Stress management using fNIRS and binaural beats stimulation

Abstract

In this study, we investigate the effectiveness of binaural beats stimulation (BBs) in enhancing cognitive vigilance and mitigating mental stress level at the workplace. We developed an experimental protocol under four cognitive conditions: high vigilance (HV), vigilance enhancement (VE), mental stress (MS) and stress mitigation (SM). The VE and SM conditions were achieved by listening to 16 Hz of BBs. We assessed the four cognitive conditions using salivary alpha-amylase, behavioral responses, and Functional Near-Infrared Spectroscopy (fNIRS). We quantified the vigilance and stress levels using the reaction time (RT) to stimuli, accuracy of detection, and the functional connectivity metrics of the fNIRS estimated by Phase Locking Values (PLV). We propose using the orthogonal minimum spanning tree (OMST) to determine the true connectivity network patterns of the PLV. Our results show that listening to 16-Hz BBs has significantly reduced the level of alpha amylase by 44%, reduced the RT to stimuli by 20% and increased the accuracy of target detection by 25%, (p < 0.001). The analysis of the connectivity network across the four different cognitive conditions revealed several statistically significant trends. Specifically, a significant increase in connectivity between the right and left dorsolateral prefrontal cortex (DLPFC) areas and left orbitofrontal cortex was found during the vigilance enhancement condition compared to the high vigilance. Likewise, similar patterns were found between the right and left DLPFC, orbitofrontal cortex, right ventrolateral prefrontal cortex (VLPFC) and right frontopolar PFC (prefrontal cortex) area during stress mitigation compared to mental stress. Furthermore, the connectivity network under stress condition alone showed significant connectivity increase between the VLPFC and DLPFC compared to other areas. The laterality index demonstrated left frontal laterality under high vigilance and VE conditions, and right DLPFC and left frontopolar PFC while under mental stress. Overall, our results showed that BBs can be used for vigilance enhancement and stress mitigation.

1. Introduction

Today, mental stress is one of the major contributors to health and economic problems. Many people spend a lot of time at work, which contributes to higher stress levels. Chronic stress has been linked to serious health problems such as heart diseases, obesity, diabetes, stroke and depression [1]. High workplace stress also impairs creativity, problem-solving ability, decision-making process, and working memory function [2]. In a variety of professional situations, such as surveillance, airport security, industrial control, and medical monitoring, this is a severe issue [3]. Stress has been listed as the second most severe work-related health issue in Europe [4], after musculoskeletal disorders. This heightened stress level costs the world economies billions of dollars in lost productivity and health-related problems [5]. It has been suggested that the annual economic cost of mental illness globally is $\mathrm{\$}$ 2.5 trillion, with a projected increase to $\mathrm{\$}$ 6 trillion by 2030 [6]. According to the American Institute of Stress, the United States spends $\mathrm{\$}$ 300 billion per year on stress-related disorders [5]. Similarly, work-related stress cost European businesses $\mathrm{\$}$ 240 billion in a year [7]. All these occurrences harm economic output, which affects everyone in society. Thus, stress management is critical for safety, productivity, and quality of life. Despite the serious consequences of stress on the human body, stress treatment has remained marginal in medical practice due to the lack of approved interventions for stress mitigation. In this research work, we aim at exploring interventions to enhance vigilance and mitigate mental stress. To achieve this, we propose using Functional Near-Infrared Spectroscopy (fNIRS) as a potential biomarker of stress level and binaural beats stimulation (BBs) as a method to enhance vigilance and mitigate stress.

fNIRS is noninvasive optical imaging technique used to measure the cerebral hemodynamics associated with neural activity. It has a temporal resolution in sub-second range, and the spatial resolution is in the order of 1 cm² [8]. Hence, fNIRS is a promising complementary mode of measurement, achieving some middle ground between Electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) techniques in terms of spatial and temporal resolution as well as mobility [9,10]. More recently, fNIRS has been accepted as an assistive tool to differentiate between depression, bipolar disorder and schizophrenia [11,12]. To date, fNIRS has only been used to investigate the neural response of stress in a limited number of studies. For example, Rosenbaum et.al [13] monitored cortical neuronal activity during the Trier Social Stress Test (TSST) and found significant differences in the dorsolateral prefrontal cortex (DLPFC), inferior frontal gyrus, and superior parietal cortex. The study was reproduced by the same research group, which revealed that high ruminators exhibited lower responses in the inferior frontal and DLPFC [14]. Similarly, a recent study looked at the prefrontal cortex's hemodynamic changes during the Maastricht Acute Stress Test (MAST) and found significantly increased neuronal activity in the DLPFC and orbitofrontal cortex (OFC) in response to the MAST compared to the control condition [15]. Similar findings were reported on workers’ prefrontal cortex (PFC) who underwent construction hazard [16]. Meanwhile, in our earlier research, applying mental arithmetic tasks with negative feedback to stress participants, we found that the stress condition lowered activity in the right PFC relative to the control condition [17].

While previous fNIRS stress studies look at distinct brain regions separately, brain connectivity examines the relationship between brain regions by measuring the dependencies of brain activity such as coactivation and causal relationships [18,19]. Functional connectivity network has been reported as a robust biomarker for stress and anxiety [20,21]. In this context, fNIRS studies have shown that the PFC functional connectivity significantly decrease during sustained attention and increase with decision-making [22,23]. Inconvenient workplace circumstances have been shown to reduce workers’ comfort and functional connectivity on PFC at non-ergonomic workstations and irregular working shifts [24,25]. The stability of fNIRS functional connectivity has been recently validated and suggested using the PFC connectivity as a potential biomarker of emotional sensitivity, and major depression [21, 26,27]. However, one fNIRS study reported higher connectivity network while training shutdown maintenance workers [28]. This finding could be due to the lack of inducing stress in the workplace [29]. In particular, studies that validated their protocols using salivary cortisol or salivary alpha amylase (SAA) both widely used measures in the clinical diagnosis of emotional stress events are optimal [30,31]. Nevertheless, the aforementioned fNIRS connectivity studies filtered the connectivity network using absolute or proportional thresholding methods. These methods of network filtering required the user to decide the threshold value [32]. Fully automated thresholding method is hence required to enhance the estimation of connectivity network for the assessment of vigilance and mental stress. To the best of our knowledge, only one fNIRS study has automated the thresholding of the connectivity network [33]. The fNIRS study estimated the connectivity network using the Pearson correlation and used the orthogonal minimum spanning tree (OMST) for thresholding the network. However, the method was only validated on a dataset of Alzheimer patients who are known to have lower connectivity networks. Hence, validating the thresholding connectivity method on a dataset of healthy people is plausible for a better estimate of connectivity network in vigilance enhancement and stress mitigation at the workplace.

