analyzed data cross-subjects based on their EEG characteristics.
3) Time-Frequency Decomposition and Band Power Analyses: We conducted the event-related spectral perturbation (ERSP) analysis [30] on the independent component activities within the extracted clusters from the group analysis discussed in the previous section.To assess the brain dynamics over time following specific task events under SND, SD, WND, and WD conditions, we used the "newtimef" function of the EEGLab Toolbox [22], [26].In this analysis, the cue acted as the starting point for each trial.We defined a baseline at -300 ms, with a window size set to 128 points and an overlap of 50%.Subtracting the mean of the baseline spectrum from the power spectrum enabled us to visualize spectral "perturbations."Each trial underwent this process, and the results were averaged to produce ERSP images showing statistically significant changes in spectral characteristics (p < 0.05).
Furthermore, based on the average ERSP results, we extracted theta (4-7 Hz) and beta (13-28 Hz) power under different conditions.To assess the correlation (r ) between SND and WND conditions, SD and WD conditions, SND and SD conditions, and WND and WD conditions for both theta and beta power, we computed Pearson's correlation coefficient.Initially, we segmented the frequency band values of each trial from -300 ms to 1250 ms to calculate multiple r values.This calculation used a moving window of 64 ms with an overlap rate of 50%, allowing us to observe the trend in band correlation.Subsequently, we determined the r value for the entire duration from the target presentation to the end of the trial event (600 ms to 1250 ms).This measurement helped quantify the difference in spectral changes between the two conditions during the target search task.A higher r value indicates a stronger correlation and a smaller disparity in the dynamic changes of the frequency bands between the two conditions.Significant differences (p < 0.05) between two of the four multitasking conditions were identified using the Wilcoxon signed-rank test, denoted by an asterisk.

A. Behavioral Performance in Four Multi-Tasks
Figure 4 plots the reaction time (RT) and accuracy data for four multi-task conditions across 15 subjects to assess the impact of walking and distractions on human performance.Figure 4(a) illustrates that, under all no-distraction tasks, the response accuracy for WND tasks is 0.806±0.96,significantly lower ( p = 0.0019) than that for SND tasks (0.905±0.05).In distraction tasks, the response accuracy for WD tasks (0.757±0.73) is significantly ( p = 0.0415) lower than that for SD tasks (0.815±0.08).A significant difference exists between standing and walking tasks ( p = 0.0019, p = 0.0415), irrespective of the presence of a distraction task.
We then investigated the influence of distractions on human performance.In both standing tasks, the accuracy of SD tasks is significantly ( p = 0.014) low than that of SND tasks.However, in both walking tasks, the target detection accuracy under the WND task is not significantly ( p = 0.13) different from that under the WD task.A walk and distraction (WD) multi-task results in the lowest accuracy.As depicted in Figure 4

B. Dipole Locations and Scalp Maps of ICs
We used ICA to separate statistically independent brain sources.Based on previous studies [13], [16] [19], we selected the frontal and motor cortices for further analysis.Considering the visual searching tasks conducted in the IPAR scenarios, we also included the occipital IC cluster in the subsequent analysis.IC clustering grouped all 153 brain-related components from 15 subjects into six significant clusters.These clusters comprised the left/right frontal, middle motor cortex, and left/middle/right occipital areas.To analyze the effects of multi-motion distraction, we merged six distinct components into three clusters, including 35 components.For instance, we merged the left and right occipital clusters into the occipital lobe cluster.The EEG sources within the same cluster across different subjects originated from the same physiological component.Figure 5 illustrates the equivalent dipole source location and scalp maps for the left frontal lobe (N = 10, 10 ICs), middle premotor cortex (N = 11, 11 ICs), and occipital lobe (N = 14, 14 ICs).Furthermore, Table I presents the coordinates of the dipole source locations for each cluster and the Montreal Neurological Institute (MNI) coordinates.

