For this study, 20 healthy students of Tecnológico de Monterrey (six women and 14 men) aged between 19 and 24 years old were recruited (i.e., a convenience sampling method was undertaken). All of them reported not having musical experience and voluntarily consented to their participation in the study. This study was previously approved by the Ethical Committee of the Medicine School at Tecnológico de Monterrey (CONBIOETICA-19-CEI-011-20161017).

To record EEG activity, the mBrainTrain system was used. This is a Bluetooth-interface EEG device of 24 channels (Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T7, T8, P7, P8, Fz, Cz, Pz, M1, M2, AFz, CPz, POz) positioned according to the 10/20 international system, as shown in Figure 1. The channels M1 and M2 were set as references and the channel FCz as ground. The sampling frequency was 250 Hz. The mBrainTrain has the Smarting Streamer software, which was used to verify electrode impedances to be below 5 kΩ. To set up the experimental paradigm, OpenVibe was employed. OpenVibe is a free software commonly used in the neuroscience field to design and test brain-computer interfaces and to develop experimental paradigms for offline records (Renard et al., 2010).

To perform the audiometry, an audiometer Interacoustics-AD226 was utilized. Tonal audiometry within ranges from 250 Hz to 2 kHz was used to determine auditory thresholds. Measurements were taken in dB HL. Finally, HR was taken with pulse oximeter Hergom-MD300 and auditory stimuli was given through open-back headphones SHURE-SRH1840.

According to Klimesch (2013), biosignals do not vary randomly or arbitrarily. Namely, brain and body signals oscillations are aligned with each other and form a single frequency architecture. The interaction between brain and body may be described as a complex system that couples and decouples according to a specific harmony frequency described by,

where s is the scaling factor, i refers to the biosignal of interest, and f is the fundamental frequency of the biosignal oscillation. When i = 0, f_d refers to cardiac activity. When i < 0, f_d refers to breathing rhythms (including Mayer waves that are the lowest frequency in the respiratory process), blood pressure waves, rhythmic fluctuations in the blood oxygen level-dependent (BOLD) signal at intrinsic mode fluctuations, and gastric waves. When i > 0, f_d refers to brain oscillations [delta (i = 1), theta (i = 2), alpha (i = 3), beta (i = 4), gamma (i = 5)]. In addition, upper, and lower frequencies of each fundamental frequency can be, respectively, estimated by,

Recently, Klimesch (2018) empirically demonstrated that the resonance of a biosignal is harmonized with other ones at resting state. Some examples are:

In the light of the above evidence, this research proposes the design of personalized BB based on individual HR. That is, when i = 0 in Eq. (1) then,

Having estimated s, central frequency (f_c) of theta and beta EEG rhythms were calculated in accordance with the individual HR at hand. Note that resulting fundamental frequencies were not subjective, they were rather relative to the human body function, and their values are around the well clinically established frequency bands.

As theta BB consisted of pure tones of 500 Hz for the left ear, f_c of theta EEG band was used to adjust the frequency for the right ear (F_{R_ear}). Similarly, beta BB consisted of pure tones of 500 Hz for the left ear. Then, f_c of beta EEG band was used to adjust the frequency for the right ear. That is,

The resulting EEG frequency bands for each volunteer, and the corresponding BB produced after the individual EEG frequency band identification, are reported in Table 1. The computational algorithm to generate personalized BB in accordance with the BBC theorem was programmed in MATLAB programming language and was published in MathWorks File Exchange Forum¹.

Each volunteer participated in two BB sessions on different days. In the first session (S1), theta BB was applied, and in the second session (S2), beta BB was used. Both BB exposures were for 20 min, since listening to BB longer than 20 min may lead to mental fatigue (Jirakittayakorn and Wongsawat, 2017a). All the participants were seated on a comfortable chair and in a quiet room. All of them were asked to keep their eyes closed during BB stimulation. The procedure was conducted in two steps: (1) environment for binaural stimuli and (2) presentation of binaural stimuli.

Power spectral density is a method to extract the power content of a signal in the frequency domain. Power spectral density (PSD) utilizes the Discrete Fourier Transform to obtain the periodogram. Welch’s method is one of the most applied algorithms to estimate PSD (Zhang, 2019). In this study, PSD was calculated for the three different mental states: (1) S_R, (2) S1, and (3) S2. The analysis was carried out in two steps: (1) preprocessing and (2) processing.

Statistical analysis was carried out for AP and RP estimates. With respect to AP estimates, the comparison was between sessions (1) S_R with S1, and (2) S_R with S2. First, the Shapiro–Wilk test for normality (Ahad et al., 2011) was applied for these two pairs of data. Once normality was confirmed for both cases, a separate t-test was performed to find significant differences between S_R-S1 and S_R-S2 through all channels. Afterward, Cohen’s d effect size was utilized to estimate the magnitude of these differences as “negligible” (d < 0.2), “small” (0.2 ≤ d < 0.5), “medium” (0.5 ≤ d < 0.8), or “large” (d ≥ 0.8) (Magnusson, 2021). Regarding RP estimates, normality was also tested using the Shapiro–Wilk test for the power difference between theta, alpha, and beta bands in both S1 and S2 in every channel. Subsequently, two alternately two-way ANOVA tests were performed to RP in S1 and S2. The tested values included averaged RP values, channels, and bands. These two ANOVAs were aimed to verify if statistically significant changes in RP amongst EEG bands and channels were achieved after BB stimulation. Finally, a Tukey test was realized (Daniel and Cross, 2013) to locate these differences. A significance level of 0.05 was used in all statistical tests.