To mitigate elevated mental stress, the literature contains a large number of methods including biofeedback (BFB), neurofeedback (NFB) and indirectly through cognitive vigilance enhancement methods [34,35]. BFB provides a way to regulate physiological variations and display the information in real-time [36]. NFB is a type of biofeedback technique that uses real-time recordings of brain activity to enhance the self-regulation of specific brain functions in connection with behavior. The most used NFB techniques are based on EEG, fMRI and fNIRS. The underlying concept is that through brain training with such feedback, one may entrain, modify, and regulate neural activity. However, most of the BFB and NFB studies were hardly translated from cognitive neuroscience lab into real-life practices [37]. Thus, although these methods give real-time feedback, they are not suitable for workplace environments.

BBs techniques have recently attracted much interest in therapeutic and basic research applications. When two pure tones with a slight frequency difference are presented to each ear separately, the brain perceives a third tone, named BBs, with a frequency identical to the frequency difference of the presented tones. EEG and Magnetoencephalography (MEG) studies showed synchronization of brain activity at the BBs frequency of stimulation and its harmonics in and beyond the corresponding sensory brain areas [38,39]. Brain oscillations entrained by the BBs have been recorded with EEG and MEG similar to the responses elicited by amplitude-modulated sounds at theta frequencies around 4 Hz [40,41], gamma frequencies around 40 Hz [38, 41] and beta frequencies within the range of 15 to 16 Hz [42,43]. The BBs in general have been proven to enhance vigilance levels [44], attentional control [45], long-term memory [46], focusing [47], working memory [48], dual task crosstalk [49], sleep quality [50], and relaxation [51]. Furthermore, BBs have proven to be useful in clinical and research work related to cognition, anxiety, and pain [52,53,54,55] . A significant decrease in anxiety levels was observed when compared to the control group. Likewise, McConnell et al. [56] assessed the heart rate variability while individuals were exposed to theta-frequency BBs and found significant variations in sympathetic and parasympathetic activities.

Interestingly, theta BBs in the parietal, frontal, and temporal cortices, including the auditory cortex have been reported to produce frequency following response [41]. The assumption is that the brain can adapt its brainwave frequency to the frequency of the auditory beat by a synchronization process between neural activity and the auditory stimuli. It was found that BBs at alpha band of 10 Hz increased interhemispheric coherence between the auditory cortices [57]. Meanwhile, BBs at beta band of 15 Hz strengthened cortical networks while performing working memory tasks [46].

Entrainment of beta oscillations is of specific interest because of the role of beta oscillations in vigilance enhancement [44], visuospatial [43], working memory [58], anxiety and pain [55]. Previous study found that listening to beta oscillation at 15 Hz BBs during a visuospatial working memory task not only increased the response accuracy, but also modified the strengths of the cortical networks during the task [43]. Likewise, beta oscillation at 16 Hz BBs demonstrated its effectiveness in mitigating the attentional blink. In particular, the 16 Hz BBs reported as a promising method of non-invasive brain stimulation for enhancing training and learning which is well-suited to rehabilitation training [42]. These works motivated us to use the 16 Hz to entrain multiple mental states, i.e., enhance vigilance and mitigate stress. Nevertheless, the method of assessment on previous BBs studies, in general, were limited to the behavioral responses, and recently introduced EEG and MEG modalities. In this regard, we hypothesize that BBs at 16 Hz could enhance vigilance and mitigate mental stress levels at the workplace. To test this hypothesis, we assessed the levels of vigilance and stress by utilizing simultaneous recording of fNIRS with behavioral responses and salivary alpha-amylase. We then quantified the vigilance and stress levels using the reaction time (RT) to stimuli, accuracy of detection, and the functional connectivity metrics of the fNIRS estimated by Phase Locking Values (PLV). Based on the above, we summarize the main contributions of our work below:

• •Develop an experimental protocol to induce four cognitive conditions: high vigilance (HV), vigilance enhancement (VE), mental stress (MS) and stress mitigation (SM). Explore the effectiveness of using 16 Hz BBs to enhance vigilance and mitigate stress levels
• •Quantify vigilance and stress levels based on fNIRS connectivity networks, salivary alpha amylase, and behavioral responses
• •Investigate two thresholding connectivity network methods based on maximum global cost efficiency (GCE) and orthogonal minimal spanning tree (OMST) for the evaluation of cognitive vigilance and mental stress levels
• •Study the Laterality Index (asymmetry) under the four cognitive conditions above

2. Methodology

2.1. Participants

Thirty healthy right-handed adults (25 males, 5 females; aged 25 ± 5) participated in this study. The inclusion criteria required having normal hearing, normal or corrected-to-normal vision, no history of neurological or psychiatric illness and no intake of long-term medications. None of the participants had prior experience with binaural beats stimulation nor the experiment protocol. They were also requested not to smoke, exercise, or take any caffeine or alcohol 12 hours prior the experiment. After introducing the participating subjects to the experiment details, they signed the informed consent. The study protocol was prepared following the principles of ethical research as of the declaration of Helsinki. This study was approved by the Institutional Review Board of the American University of Sharjah. The IRB protocol was entitled “Cognitive Assessment and Enhancement” (IRB-19-513) and was approved on 31 March 2020.

2.2. Stress task

In this study, the Stroop Color-Word Test (SCWT) was used to induce stress levels. Six color words were displayed in a random manner on the computer monitor as depicted in Fig. 1(A).

Fig. 1.

Task presentation and time window. (a) Interface of the SCWT questions. The variables ‘Others’ and ‘You’ display the performance of individual in real-time. (b) Experiment conditions and sampling, and (c) Time windowing for each experiment condition. The SCWT stands for Stroop color word task and the ‘+’ sign is for rest state.

Only one word was displayed at a time and the answers of the color word to be matched are presented in random sequences in which participants have to select the right answer. Answering incorrectly or failing to answer each question within the allocated time, results in receiving a feedback message of “Correct’’ or “Incorrect” or “Time is out”. The behavioral data of reaction time and accuracy of detection were recorded simultaneously for statistical analysis.