C. Brain Neural Dynamics in the Four Multi-Tasks
1) Left Frontal Clusters: Figure 6 shows the averaged ERSP and spectral trends in the left frontal cluster under the four multi-task conditions.Figure 6 (a) shows a notable decrease in alpha and beta power between 600 ms and 1250 ms in Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.the frontal cluster.This decrease is associated with enhanced performance in the target search task under SND conditions, indicating heightened attention.Moreover, it reveals an increase in theta band activity from 600 ms to 900 ms in WND compared to SND, accompanied by less suppression in beta band power from 600 ms to 1250 ms.SND demonstrates greater theta augmentation and beta suppression compared to SD. Figure 6 (b) (c) further illustrates the trends in theta and beta power.Notably, the correlations between SND and WND (theta: r = 0.689; beta: r = −0.24)are weaker than those between SND and SD (theta: r = 0.9; beta: r = 0.93).This implies that the changes in brain activity induced between standing and walking conditions were more pronounced than between undistracted and distracted conditions.Similarly, in Figure 6 (a), we observe an increase in beta power in the frontal cluster during the WD condition, which involves multitasking with walking and distraction, compared to the SD condition.A slight decrease in theta band power is noted from 600 ms to 900 ms, while theta band power increases from 0 to 600 ms in WD than WND. Figure 6 (b) and (c) depict the temporal changes in beta band power for WD and WND.The correlation of beta band power between WND and WD (r = 0.63) surpasses that between SD and WD (r = 0.13).This higher correlation between WND and WD implies that more pronounced EEG differences exist between the walking and standing conditions than between the distracted and nodistraction conditions.
2) Premotor Cortex Cluster: Figure 7 shows the averaged ERSP and spectral trends in the middle premotor cortex under the four multi-task conditions.Figure 7 (a) highlights the pronounced suppression of alpha and beta power within the 600 ms to 1250 ms timeframe in the SND condition.Notably, following button press in all four multitasking conditions, the premotor cortex exhibited significant increases in theta and alpha power.Furthermore, in both WND and WD conditions, compared to SND and SD conditions, there is a noticeable decrease in alpha and beta band power suppression within the 600 ms to 1250 ms window while subjects are simultaneously walking and performing target detection tasks.Figure 7 (b) (c) reveals consistent trends in theta and beta power over time in the case of SND and SD post-target appearance.The correlation trend between SND and SD (theta: r = 0.98; beta: r = 0.27) exceeds that of SND and WND (theta: r = 0.89; beta: r = −0.13)after the target appearance, indicating that a standing compared to a walking condition elicits more pronounced brain dynamic changes than a no-distraction compared to a distraction condition.Moreover, in Figure 7 (a), an increase in theta and beta band power in the premotor cortex is evident in the WD condition, which involves multitasking with both walking and distraction, compared with the SD condition.A slight increase in theta band power is observed within the 300 ms to 1000 ms window in WD compared to WND. Figure 7 (b) (c) also shows the trends of theta and beta power changes in WD and WND.The theta correlation trend between WND and WD (r = 0.96) mirrors that of SD and WD (r = 0.95).However, the beta trend correlation between WND and WD (r = 0.63) is lower than that of SD and WD (r = 0.77).Notably, beta band power exhibits the least suppression in WD compared to the SND condition.
3) Occipital Lobe Cluster: Figure 8 shows the averaged ERSP and spectral trends in the occipital cortex under the four multi-task conditions.We observed a decrease of theta and low beta power from 600 to 1250 ms and an increase in the alpha power within 300 to 800 ms while performing a target search task in the SND condition, as is shown in Figure 8 (a).Furthermore, compared to SND, we noted increases in theta and beta band power across the WND, SND, and WD conditions, indicating the influence of visual, walking, and distracting effects in the occipital lobe.

