High and low pitch sound stimuli effects on heart-brain coupling

Abstract

This study aimed to explore the influence of sound stimulation on heart rate and the potential coupling between cardiac and cerebral activities. Thirty-one participants underwent exposure to periods of silence and two distinct continuous, non-repetitive pure tone stimuli: low pitch (110 Hz) and high pitch (880 Hz). Electroencephalography (EEG) data from electrodes F3, F4, F7, F8, Fp1, Fp2, T3, T4, T5, and T6 were recorded, along with R-R interval data for heart rate. Heart-brain connectivity was assessed using wavelet coherence between heart rate variability (HRV) and EEG envelopes (EEGE). Heart rates were significantly lower during high and low-pitch sound periods than in silence (p < 0.002). HRV-EEGE coherence was significantly lower during high-pitch intervals than silence and low-pitch sound intervals (p < 0.048), specifically between the EEG Beta band and the low-frequency HRV range. These results imply a differential involvement of the frontal and temporal brain regions in response to varying auditory stimuli. Our findings highlight the essential nature of discerning the complex interrelations between sound frequencies and their implications for heart-brain connectivity. Such insights could have ramifications for conditions like seizures and sleep disturbances. A deeper exploration is warranted to decipher specific sound stimuli’s potential advantages or drawbacks in diverse clinical scenarios.

Introduction

Investigating the synchronization of frequency bands from two distinct biological signals offers a valuable lens into the mechanisms mediating these interactions [1]. However, the synchronization between heart rate variability (HRV) - representing the variability of successive RR intervals - and electroencephalography (EEG) activity has been primarily studied in specific contexts such as neurologically healthy premature infants [2], during epileptic seizures and in anesthetized individuals [3, 4]. Notably, many sensory inputs, including auditory stimuli, can instigate cardiovascular, cerebrovascular, and respiratory changes, predominantly mediated by the autonomic nervous system (ANS) [5]. Auditory stimuli have been known to induce myriad cardiovascular alterations, including shifts in HRV [5, 6]. Synchronized fluctuations in EEG and HR are postulated to shed light on the interplay between cortical, thalamocortical, and central autonomic brain regions [3, 7]. However, the relationship between frequency band synchronization from ECG (specifically RR interval variability) and EEG signals in healthy individuals remains an area rife with questions, especially considering the constant sensorial stimuli the central nervous system encounters, potentially influencing heart-brain communication.

The intricate interactions between the heart and brain, two paramount human organs, are central to holistic health [8]. Yet, there’s a discernible void in literature delving into coherence measurements between these two vital signals [9, 10], highlighting an unmet research need. While EEG and HRV are established modalities to study brain and heart activities, their distinct frequency characteristics pose challenges [10, 11]. For instance, EEG signals are typically divided into five frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), Beta (12–30 Hz), and gamma (30–80 Hz) [8, 10]. Conversely, HRV signals mainly bifurcate into high-frequency power (HF: 0.15–0.4 Hz) and low-frequency power (LF: 0.04–0.15 Hz) domains [11]. These discrepancies necessitate the application of an envelope to the EEG signal to render it commensurate with HRV data [3, 12]. The Hilbert transform, a preeminent technique, facilitates the extraction of a signal’s envelope [13, 14]. Applying this transform to EEG data allows for deriving its amplitude envelope, aligning it for coherence analysis with HRV [3].

Probing into the effects distinct auditory stimuli exert on heart-brain interactions could unearth nuanced insights into their communication dynamics. Accordingly, this study aims to discern the impacts of high- and low-frequency sound stimuli on heart-brain synchrony, as examined through wavelet coherence between heart rate variability (HRV) and electroencephalogram envelopes (EEGE).

Methods

Thirty-one volunteers participated in this study, including 19 females (average age 29.4 ± 5.8 years) and 12 males (27.8 ± 6.6 years). The University Ethics Committee (CEP/UNITAL # 507/11) comprehensively reviewed all aspects of this research, and every participant provided written informed consent before testing. All participants possessed normal hearing capabilities and affirmed they were in good health without any diagnosed neurological conditions.