To mark the start and the end of the SCWT in each block and experiment condition, different markers were used. For more details on the SCWT, the reader can refer to our previous vigilance studies [3], [59].

Figure 1(B) shows the four different cognitive conditions; high vigilance, vigilance enhancement, mental stress and stress mitigation. To avoid the order effect or fatigue on the participants, we conducted the experiment in counterbalanced manner. As such, half of the participants started their experiment without BBs and the other half started their experiment with BBs. The experiment began by collecting SAA levels for the baseline followed by training each participant on the SCWT. Then, each participant wore the fNIRS cap and recorded 20 seconds of rest followed by ten alternating trials of task and rest as shown in Fig. 1(C). In each condition, the task was presented for 30 s followed by 20 s rest. During the 30 seconds task, subjects were asked to promptly and accurately answer as many SCWT questions as possible. While resting, subjects were asked to relax by concentrating on the displayed white cross. In the high vigilance condition (HV), all participants performed the SCWT in block design as shown in Fig. 1(C). The HV condition is a state in which participants are fully alert.

Similarly, all participants underwent the same SCWT with a time limit of -20% of the time allocated during the HV to induce mental stress (MS). Meanwhile, in the vigilance enhancement condition (VE), the participants performed the task while listening to 16 Hz BBs. In this scenario, VE is a state of increased alertness due to the integration of audio stimulation with the SCWT. In the stress mitigation condition (SM), the participants performed the same mental stress task while listening to the 16 Hz BBs. In each condition, the experiment lasted for approximately 10 min .The BBs were generated by applying 250 Hz and 266 Hz pure tones to the right and left ears respectively using stereo headphones (MDR-NC7, Sony). The tones were played continuously while performing the SCWT. The sound intensity was set to 60 decibels which is equivalent to normal hearing level of 48 decibels. Figure 1 (C) shows the distribution of the experimental period for all sessions, which was roughly 60 minutes.

2.3. Salivary alpha-amylase (SAA)

Previous studies have utilized SAA as a non-invasive biomarker of mental stress [60]. It has demonstrated that SAA level significantly increase after performing stress task such as the Trier Social Stress Test (TSST) for approximately 10 minutes [61]. In line with this, we used the SAA to quantify the levels of stress induced while performing SCWT under time pressure and negative feedback. In this study, the amylase activity was measured using a hand-held meter (COCORO meter, NIPRO, Osaka, Japan). Saliva was sampled by immersing a salivary-sampling strip in saliva under the tongue for ∼40 seconds. Then the strip was immediately inserted in an automatic saliva transfer system device, where it was compressed into alpha-amylase test paper. The saliva intensity reading was calculated and displayed on the COCORO meter monitor, and the level of stress was assessed. We collected the salivary alpha-amylase samples to ensure that the performed SCWT with or without BBs induced stress/mitigation to all participants. Throughout the experiment, five samples of salivary alpha-amylase were taken from each participant. The first sample was collected five minutes before the beginning of the experiment as a baseline sample. The second sample was collected immediately at the end of the high vigilance condition. The third sample was collected immediately at the end of the vigilance enhancement condition. The fourth sample was collected immediately after the stress condition and fifth sample was collected after the stress mitigation condition. The sample collection time frame is shown in Fig. 1(B).

2.4. fNIRS Data acquisition

The cerebral activity (HbO2 and Hb hemoglobin) was measured using the NIRSport2 system (tandem NIRSport 2; LLC NIRx Medical Technologies). Eight source probes (S1- S8) and seven detector probes (D1- D7) were plugged into holders and arranged into a cap based on the international 10–20 system as shown in Fig. 2(a). A total of 20 channels were defined covering the prefrontal cortex, ventrolateral cortex and dorsolateral cortex areas. Sources and detectors are designed to have a 3 cm distance between each other. The intersection between the left and right tragus and the Nasion and Inion was the center of the cap, which was denoted by the Cz position. A dark black over-cap covered the cap to block external light luminance. The infrared signal is emitted in two wavelengths (760 nm, 850 nm) and collected at a sampling frequency of 10.17 Hz. Figure 2 (b) shows an example of the experimental setup while doing the SCWT.

Fig. 2.

a) fNIRS source and detector configuration in the international 10–20 system. The labels S1 to S8 represent the fNIRS sources and the Label D1 to D7 represent the fNIRS detectors. b) Experimental setup and data acquisition.

2.5. fNIRS data preprocessing

No standardized method has yet been established for fNIRS data analysis, however, various approaches have been reported [62,63]. In this study, the fNIRS data was preprocessed using the NIRS Brain AnalyzIR toolbox [64] with a custom script developed by our research group [65,66]. First, the absorbed NIR-light was transformed into oxy-hemoglobin (HbO2) and de-oxy hemoglobin (Hb) levels using the modified Beer-Lambert law. Second, we used Temporal Derivative Distribution Repair (TDDR) [67] and principal component analysis (PCA) [68] for movement artifact correction of sharp peaking signals. Third, band-pass filtering between 0.01 Hz to 0.2 Hz was applied to reduce the physiological noises caused by cardiac activities and respiration. Since our previous findings on stress studies [65,69] supported the notion that the HbO2 signals had more vigorous effects than the Hb signals, only the HbO2 concentration signal is chosen to compute the functional connectivity network (FCN).

2.6. Functional connectivity network

The functional connectivity network (FCN) is measured using phase-locking value (PLV) [70] by applying the Hilbert Transform on the bandpass-filtered epochs. We proposed the PLV due to its simplicity and since it does not require prior assumption about the data, thus it is better suited for the analysis of non-linear and non-stationary signals [71]. Besides, the PLV is less sensitive to noise and allows the evaluation of phase coupling without the influence of zero-lag interference. The instantaneous phase $\varphi {(t)}_H$ of signal $x(t)$ is calculated using the Hilbert transform as:

$$\varphi {(H)}_t=\text{arc tan}\left[\frac{x_H(t)}{x(t)}\right],$$ (1)

where $x_H(t)$ is the Hilbert transform of $x(t)$ and it is defined as:

$$x_H(t)=\frac{1}{\pi }P.V.\underset{-\mathrm{\infty }}{\overset{+\mathrm{\infty }}{\int }}\frac{x(\tau )}{t-\tau }d\tau ,$$ (2)

where the P.V is Cauchy’s principal value.