Absolute power from S1 and S2 were compared with S_R to identify if BB exposure elicited changes in neural activity. In Figure 5, it is exhibited average AP from the three conditions in dB within the frequency range of 0–30 Hz. PSD was estimated across all channels.

At an initial glance, it seemed that average S1, S2, and S_R triggered similar brain activity. To dismiss this issue, a paired t-test was applied to AP values of (1) S1 with S_R and (2) S2 with S_R, for all channels. Significant differences were found across all the channels in both comparisons (p < 0.05). However, a Cohen’s d effect size test was implemented to estimate the magnitude of these differences. Cohen’s d values are graphically expressed in Figure 6.

In order to confirm if brain entrainment was achieved, the following conditions must be met: (1) for theta BB, an increase in S1/S_R ratio in the theta band, or (2) for beta BB, an increment in S2/S_R ratio in the beta band. Thus, RP from S1/S_R and S2/S_R were compared across theta, alpha, and beta bands. Figure 7 shows RP from all channels in dB, throughout EEG frequency ranges where theta, alpha, and beta were colored in blue, green, and red, respectively.

So far, it is known that statistical differences in all channels were identified between BB sessions and resting state, but it does not confirm if brain entrainment was attained. For this reason, a two-way ANOVA was consecutively applied to RP from S1 and S2 to verify if neural activity between bands were dissimilar due to BB. The ANOVA data for each matrix contained RP values from S1 or S2, channels, and bands (theta, alpha, and beta). Significant differences were found for alpha-theta bands (p = 0) and alpha-beta bands (p = 0). Nevertheless, no statistical difference was obtained for theta-beta (p = 0.5577).

As can be seen from Figure 6, T7 and T8 showed the highest differences when BB sessions were compared against S_R, followed by medium differences in P7, P8, and O2. After listening to either theta or beta BB, small to medium changes in AP occurred in the remaining channels, suggesting that significant effects due to BB were not entailed or, in other words, brain entrainment was not achieved. As the greatest changes were seen over the temporal lobe, followed by the parietal one, they may be associated with auditory perception and sound location (Benarroch, 2006; Bizley and Cohen, 2013; Goldstein, 2014).

Since data standardization was carried out, a 0 dB value means that BB session and baseline neural activity were comparable. On the contrary, when dB > 0, brain activity was stronger during BB exposure than baseline. Thus, either theta or beta BB elicited neural synchronization. Similarly, given a dB < 0, it implies that brain activity was higher in the resting state in comparison with the BB session, which means that neural desynchronization was induced (Watkinson and Clarke, 2018).

According to Cohen (2014), an oscillation from −2 dB to +2 dB implies a percentage change from −36.9 to +58.8%. In other words, if RP was equal to −2 dB, it means that baseline activity was higher than in BB session by 36.9%, (i.e., BB induced 36.9% of neural desynchronization). Moreover, a +2 dB change means that the BB session had stronger brain activity than baseline by 58.8% (i.e., BB triggered 58.8% neural synchronization) (Cohen, 2014).

Interestingly, RP in the parieto-occipital region was lesser than −2 dB for alpha EEG band while listening to theta BB, specifically over Pz, P4, P8, O1, POz, and O2 recording sites. This decrease in dB value is explained by greater brain activity at the resting state regarding theta BB. For these channels, as alpha desynchronization occurred only for S1, we suggest that theta BB probably disturbed volunteers instead of inducing them into meditative or relaxed states, even though when they were just seated with closed eyes and doing no task. A theory of this behavior may be related to volunteers feeling uncomfortable when listening to the theta BB (Crespo et al., 2013; Lee et al., 2019).

In conclusion, neural resynchronization was met with both personalized theta and beta BB, but no different mental conditions seemed to be achieved.

The aim of the study was mainly limited due to (1) conditions of participants, and (2) the use of open-back headphones. First, for more specificity in sample selection, hours of sleep in the night before the study and psychological conditions, such as stress or anxiety, should have been considered. These factors can disturb brain oscillations. Second, as BB was delivered through open-back headphones, the environment auditory stimuli could bias neural information. Therefore, the study should have been implemented in an isolated room.

In order to develop a full picture of brain activity due to BB exposure, a suggestion would be that the application of other kinds of BB not used in this study such as alpha BB (Park et al., 2018; Shekar et al., 2018) or gamma BB (Colzato et al., 2017; Shekar et al., 2018). Additional studies that give an insight into brain signal modulation are needed for further understanding of which effects BB induces on humans. Empirical studies where BB effects are behaviorally measured are not enough to demonstrate binaural sound influence on the human mental state.