IV. DISCUSSIONS A. Decrease in Behavioral Performance Under Multi-Tasking
The impact of multitasking on daily performance has been a subject of prior research [31].For instance, Strayer and Drews [32] examined the use of mobile phones while driving and found that participants' driving performance was significantly impaired due to divided attention.Several studies have also investigated the performance of students habitually multitasking in the classroom, observing significant performance losses in the primary learning tasks in various experimental settings, such as texting during lectures [33], [34].Recently, Chuang, et al.'s study [35] pointed out that when pedestrians focus on smartphones, they will be distracted and unable to maintain situational awareness; multitasking will lead to decreased behavioral performance and changes in walking patterns.In our study, the accuracy of responses under WND (0.806±0.96) and SD tasks (0.815±0.08) significantly declined compared to the single-task SND condition (0.905±0.05).The WD condition, which involved walking and distraction tasks, showed the biggest decrease in accuracy (0.757±0.73).Simultaneously, there was no significant difference in reaction time (RT) between any two pairs of SND, WND, SD, and WD conditions.This result may be attributed to the limited response time in the experimental setup, with a brief interval of only 650ms between the appearance of the target and the next trial.A failure to respond within the stipulated time frame was considered an error.Some studies suggest that restricting response time can lower the accuracy of the SSVEP target classification [36].Therefore, despite the similar RTs across different conditions, the accuracy of behavioral performance significantly decreased when engaging in walking and distracting tasks.

B. Pronounced Brain Dynamic on Target Detection of Walking Task in Embodied Cognition
This study found that the comparison between standing and walking from target appearance to subject response induced more substantial changes in brain dynamics than the comparison between a no-distracted and distracted state.Specifically, in the frontal lobe, the correlations of theta (r=0.689) and beta (r=−0.24)power between the SND and WND conditions was lower than the correlations of theta (r=0.9) and beta (r=0.93) between the SND and SD conditions.Similarly, in the premotor cortex, the correlations between the theta (r=0.89) and beta (r=−0.13)frequency bands in the SND and WND conditions were lower than the correlations of theta (r=0.98) and beta (r=0.27) between the SND and SD conditions.Lower correlations in event-related power suggest greater brain turbulence.Notably, the dynamic changes in the brain triggered by a single signal distraction light are relatively small compared to the impact of walking conditions.This implies that interference from body motion induces more significant oscillations in the frontal and motor cortex than distraction from the social environment.Beurskens et al.'s study [37] also delved into the effects of cognitive and motor interferences on walking performance, revealing a significant impact of motor interference tasks on walking performance.Furthermore, under both cognitive and motor interference walking conditions, there was a noticeable modulation in the average activity of alpha and beta frequencies in the frontal and central brain regions.Concurrently, our study provides further support for the notion that in an embodied cognitive environment, body motion manifests in behavior and reflects dynamic changes in the brain.
Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

C. Diminished Beta Suppression While Multitasking During Walking
Several studies have shown that before movement, the motor cortex and frontal lobe will exhibit mu and beta band suppression [38], [39].The frontal lobe and motor cortex showed beta suppression under the SND condition.This may be attributed to the continuous muscle contractions required for maintaining a standing posture, positioning the body in a state of readiness for potential movement, thereby suppressing beta frequency activity [38].It is worth noting that, in our study, under the conditions of WND and WD, there was a beta power rebounded in the frontal lobe and premotor cortex.Beta-power rebound is a critical indicator of somatosensory and motor integration, typically observed after completing a motor task [40].During walking, there is a phase involving partial completion of movement, and the suppressed beta power rebound occurs after a movement is completed.On the other hand, this could be because of performing multiple tasks simultaneously, such as target detection while walking or mixing walking and distracting tasks, which increases cognitive load or brain resources competition.Beta activity generally increases when the cognitive load increases [41].One study pointed out [37], beta activity increased when performing motor-demanding tasks (i.e., grasping tasks) while walking.That is, the demands of the walking and distraction tasks used in our study affected participants' performance in executing target detection in an upright position.Additionally, there was a need for increased attentional resources to compensate for this interference, as indicated by the heightened beta activity and the attenuation of beta suppression.Furthermore, on-task behavior emerges from interactions among the brain networks.There are competing brain sources when we process multiple tasks simultaneously [42].The limited multitasking capacity of the brain often cannot meet the demands of simultaneous execution of controls [43].The attenuation of event-related walking spectral and the disappearance of beta suppression indicate the competition for neural resources during multitasking [44].