Experimental design

Study volunteers laid supine in a silent, darkened room with their eyes closed. Using wired noise-canceling headphones (Bose QuietComfort 15, Bose Corporation), participants were exposed to either silence or one of two continuous, non-repetitive pure tone stimuli with distinct base frequencies: low pitch (110 Hz) and high pitch (880 Hz). Each exposure lasted 120 s (as depicted in Fig. 1). The sequence of pitch exposure was randomized among the study volunteers.

Fig. 1.

Experimental Design: Participants were positioned lying down with their eyes closed in a quiet, dark room. Using noise-canceling headphones, they were exposed to one of three auditory conditions: silence, low-pitch pure tone stimulus (110 Hz), or high-pitch pure tone stimulus (880 Hz). Each sound stimulus lasted for 120 s, with the order of pitch presentation randomized across participants. Concurrently, electroencephalography (EEG) from 10 distinct channel locations and heart rate R-R intervals were recorded

Data collection

EEG data were recorded at a 200 Hz sampling rate, with a band-pass filter between 0.01 and 100 Hz, a 60 Hz Notch filter, and a gain setting of 20,000. A BrainNet BNT 36 system was employed, equipped with 32 channels and 1.5 m lead/tin electrodes. To focus on our regions of interest and to ensure statistical significance, data from ten specific channels were collected, all of which maintained an impedance of 5-10 K ohms. The electrode placements adhered to the International 10–20 system as outlined by Jasper, with reference electrodes located in both ears [15]. The EEG data specifically emanated from regions F3, F4, F7, F8, Fp1, Fp2, T3, T4, T5, and T6. Based on existing research, these channels were chosen due to their significance in auditory processing and emotional regulation [16, 17]. Notably, these regions are largely associated with the frontal and prefrontal cortex – zones instrumental in executive functions, decision-making, and emotional regulation [17]; the temporal lobes – pivotal in auditory processing, memory, and emotion [16]; and the posterior facet of the temporal lobes adjacent to parietal regions – significant in auditory processing and sensory information amalgamation [18]. Besides EEG data, heart rate R-R intervals were monitored using a Polar V800 device.

Data analysis

MatLab (R2017b, The Mathworks, Inc.) was employed for data processing. Analyses were structured around the five stages delineated in Fig. 1.

HRV signal processing

Initially, for the HRV frequency domain analysis, heart rate beat-to-beat (R-R) interval data were converted into binary sequences with a 1 ms bin width. This yielded a time series of zeros and ones sampled at 1000 Hz [19]. This method accentuates the frequency distribution of heartbeats while discarding waveform nuances of the cardiac electrical activity. This data transformation hinges on the insight that the primary interest in HRV analysis is the interval distribution between heartbeats, not the precise waveform or amplitude of signals [19, 20].

EEG signal processing

Filtered EEG data were segmented into epochs. Before artifact detection, the epoch EEG data underwent Independent Component Analysis (ICA) using the FastICA algorithm (Version 2.5, October 19, 2005) [21]. This step aimed to decompose the EEG signal into a series of maximally independent components by using a multidimensional random vector as a linear blend of non-gaussian random variables. This decomposition aids in spotting and eliminating noise and artifacts from the primary EEG dataset. The ICA algorithm utilized a symmetric methodology with a hyperbolic tangent (‘tanh’) nonlinearity.

Upon retrieving the independent components, the EEG data were mapped to a predefined set of channels in the 10–20 system. Each channel’s position was represented using its spherical (theta, radius) and Cartesian (X, Y, Z) coordinates. Subsequently, the ADJUST algorithm was invoked to spot components tainted by artifacts [22]. ADJUST leverages statistical attributes and spatial configurations to detect and label components from common EEG disturbances like eye movements and other non-cortical events.

EEG and HRV wavelet analysis

EEG and HRV frequency domain analysis were conducted using the Morlet wavelet transform. Given the duration constraints of the sound stimulus, our analysis did not include the very low-frequency band of the HRV. Our focus remained on two primary frequency bands: the low-frequency band (LF: 0.04–0.15 Hz), indicative of baroreflex activity and influenced by both the sympathetic and parasympathetic nervous systems and the high-frequency band (HF: 0.15–0.4 Hz), which primarily reflects parasympathetic activation [23].