The PLV between channels u and v in a time window (PLV _uv ) is determined by:

$$PL{V}_{uv}=\frac{1}{n}\times \left|\sum _{k=1}^n{e}^{j[{\varphi }_u(t_k)-{\varphi }_v(t_k)]}\right|,$$ (3)

where ${\varphi }_u$ and ${\varphi }_v$ are the instantaneous phase of HbO2 concentration of channels u and v at time $t_k$ respectively, and n is the length of the time window within each block (here, n = 305).

The statistical significance of the estimated PLVs was determined by comparison with an empirical distribution of 100 surrogates, PLV maps (p < 0.05). For each subject, a 3D connectivity matrix contains the window-number × (channels× channels) was calculated. We then performed matrix binarization on the PLV using the orthogonal minimum spanning trees (OMSTs) method [33] to define the status of connection between channels. To the best of our knowledge this is the first study to use OMST with the fNIRS PLV connectivity for the assessment of vigilance and mental stress.

First, we constructed undirected weighted graph by assigning each pairwise PLV to the link (edge) between the corresponding channel (node) pair, as its weight. Then, a minimum spanning tree (MST) was generated using the Kruskal’s algorithm [72] for each iteration (MST_iter). From the weighted network (FW₁), the highest PLV value (short path length) in each node was chosen to form MST₁. Subsequently, n−1 edges were chosen in each iteration of MST except the last MST satisfying the constraint that is orthogonal, i.e., shares no common edges with the previous MSTs. With this iterative approach, we can get orthogonal MST and topologically filtering brain network by optimizing the global efficiency of the network constrained by the cost of keeping its connections [33]. The final MST is thus an undirected unweighted graph, including all N nodes (i.e., 20) with N − 1 links (i.e., 19).

The MSTs from 1 to iter were subsequently summed to form the OMSTs_iter according to Eq. (4) and its equivalent objective function of Global Cost Efficiency (GCE) was computed using Eq. (5): $J_{GCE}^{OMSTs}$ . The GE is the ratio of the weighted global efficiency of existing edges to that of the original weighted network FW₁ and the cost is the total strength of the existing edges divided by the total strength of original weighted network FW₁. The last iteration, k, would be the round-up integer of total edges (in the FW₁) divided by n−1. The algorithm would automatically identify the weighted OMSTs-filtered network with optimum $J_{GCE}^{OMSTs}$ as the true brain network [73] as given by Eq. (6). We only considered the number of edges instead of the weights; thus, the binary network (G) was extracted.

$$OMST{s}_{iter}=\sum _{i=1}^{iter}MS{T}_i,$$ (4)

$$J_{GCE}^{OMST{s}_{iter}}=GE-\text{cost},$$ (5)

$$G_w=OMST{s}_j$$ (6)

where $j\iff J_{GCE}^{OMST{s}_j}=\underset{i\in iter}{max}{J}_{GCE}^{OMST{s}_i},$ and G_w is the weighted MOSTs-filtered network [73]. We further compared the proposed OMST algorithm with the previous state-of-the art algorithm in [74] described below.

Global Cost Efficiency (GCE) [74]: A data-driven thresholding scheme with an objective criterion to maximize the formula of cost-efficiency in Eq. (7) of a network versus its cost that typically falls between the upper and lower limits of an economical small-world network. The cost is defined as the ratio of existing edges divided by the total number of possible edges. We searched for the optimum threshold that maximizes the GCE by iterating the absolute values between (0–1, with step size of 0.01).

$$GCE=GE-\cos t\text{ = }\frac{1}{n(n-1)}\left[{\sum }_{i,j\in N,j\ne i}\frac{1}{d_{ij}}-{\sum }_{i,j\in N}{a}_{ij}\right]$$ (7)

where GE is the global efficiency, N is the number of nodes, $d_{ij}$ is the short path length, and cost is the total edges of the graph. The basic difference of these two algorithms is the sampling of connections from the given network. The proposed algorithm is based on OMST while GCE is based on absolute threshold searching approach without any topological constraint. Besides, the GCE distinguishes weak from strong connections where mixture of both is essential to improve the design of reliable connectomic biomarkers [75].

Performance evaluation

• 1) Topological criterion:

  The topological criterion of the two algorithms were evaluated using the general quality formula given in Eq. (5). The values obtained of this formula range within the limits of an economical small-world for healthy subjects [76]. In the evaluation, we utilized the weighted networks after they were generated by masking the binary networks over the original PLV networks. In fact, the J_GCE is the measure of how well the essential structure of a network was captured while preserving its sparsity. Thus, the higher the J_GCE, the better is the network efficacy and the lower is the cost.

• 2) Graph Theoretical Analysis:

  To differentiate between the four mental states (high vigilance, vigilance enhancement, mental stress and stress mitigation) based on the proposed thresholding connectivity methods, we used network metrics. We employed the signal strength, nodal degree (ND), clustering coefficient (CC), modularity (Q), and local efficiency (LE). The full description of the graph theory analysis (GTA) measures can be found in our previous vigilance and emotion studies [59,70].

2.7. Laterality index (LI)

The lateralization of the functional connectivity network was evaluated using Eq. (8), where R and L correspond to the estimated local GTA metrics for the right and left networks, respectively. We computed the LI for each homologous channel pair (e.g., ch10–ch8, ch17–ch2, ch14–ch7, etc.). Equation (8) yields a LI value within the interval [−1, 1], with positive values indicating right lateralization and a negative value represents left lateralization. Significant leftward or rightward were only considered for LI distributions with means statistically different form zero (p < 0.05), after correction for multiple comparisons using the Holm-Bonferroni method.

$$LI=\frac{R-L}{R+L}$$ (8)

2.8. Statistical analysis

We performed multiple statistical analyses using MATLAB 2020a and the Statistical Package for Social Sciences (SPSS) version 25.0 (SPSS Inc., Chicago, IL, USA). To evaluate the effects of BBs on alpha amylase and behavioral responses of the group differences, we used One-way ANOVA with repeated measures. Before conducting the statistical analysis, we used the Kolmogorov-Smirnov method to test if the data is normality distributed [77]. The statistical analysis between fNIRS groups (HV vs VE; HV vs MS and MS vs SM) were conducted within the channel level using paired sample t-tests. The LIs were tested using one sample t-tests on each measure of the four groups compared to an alternative hypothesis that the mean is different from zero (0). All parametric statistics were carried out after confirming the normal distribution. The statistical differences were assumed to be significant if p-value was less than 0.05, p < 0.05.The p-values for all the measurements were corrected using Holm-Bonferroni method.