V. CONCLUSION
This study investigated the embodied cognition from the interactions between the brain, body, and social environment of fifteen human participants who undertook a target-detection mission under multitasking with body motion interference and social environment distraction in an immersive mixed scenario with wearable EEG.The embodied cognition experiment comprises a self-made random social environment distraction red-green light signal, a treadmill, an IPAR for showing immersive target detection scenario, and a wearable EEG system.Participants' behavioral performance and EEG dynamics of the IC cluster near the left frontal, mid-premotor cortex, and occipital lobes were analyzed while performing four (SND, WND, SD, WD) multi-tasks.This study demonstrates a decline in behavioral performance when executing multiple tasks simultaneously.In line with previous research, the frontal lobe and motor cortex showed decreased alpha and beta power during a standing target search task.This occipital lobe decrement in theta and increment in alpha power are associated with visual processing.The perturbations in theta band power during tasks involving distractions were inconsistent, likely because of the infrequent and random occurrence of interruptions.Compared to distractions from the social environment signals, physical movement elicited more pronounced fluctuations in the motor cortex, especially in scenarios involving walking and multitasking, beta suppression was diminished, potentially implicating an increased cognitive load or competition for brain resources.This study challenges traditional views of cognition as merely behavior or braincentered, emphasizing the role of brain and body systems in shaping thought and behavior.Additionally, integrating immersive technologies, such as virtual and augmented reality, into this field further expands our understanding of cognitive processes.These technologies allow deeper insights into how our bodies and surroundings interact and influence our cognitive functions.

Fig. 1 .
Fig. 1.The overall structure diagram of embodied cognition perspective experiment in an immersive projector-based augmented reality (IPAR) scene.

Fig. 2 .
Fig. 2. The detailed experimental paradigm for target detection under four multi-tasks in IPAR scenarios.(a) Overall experiment timeline.(b) The experimental flow of a single trial includes the fixation, cue, target searching, and response phase.(c) An example of one subject performing the experimental for target detection under multi-tasks in an IPAR environment on the treadmill.The four different multi-tasks: stand no distraction condition (SND), walk no distraction condition (WND), stand distraction condition (SD), and walk distraction condition (WD).

Fig. 3 .
Fig. 3.A flow chart detailing the analysis processing the EEG data in all subjects.

Fig. 4 .
Fig. 4. Behavioral response for target detection under four multi-task conditions across 15 subjects.(a) The response accuracy for different multi-task sessions.(b) Overall response time (RT) across different multi-task sessions.Bars represent the standard error of the mean.The dark gray bar: SND; dark green bar: SD; light cyan bar: WND; the dark red bar: WD.Significance is indicated by * p < 0.05, with the asterisk positioned at the top of the figure (a).

Fig. 5 .
Fig. 5.The component scalp maps and dipole sources of the three selected clusters in or near the left frontal lobe, middle premotor cortex, and middle occipital lobe of all the participants.Cluster 1: Left frontal lobe (N = 10).Cluster 2: Middle premotor cortex (N = 11).Cluster 3: Middle occipital lobe (N = 14), where N is the number of dipoles projected in a cluster.