For EEG, four frequency bands were considered: delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), and Beta (12–30 Hz) [22, 24]. Although our equipment’s sampling rate could represent frequencies up to 100 Hz based on the Nyquist theorem, we chose not to consider the gamma band (30–80 Hz) to ensure our analysis’s robustness and minimize potential signal artifacts.

To quantify the coherence between the HRV and various EEG frequency bands, we determined EEG envelopes (EEGE) using a frequency-selective Hilbert transform of our EEG data. We ran wavelet coherence between the EEGE data and the HRV data. Each EEG frequency band employed a frequency band selective Hilbert transform, calculated using the Fourier transform, as outlined in Piper et al. (2014) [3], Eq. (1). By transforming the EEG signal, its amplitude envelope can be extracted and compared with the HRV signal for coherence analysis [3].

1

where EEGE_x(t) represents the Hilbert transform of the EEG data; F^− 1 the inverse Fourier transform, which converts a signal from the frequency domain back to the time domain; −𝑖 the negative imaginary unit, used to shift the phase of the Fourier-transformed data, 𝑠𝑖𝑔𝑛(𝑓) the signum function, which returns − 1 for negative values of 𝑓, 0 when 𝑓 is zero, and 1 for positive values of 𝑓 (It ensures that the negative frequencies are negated, and the positive frequencies are unaffected); BP(f) the band-pass filter operator (It selects specific frequency bands of interest from the EEG data); and F_x(f) the Fourier transform of the EEG data, which represents the EEG data in the frequency domain.

In simpler terms, Eq. (1) is a frequency-selective Hilbert transform. It selectively retains specific frequency bands from the EEG data, then shifts the phase, and finally transforms the data back to the time domain.

Morlet wavelet coherence was used to identify localized correlation coefficients in the wavelet time-frequency space between HRV and EEGE data. Wavelet coherence measures the correlation between the common frequencies of two signals, with values ranging between 0 and 1. We quantified HRV-EEGE coherence as the mean coherence in time of the wavelet coherence [25–27]. Mean coherences across EEG bands were determined using the two HRV frequency bands: LF: 0.04–0.15 Hz, and HF: 0.15– 0.4 Hz.

2

where R_n²(s,τ)^XY represents the squared wavelet coherence between signals X and Y for a given scale s and time shift τ; s the scale factor, which represents frequency dilation (a higher scale corresponds to a lower frequency) τ the time shift parameter; W^XY_(s,τ) the cross wavelet transform between signals X and Y (it captures the common power and phase between the two signals at each scale and time shift); ∣W^X_(s,τ)∣ the absolute value of the wavelet transform of signal X (representing its power); ∣W^Y_(s,τ)∣: The absolute value of the wavelet transform of signal Y; and S a is naturally to designed smoothing operator that has a similar footprint as the Morlet wavelet (it helps in making the coherence measure more robust and less affected by noise).

In simpler terms, the squared wavelet coherence measures the correlation of the two signals in the time-frequency space. A value closer to 1 indicates strong coherence, while a value closer to 0 signifies weak coherence.

Statistics and data presentation

For all statistical analyses, results from the three silence stages were averaged. To analyze study volunteers’ heart rates across conditions, a two-way mixed analysis of variance (ANOVA) (2 sexes x 3 conditions) was applied with sex as a between-individuals factor and repeated measures on conditions. For the HRV-EEGE coherence analysis, brain channels were categorized by their location into left and right sides, and data from each side were averaged. A four-way ANOVA (2 brain sides x 4 EEG frequency bands x 2 HRV frequency bands x 3 conditions) with repeated measures on all factors was then used. The statistical software IBM SPSS (version 20.0) was employed for these analyses (IBM Corp., Chicago, Illinois). ANOVA models with significant interactions were followed by suitable post hoc analysis. The Bonferroni adjustment corrected multiple t-test comparisons. All statistical tests were set with an alpha level of 0.05. Data are presented in the text as mean ± SD and in figures as mean ± standard error of the mean (SEM). Unless otherwise stated, only the significant main effects and interactions are provided. Based on the statistical analysis, HRV-EEGE coherence was visualized using a topographic plot considering HRV-EEGE coherence values and EEG channel positions [28].