3. Results and analysis

3.1. SAA and behavioral data analysis

The measured values of the salivary alpha amylase collected five minutes before the experiment and immediately after each of the high vigilance, vigilance enhancement, stress, and stress mitigation conditions are shown in Fig. 3. The statistical analysis showed significant differences between all conditions, (F = 33.88, p < 0.00001, one-way ANOVA with repeated measures). The post hoc analysis showed a significant increase in the salivary alpha amylase level from the baseline to the high vigilance and to the stress condition with F= 6.44, p = 0.00011, and F= 15.00, p < 0.00001, respectively. Meanwhile, the analysis showed significant decrease in the salivary alpha amylase level from high vigilance to vigilance enhancement, (F= 4.74, p = 0.00884, one-way ANOVA with pairwise comparisons), and significant increase from high vigilance to stress, (F= 8.56, p < 0.00001, one-way ANOVA with pairwise comparisons). Similarly, we found significant increase in the salivary alpha amylase level from vigilance enhancement to stress condition, (F= 13.30, p < 0.00001, one-way ANOVA with pairwise comparisons). In addition, our statistical analysis showed significant decrease in the salivary alpha amylase level from stress to stress mitigation condition (F= 9.47, p < 0.00001, one-way ANOVA with pairwise comparisons). The overall results show that alpha amylase increases with increasing the level of stress and decreased by 44% with the stress mitigation method across all the participants.

Fig. 3.

Alpha amylase activity obtained from samples under different mental states. Error bars represent standard deviation of the mean across subjects. The asterisk ‘**’, ‘***’, and ‘****’ indicates the differences is significant with p < 0.01, p < 0.001 and p < 0.0001.

Likewise, the behavioral responses collected while performing the SCWT in the form of reaction time (RT) and accuracy of answering the SCWT questions under high vigilance, vigilance enhancement, stress and stress mitigation conditions are shown in Fig. 4. The statistical analysis showed significant differences between all conditions, (F = 36.53, p < 0.00001, one-way ANOVA with repeated measures). The post hoc analysis showed significant decrease in the RT from high vigilance to vigilance enhancement condition, (F= 7.08, p = 0.00001, one-way ANOVA with pairwise comparisons), and from stress to stress mitigation condition, (F= 3.80, p = 0.040, one-way ANOVA with pairwise comparisons), respectively. We found 20.18% reduction in the RT from high vigilance to vigilance enhancement and 14% from stress to stress mitigation condition. Similarly, the mean accuracies of answering the questions of SCWT under the high vigilance, vigilance enhancement, stress and stress mitigation conditions, are 96.0416%, 97.3113%, 54.7126%, and 72.4877% respectively. The obtained results showed that stress reduced the accuracy by 70% with (F= 23.05, p < 0.00001, one-way ANOVA with pairwise comparisons). Besides, the mitigation method improved the accuracy by 25% (F= 10.26, p < 0.00001, one-way ANOVA with pairwise comparisons). The reduction in RT and the improvement in the accuracy of target detection indicate that BBs significantly improve the behavioral performance of people under high vigilance and high stress conditions.

Fig. 4.

Behavioral reaction time and accuracy under different mental states. Error bars represent standard deviation of the mean across subjects. The asterisk ‘*’, and ‘****’ indicates the differences is significant with p < 0.05, and p < 0.001.

3.2. Topological criterion

In the topological criterion assessment, we compared the two thresholding algorithms OMST and GCE using the general quality formula J_GCE. Figure 5 shows the J_GCE values for the OMST (green violin) and for the GCE (yellow violin) thresholding algorithms across all the four cognitive conditions. The figure shows that, the OMST produced the highest J_GCE in all the cognitive conditions compared to GCE with p < 0.001, corrected with Holm-Bonferroni. The higher the J_GCE is, the better is the thresholding. Since OMST performed better in thresholding the PLV connectivity network, the rest of the analysis in this work is limited to the PLV connectivity network thresholded by OMST, (OMST- PLV). The detailed analysis of the GCE can be seen in Fig. 1, Fig. 2 and Table 1 in the supplementary material.

Fig. 5.

Quality assessment of OMST and GCE thresholding methods. The asterisk ‘***’ indicates the differences is significant with p < 0.001.

Table 1. Statistical Analysis between Stress Levels, corrected with Bonferroni.

High Vigilance vsVigilance Enhancement			High Vigilance vs Stress			Stress vs Stress Mitigation
Channel Pair	p-value	t-value	Channel Pair	p-value	t-value	Channel Pair	p-value	t-value
Ch1-Ch4	0.0227	-2.4061	Ch1-Ch5	0.0125	-2.6644	Ch1-Ch8	0.0439	2.1065
Ch1-Ch5	0.0227	-2.4075	Ch2-Ch6	0.0242	-2.3779	Ch2-Ch7	0.0460	2.0847
Ch1-Ch7	0.0269	2.3304	Ch2-Ch9	0.0101	-2.7536	Ch2-Ch9	0.0049	3.0496
Ch5-Ch8	0.0103	2.7446	Ch2-Ch11	0.0441	2.2011	Ch2-Ch11	0.0102	-2.7491
Ch7-Ch14	0.0024	3.3281	Ch2-Ch13	0.0017	3.4584	Ch5-Ch13	0.0383	-2.1710
Ch8-Ch14	0.0494	2.0508	Ch5-Ch12	0.0251	-2.3628	Ch8-Ch17	0.0413	-2.1359
Ch8-Ch17	0.0071	2.8985	Ch5-Ch13	0.0367	2.2001	Ch9-Ch15	0.0267	-2.3347
Ch9-Ch16	0.0407	-2.1422	Ch6-Ch13	0.0441	-2.1500	Ch10-Ch20	0.0451	-2.0940
Ch10-Ch13	0.0310	2.2678	Ch6-Ch17	0.0080	-2.8511	Ch13-Ch19	0.0202	2.4575
Ch10-Ch14	0.0422	2.1252	Ch9-Ch12	0.0347	-2.2256
Ch11-Ch19	0.0332	2.2358	Ch13-Ch19	0.0342	-2.235
Ch14-Ch20	0.0235	2.3905	Ch16-Ch17	0.0056	-3.0010
Ch16-Ch17	0.0410	-2.1386	Ch18-Ch20	0.0050	-3.0550