Fig. 6 .
Fig. 6.Averaged component-based time-frequency decomposition in the left frontal lobe under the SND, SD, WND and WD conditions in all subjects.(a) The event-related spectral perturbation (ERSP) images of the left frontal cluster with four multi-tasks conditions.The column contains the stand and walk conditions, and the row contains the no distraction and distraction conditions.The first black line denotes the command cue onset.The second red dashed line reveals the target appears.The third black dashed line denotes the response time line.The ERSP difference is significant at p < 0.05.The colored bars indicate the scale of ERSP.(b) The time-frequency band power and correlation (r value) comparison under row multi-tasks, including theta (θ) and beta (β) band power and correlation comparison between SND and WND, and theta and beta band power and correlation comparison between SD and WD.The left y-axis is band power, and the right y-axis is two task correlations.SND (the dark gray line), WND (the light cyan line), SD (the dark green line), and WD (the dark red line).(c) The time-frequency band power and correlation (r value) comparison under column multi-tasks, including theta and beta band power and correlation comparison between SND and SD, and theta and beta band power and correlation comparison between WND and WD.The left y-axis is band power, and the right y-axis is two task correlations.SND (the dark gray line), SD (the dark green line), WND (the light cyan line), and WD (the dark red line).Significance is indicated by * p < 0.05, with the asterisks positioned at the bottom of figures (b) and (c).

Fig. 7 .
Fig. 7. Averaged component-based time-frequency decomposition in the middle premotor cortex under the SND, SD, WND and WD conditions in all subjects.(a) The ERSP images of the middle premotor cortex cluster with four conditions.The column contains the stand and walk conditions, and the row contains the no distraction and distraction conditions.The first black line denotes the command cue onset.The second red dashed line reveals the target appears.The third black dashed line denotes the response time line.The ERSP difference is significant at p < 0.05.The colored bars indicate the scale of ERSP.(b) The time-frequency band power and correlation (r value) comparison under row multi-tasks, including theta (θ) and beta (β) band power and correlation comparison between SND and WND, and theta and beta band power and correlation comparison between SD and WD.The left y-axis is band power, and the right y-axis is two task correlations.SND (the dark gray line), WND (the light cyan line), SD (the dark green line), and WD (the dark red line).(c) The time-frequency band power and correlation comparison under column multi-tasks, including theta and beta band power and correlation comparison between SND and SD, and theta and beta band power and correlation comparison between WND and WD.The left y-axis is band power, and the right y-axis is two task correlations.SND (the dark gray line), SD (the dark green line), WND (the light cyan line), and WD (the dark red line).Significance is indicated by * p < 0.05, with the asterisks positioned at the bottom of figures (b) and (c).

Fig. 8 .
Fig. 8. Averaged component-based time-frequency decomposition in the occipital lobe under the SND, SD, WND and WD conditions in all subjects.(a) The ERSP images of middle occipital lobe cluster with four conditions.The column contains the stand and walk conditions, and the row contains the no distraction and distraction conditions.The first black line denotes the command cue onset.The second red dashed line reveals the target appears.The third black dashed line denotes the response time line.The ERSP difference is significant at p < 0.05.The colored bars indicate the scale of ERSP.(b) The time-frequency band power and correlation (r value) comparison under row multi-tasks, including theta (θ) and beta (β) band power and correlation comparison between SND and WND, and theta and beta band power and correlation comparison between SD and WD.The left y-axis is band power, and the right y-axis is two task correlations.SND (the dark gray line), WND (the light cyan line), SD (the dark green line), and WD (the dark red line).(c) The time-frequency band power and correlation comparison under column multi-tasks, including theta and beta band power and correlation comparison between SND and SD, and theta and beta band power and correlation comparison between WND and WD.The left y-axis is band power, and the right y-axis is two task correlations.SND (the dark gray line), SD (the dark green line), WND (the light cyan line), and WD (the dark red line).Significance is indicated by * p < 0.05, with the asterisks positioned at the bottom of figures (b) and (c).

TABLE I THREE
IC CLUSTERS IN THE BRAIN AND THE MONTREAL NEUROLOGICAL INSTITUTE (MNI) COORDINATES OF THEIR DIPOLE SOURCE LOCATIONS 509 ± 22.54 ms, respectively.There is no noticeable difference between them.