Results

Our statistical analysis of participants’ heart rates across different conditions showed a significant main effect for sound (F_2,58 = 3.565; p = 0.035), no main effect for sex (F_1,29 = 2.283; p = 0.142), and no significant interaction (F_2,58 = 0.664; p = 0.422). Post hoc analysis revealed that heart rate was significantly lower during high and low-pitch sounds than silence (p < 0.002, Fig. 2).

Fig. 2.

Heart rate (in bpm) for three conditions: silence, low-pitch sound stimulus, and high-pitch sound stimulus. Statistical analyses showed that heart rates were notably lower during high and low-pitch sound periods than in silence (* p < 0.002)

In the heart-brain coherence analysis, we noted a significant three-way interaction among EEG bands, HRV bands, and conditions (F_6,180 = 2.150; p = 0.050). This interaction implies that irrespective of the brain hemisphere, no significant variances were observed across conditions and EEG bands for the HRV’s high-frequency range (Fig. 3a). Whereas, for the HRV’s low-frequency range (Fig. 3b), HRV-EEGE coherence was significantly lower in the Beta band during the high-pitch intervals compared to silence (p = 0.033) and low-pitch sound intervals (p = 0.048). Figure 4 vividly demonstrates these variations through a topographical representation of coherence values and EEG channel locations based on the 10–20 Jasper international system.

Fig. 3.

Wavelet coherence analysis of heart rate variability (HRV) low frequency (LF: 0.04–0.15 Hz; a) and high frequency (HF: 0.15–0.4 Hz; b) and electroencephalography envelopes (EEGE) of 4 EEG frequency bands (Delta, Theta, Alpha, Beta) across three auditory conditions: silence, low-pitch sound stimulus, and high-pitch sound stimulus. Statistical tests indicated no significant differences across conditions and EEG bands for the HRV-HF (a). Furthermore, analyses displayed that in the low-frequency band of HRV (b), the coherence between HRV and EEGE was substantially lower for the Beta band during the high-pitch sound compared to both silence and low-pitch sound (* p < 0.048)

Fig. 4.

Topographic representation of the wavelet coherence analysis between electroencephalography (EEG) and heart rate variability (HRV) for the low frequency (LF: 0.04–0.15 Hz) across the three conditions: silence (a), low-pitch sound stimulus (b), and high-pitch sound stimulus (c). The topographic plot integrates coherence values with EEG channel positions: F3, F4, F7, F8, Fp1, Fp2, T3, T4, T5, and T6, as per the 10–20 Jasper International System

Discussion

Our primary research objective examined the influence of high- and low-frequency sound stimuli on heart-brain coupling. By quantifying the wavelet coherence between heart rate variability and electroencephalogram envelopes from thirty-one study volunteers exposed to periods of silence or high (880 Hz) or low pitch (110 Hz) sounds, we discerned two crucial insights: both sound stimuli types were associated with reduced heart rates, and HRV-EEGE coherence was significantly lower for the high-pitch sound compared to silence and the low-pitch sound for the heart rate variability’s low-frequency band and EEG’s Beta band.

Our findings suggest that sustained high and low-pitch sounds might decrease heart rate relative to silence, potentially pointing to a calming or focusing influence of these sounds on the autonomic nervous system. The broader implications of these insights for real-world scenarios are considerable. While sound-induced heart rate reduction might be therapeutic, potentially aiding relaxation or anxiety management, it might also pose challenges. For instance, consistent exposure to specific frequencies during tasks demanding precision and concentration might adversely affect performance due to physiological changes. While few studies directly assess the effect of simple sound stimuli on heart rate, relevant research corroborates our observations. Specifically, findings from Nakajima et al. [29] and Veternik et al. (2008) [30] mirrored our observations, but a cause-effect relationship between sound and heart rate modifications necessitates more extensive research [31, 32].