3.3. Connectivity analysis

We calculated the PLV connectivity matrix of the high vigilance, vigilance enhancement, stress and stress mitigation groups for each subject in each block. We then filtered the PLV connectivity networks using OMST and evaluated their performance using global criteria and GTA. The mean results of the filter PLV (OMST- PLV) for the four groups of stress are shown in Fig. 6. Among them, the dark-red area indicates a relatively more active node which shows that it has a higher synchronization level, and the white area indicates a lower degree of relevance between channels. Figure 6 (a), and (b) which represent the high vigilance, and vigilance enhancement groups, respectively, show that the red area for the vigilance enhancement group is larger than that of the high vigilance/alertness group. Similarly, the red area in Fig. 6 (d) is larger than that in Fig. 6(c) which represents the stress mitigation and stress groups respectively.

Fig. 6.

OMST-PLV connectivity network map, (a) high vigilance, (b) vigilance enhancement, (c) stress, (d) stress mitigation.

The difference in the OMST-PLV connectivity matrices between the four groups: high vigilance vs vigilance enhancement, high vigilance vs stress and stress vs stress mitigation were filtered out by a t-test at p < 0.05. The difference between each two groups has a range of values between -1 and +1. Figure 7 (a) shows the difference in connectivity network between high vigilance and vigilance enhancement groups. Channels with positive values indicate that the high vigilance group has significant higher connectivity network than the vigilance enhancement group and negative values indicate that the enhancement group has significantly higher connectivity at each pair of channels. Similarly, Fig. 7 (b) shows the difference in connectivity network between high vigilance and stress groups. Channels with positive values indicate that the high vigilance group has significant higher connectivity network than the stress group and negative values indicate that the stress group has significantly higher connectivity at each pair of channels. Likewise, Fig. 7 (c) shows the difference in connectivity network between stress and mitigation groups. Channels with positive values indicate that the high stress group has significant higher connectivity network than the stress mitigation group and negative values indicate that the mitigation group has significantly higher connectivity. Figure 7(d) shows the channel placement within the brain areas.

Fig. 7.

Difference OMST-PLV connectivity network between each two groups filtered out by a t-test (P < 0.05), (a) high vigilance and vigilance enhancement, (b) high vigilance and stress, (c) stress and stress mitigation, (d) channel configuration and placement over the head

High differences were found between high vigilance and vigilance enhancement group at the DLPFC area, frontopolar PFC, right VLPFC, and Orbitofrontal cortex. Similar significant results were also found between high vigilance and stress group at the DLPFC area, frontopolar PFC, orbitofrontal cortex and right VLPFC. Meanwhile, the significant results between stress and stress mitigation were found at DLFPC area, right VLPFC, Orbitofrontal cortex and right frontopolar PFC. The summary of all the statistical analysis in the form of t-values and p-values between the three pair of groups (high vigilance vs vigilance enhancement, high vigilance vs stress and stress vs stress mitigation) is given in Table 1. The higher the t-value, the higher the significant differences between conditions.

3.4. Graph theory analysis

The results of the GTA measures for all the fNIRS channels are presented in a topographical maps as shown in Fig. 8. The results are reported in the form of the difference between each two groups (high vigilance-vigilance enhancement; high vigilance-stress and stress-stress mitigation) with values between -0.1 and +0.1. Figure 8(a) shows the differences in the GTA measures between the high vigilance and vigilance enhancement states. The high vigilance state demonstrates significant decrease in the nodal degree in frontopolar area at channel (Ch7; t = 2.39, p = 0.023, and Ch15; t = 2.40, p = 0.022); significant increase in modularity at the right orbitofrontal area in channel (Ch13; t=-2.14, p = 0.04); and local efficiency in right frontopolar area at channel (Ch15; t=-2.28, p = 0.029), compared to the vigilance enhancement condition. There were no significant differences between high vigilance and stress state in all the GTA measures, as shown in Fig. 8(b). The stress mitigation experiment result in Fig. 8(c) shows significant increase in cluster coefficient at the right orbitofrontal area in channel (Ch14; t=-2.15, p = 0.04); in the modularity at left DLPFC in channel (Ch1; t=-2.13, p = 0.041); and in the local efficiency at the right orbitofrontal area in channel (Ch14; t=-2.39, p = 0.023), compared to stress condition.

Fig. 8.

GTA measures under different experimental conditions, (a) high vigilance and vigilance enhancement, (b) high vigilance and stress, (c) stress and stress mitigation. The asterisk ‘*’ indicates the differences are significant with p < 0.05.

3.5. Laterality

Figures 9 shows the laterality results for all groups and GTA metrics at each pair of channels. We did not find any significant lateralization for any of the four cognitive condition/group when the OMST-PLV network strength and cluster coefficient are used, p > 0.05, see Fig. 9 (a,c). Stress group shows significant right lateralization at channel paired of Ch16-Ch5 (DLPFC), p < 0.05 and high vigilant group shows left lateralization at channel paired of Ch19-Ch4 (orbitofrontal cortex), p < 0.043, when nodal degree is considered see Fig. 9(b). Likewise, the vigilance enhancement group shows significant left lateralization at channel paired of Ch19-Ch4 (orbitofrontal cortex), p < 0.042, when local efficiency is considered see Fig. 9(d). Moreover, stress group shows significant left lateralization at channel paired of Ch14-Ch7 (frontopolar PFC), p < 0.02, and high vigilance group shows significant left lateralization at channel paired of Ch13-Ch11 (orbitofrontal cortex), p < 0.025 as well as significant left lateralization at channel paired of Ch19-Ch4 (orbitofrontal cortex), p < 0.04, respectively, when modularity is considered see Fig. 9(e).

Fig. 9.

Laterality indexes under different experimental conditions, (a) OMST, (b) Node degree, (c) Cluster coefficient, (d) Efficiency, (e) Modularity. The channel labels represent the location within the brain region as demonstrated in Figure (7d).

4. Discussion

In this study, we investigated the use of 16 Hz binaural beat stimulation (BBs) to enhance vigilance/alertness and mitigate mental stress in the workplace. We conducted an experiment which included four cognitive conditions while participants are responding to the SCWT for approximately 60 minutes. We assessed the four types of cognitive conditions using the behavioral responses of answering the SCWT, alpha amylase level, and the brain hemodynamic signals measured by fNIRS. We quantified the level of vigilance and stress using multiple statistical analysis methods and analyzed the brain connectivity network using PLV with graph theory analysis (GTA). In the connectivity network estimation, we investigated two automated thresholding connectivity network methods based on orthogonal minimum spanning trees (OMSTs) and global cost efficiency (GCE). To the best of our knowledge, this is the first fNIRS study to automate the functional connectivity network using the OMST and GCE with PLV as well as to investigate the effectiveness of BBs on vigilance enhancement and stress mitigation at the same time.

We found a significant increase in the alpha amylase level from baseline to high vigilance condition, and from high vigilance to stress condition, p < 0.0001. On the other hand, we found significant decrease in alpha amylase level from high vigilance to vigilance enhancement conditions and from stress to stress mitigation conditions, p < 0.01 as shown in Fig. 3. Likewise, we found that the behavioral responses significantly change under stress as it reduced the accuracy of answering the questions of SCWT and increased the reaction time to stimuli. Additionally, we found significant improvement in the accuracy of target detection and reduction in the RT when answering the SCWT while listening to the 16 Hz BBs during the vigilance enhancement and stress mitigation states. The assessment of topological criteria of the connectivity network showed that the proposed OMST outperformed the GCE in thresholding the connectivity networks under all the four experiments/states, p < 0.001. Therefore, the analysis of functional connectivity network was limited to OMST-PLV. Thus, our estimated functional connectivity network showed significant increase from high vigilance to vigilance enhancement condition at various areas within the DLPFC, Orbitofrontal Cortex, right frontopolar PFC, and right VLPFC and reduced at the DLPFC and left orbitofrontal cortex. Also, the connectivity network significantly increased from high vigilance condition to stress condition at the DLPFC, frontopolar PFC, right orbitofrontal cortex and right VLPFC and reduced at the left DLPFC and orbitofrontal cortex, p < 0.05. Similarly, the connectivity network showed significantly increased from stress condition to stress mitigation condition at the left DLPFC, right orbitofrontal Cortex and left frontopolar PFC and decreased at the areas within the DLPFC, medial orbitofrontal cortex, right frontopolar PFC and right VLPFC area, p < 0.05.

4.1. Stress levels, alpha amylase and behavioral responses

In this study, we found that the amount of alpha amylase level changes with the experiment conditions. The alpha amylase level is low at baseline or rest state and has increased by 12 and 25 (U/ml) while performing the SCWT at high level of vigilance and under stress condition, respectively. The 25 U/ml increase under stress condition is consistent with our previous studies that utilized arithmetic task complexity with time pressure as the stressors [78]. The consistency in the results of alpha amylase across the studies confirms the inducement of stress using time constraint with negative feedback as the stressors regardless of the cognitive task, i.e in this study, we used SCWT and in the previous studies we used mental arithmetic task. Interestingly, the increase in the alpha amylase level under stress in this study was also associated with a decrease in the accuracy of target detection, (77%). This is a well-established phenomena in which stress disturbs cognitive functioning and decision making [79]. This can be explained by the failure to meet demands posed by the SCWT under time constraint. It has been demonstrated that if the task is excessively difficult, then the likelihood of the subject to make errors and disengage with the task is high [80]. Based on this, we can speculate that most of the subjects failed to stay engaged with the SCWT when presented with under time pressure, therefore resulting in an overall lower performance.

However, the amount of alpha amylase level did not significantly change from baseline to vigilance enhancement condition, i.e when people performed the SCWT while listening to the 16-Hz BBs, p > 0.05. This indicates that the binaural beats stimulation encourages certain behaviors such as increase in focus, relaxation and creativity. In the meantime, the amount of alpha amylase level showed a significant decrement from high vigilance level to vigilance enhancement (-9 U/ml, p < 0.01) and from stress to stress mitigation condition (-16 U/ml, p < 0.001). These decreases in the alpha amylase levels were concurrent with a significant increase in the accuracy of target detection (25%) and reduction in reaction time to stimuli (14%), p < 0.01, suggesting a performance improvement. The obtained objective and behavioral results (reaction time) confirm the effectiveness of BBs in enhancing cognitive vigilance and mitigating stress level. It seems that at higher vigilance condition, the BBs caused relaxation, thus it reduced the amount of alpha amylase level, while at stress condition the effects of BBs can be interpreted as that it stimulates new brain cell formation. Our obtained result is in line with previous studies that demonstrated the effectiveness of using BBs to enhance vigilance and reduce anxiety level [44,55]. Another interpretation in the reduction of alpha amylase level and improvement in the behavioral responses while listening to the BBs could be due to the activation of multisensory responses (visual by SCWT and auditory by BBs) [81]. It has been demonstrated that activating many senses at the same time drives positive mental states such as tranquility [82]. Thus, the overall objective and behavioral results in this study recommends that BBs reduces stress levels and improves reaction times during vigilant and stress work. Besides, BBs may also provide work performance improvement while performing stressful work.

4.2. Stress levels and connectivity network

The obtained functional connectivity networks showed different patterns within each experiment condition as depicted in Fig. 6. For example, at high vigilance condition, the highest connectivity networks were found across DLPFC and the left frontopolar PFC area. In the vigilance enhancement condition, the right frontopolar PFC and DLPFC areas showed the highest connectivity network compared to the other regions. In the stress condition, the highest connectivity networks were found towards the left DLPFC area. Meanwhile, at the stress mitigation condition, the right frontopolar PFC and left DLPFC showed the highest connectivity network compared to the other regions. Interestingly, we found common patterns of high connectivity networks across all the experiment conditions within the channels located at the left hemisphere. These higher connectivity networks within the left hemisphere could be due to the Stroop interference effect [83]. Previous hemodynamic studies have reported that SCWT in younger adults activate the left hemisphere more than the right hemisphere [84].

Our analysis of connectivity networks across the four different experiment conditions revealed several statistically significant trends. Specifically, we found that solving the SCWT under high vigilance state while listening to BBs increased the connectivity network between the DLPFC and the left frontopolar PFC area compared to that without BBs as demonstrated in Fig. 7. Likewise, solving the SCWT under stress condition while listening to BBs increased the connectivity network between the right frontopolar PFC and left DLPFC compared to the stress condition alone. The increment in the connectivity networks between the right frontopolar PFC and left DLPFC could be due to the activation of the auditory primary area in the cortex. Previous studies have shown that BBs stimulation improves the ability to exchange information between the auditory cortical regions [41,57]. Previous neuroimaging studies using BBs have provided valuable information on the spatial localization of brain activations during meditation [85]. A recent fNIRS study has also demonstrated a significant increase in the hemodynamic responses in the DLPFC associated with a decreased reaction time after meditation (loving-kindness), suggesting improved attentional focus and behavioral performance [86]. Another study had found that imagined music performance increased the functional connectivity of the angular gyrus with the superior frontal gyrus and medial prefrontal cortex [87]. In addition, previous meditation fNIRS studies on a veteran and a beginner practitioner during Taichi-quan twice reported a significant increase in blood flow in the prefrontal cortex [88]. Those findings indicated increased regional cerebral blood flow during meditation in the prefrontal cortex, which suggests increased attentional demand of meditative tasks and alterations in self-experience. Overall, the obtained functional connectivity results suggest that BBs resulted in improvements in cognitive, and behavioral outcomes.

Nevertheless, we found a significant increase in the connectivity network under stress condition compared to high vigilance as seen in Fig. 7(b). In particular, the utilized stressors in this work increased the connectivity networks between ventrolateral and dorsolateral prefrontal cortex. We posit that the higher connectivity could act as a defense mechanism against stress-related deterioration of cognitive functions. Meanwhile, a significant decrease in the connectivity networks was also observed under stress between left dorsolateral and frontopolar /ventromedial prefrontal cortex (vmPFC). This is in line with previous stress studies that reported reduced connectivity between the vmPFC and dorsolateral prefrontal cortex [89,90,91].

4.3. Graph theory analysis under different cognitive conditions

The results of the GTA measures under high vigilance condition showed significant decrease in node degree at the medial frontal cortex, increased in the local efficiency and modularity on the right frontal cortex as compared to the vigilance enhancement condition. Likewise, under stress mitigation condition, there was a significantly higher cluster coefficient and local efficiency on the right frontal cortex and higher modularity at the left dorsolateral cortex, compared to the stress condition. Our experimental results suggest the increase in the frontal connectivity as an indicator of improving the regional function. In particular, a more concentrated distribution of the increased connectivity within the right hemisphere would lead to the more efficient local information processing ability to resist the brain function decline. This is consistent with a previous EEG working memory study that reported significant improvement in the connectivity network while listening to 15 Hz of BBs [48]. The overall decrease/increase in the GTA measures within the right hemisphere indicates the leading role of the right hemisphere when listening to binaural beat stimulation when under stress condition.

4.4. Brain asymmetry under different cognitive conditions

Our analysis of asymmetry across the four different conditions revealed different laterality trends. The high vigilance state showed left laterality in node degree and modularity within the orbitofrontal cortex. This is in agreement with previous fNIRS activation and connectivity studies that showed left lateralization during SCWT [92]. In the stress condition, there was a right laterality at DLPFC, and left laterality at the frontopolar PFC when modularity and node degree are considered, respectively. The left laterality in our study is in line with previous stress studies that used SCWT as the stressor [93]. In the vigilance enhancement condition, there was a left laterality at the orbitofrontal cortex within the local efficiency. This is in agreement with previous fNIRS Qigong meditation study that reported increased in blood flow on the left prefrontal cortex in practitioners as compared to nonpractitioners [94]. Nevertheless, there was no asymmetry under stress mitigation condition suggesting that BBs reduced the laterality caused by stress.

4.5. Limitation of the study

This study has some limitations that could be addressed in future research. First, we only investigated the effects of one frequency of BBs (16 Hz) in enhancing vigilance and mitigating stress level. In future work, we plan to investigate other frequencies such as the gamma or alpha bands (higher or lower than 16 Hz). Besides, comparison of different binaural beats against a control stimulus should be investigated. Second, the fNIRS measured cortical hemodynamic responses as well as noises. In future studies, scalp effect and systemic noise should be further minimized using simultaneous short-distance measurement and pulse oximeter [95,96]. Third, we estimated the connectivity network using undirected method, directed connectivity can illuminate the potential cerebral activity mechanism in controlling life activities more accurately and comprehensively [97]. Fourth, our experiment was conducted on healthy young people during COVID-19 pandemic. Early evidence from the UK, US, Korea and Australia indicates that younger people have had the greatest increase in rates of psychological distress during the pandemic [98]. Finally, gender balance and preferences will be considered in our future work.

5. Conclusion

In this study, we investigated the use of 16 Hz binaural beats stimulation to enhance cognitive vigilance/alertness and to mitigate stress levels. We evaluated the effectiveness of BBs using salivary alpha-amylase, behavioral responses and hemodynamic brain responses measured by fNIRS. We found that listening to the 16 Hz binaural beat stimulation while under high vigilance enhanced the behavioral performance by 14% and mitigate the level of stress by 20%. Besides, we found 44% decrease in the stress hormone while listening to binaural beat stimulation under stress, which indicates stress mitigation. Our analysis of connectivity network across the four different conditions revealed several statistically significant trends. Specifically, the connectivity network under vigilance enhancement showed significant increase between the right and left DLPFC, and left orbitofrontal PFC. Likewise, the connectivity networks under stress mitigation condition showed significant increase between the right and left DLPFC, orbitofrontal cortex, right VLPFC and right VLPFC. Also, the connectivity network under stress condition showed significant increase between ventrolateral and dorsolateral PFC areas. The laterality index demonstrated left orbitofrontal PFC laterality under high vigilance and vigilance enhancement states, and right DLPFC and left frontopolar PFC while under stress. Overall, our results confirm that listening to 16 Hz binaural beat stimulation significantly improve the vigilance and mitigate stress levels by improving the cognitive efficacy, and behavioral outcomes.

A typographical correction was made to the sixth author's name.

Funding

American University of Sharjah10.13039/501100002724 (FRG20-L-E25).

Disclosures

The authors declare that there are no conflicts of interest related to this article.

Data availability

Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the corresponding author upon reasonable request and under a licensing agreement.

Supplemental document

See Supplement 1 ^{(1.2MB, pdf)} for supporting content.