As for the heart-brain coupling, we observed a notably diminished coherence between the low-frequency band of heart rate variability and the EEG’s Beta band for the high-pitch sound stimulus relative to both the silent and low-pitch stimuli. In our methodology, electrodes were chiefly positioned over the frontal and temporal scalp regions. The frontal lobes, corresponding to these placements, are instrumental in executive functions, decision-making, and emotional regulation [16, 33, 34], while temporal electrodes are crucial for auditory processing [18, 35]. Our data highlight a region-specific involvement contingent on auditory stimuli, consistent with prior findings that sound processing can modulate heart rate, respiratory rate, and blood pressure [35] and influence brainwave patterns [36].

The reduced coherence between the Beta band and LF HRV in studied brain areas warrants further contemplation. Beta brainwaves, typically oscillating between 12 and 30 Hz, correlate with active, analytical cognition, predominant during alertness, concentrated attention, and problem-solving. Conversely, HRV’s low-frequency band (0.04–0.15 Hz) signifies sympathetic and parasympathetic nervous system activities, reflecting baroreflex dynamics [37]. Hence, the observed coherence reduction during the high-pitch stimulus could indicate discord between cognitive alertness (as evidenced by Beta EEG patterns) and autonomic regulation (as represented by HRV-LF) under this auditory scenario, leading to asynchronous fluctuations in both attentional and autonomic states when juxtaposed against the high-pitch and silent conditions.

Our observations might hint at a distinct neurocardiac response to auditory stimuli [38], possibly implicating the frontal and temporal brain sectors in deciphering varied sound frequencies. This concurs with studies elucidating the continuous bidirectional communication between the heart and brain [4]. The salient coherence deviations, particularly in the Beta range, emphasize the need to comprehend the nuanced relationship between sound frequencies and their bearing on heart-brain interplay, in harmony with research indicating variable modulation of this interaction by different auditory stimuli [36].

Although literature regarding synchronization between EEG and HRV about sound stimuli is limited, a few studies have broached the profound heart-brain interplay during music exposure. Stuldreher et al. (2020) discerned that physiological synchrony across EEG, electrodermal activity, and heart rate can flag attention-critical temporal events [39]. Bernardi et al. (2005) [5] observed that even brief musical or specific sound exposure could instigate cardiovascular alterations in the T4 temporal region, likely linked to neurological and behavioral transitions. Similarly, Blood and Zatorre (2001) [40] cataloged discernible cerebral blood flow changes when participants listened to personally chosen music. Building upon these findings, Bigliassi et al. [41] postulated interconnected frameworks governing the modulation of specific heart rate components and distinct EEG patterns. They hypothesized that music might affect fatigue sensations via neural resynchronization, subsequently impacting exercise intensity sustainability. Importantly, the selection of sound frequency and its subsequent effects on heart-brain coherence has ramifications for auditory therapy, neurofeedback, and even binaural beats meditation. Such observations suggest that the coherence alterations are not simply serendipitous but might be leveraged for precise therapeutic or relaxation protocols.

Interestingly, HRV-EEGE coupling has been indicated in epilepsy [3] and sleep cycles [24]. Piper et al. documented significant coherence between an HRV low-frequency sub-band (0.08–0.12 Hz) and the EEG δ envelope (1.5-4 Hz), apparent in both preictal and immediate postictal seizure phases [3]. Jurysta et al., meanwhile, found coherence between normalized high-frequency HRV and all-band EEG power spectra [24]. These findings intimate that specific sound stimuli might play roles in either mitigating or inducing epileptic seizures or enhancing or degrading sleep quality.

However, establishing a linear cause-effect triad amongst sound, cerebral activity, and heart rate is intricate [42]. Asserting, for example, that heart rate invariably ‘ascends’ during silence or ‘descends’ upon sound exposure compared to a baseline remains premature. Various factors might influence these mechanisms, including sound stimuli attributes (like frequency, volume), individual physiological sound reactions, and the auditory presentation context. A sound frequency might activate cerebral regions or neural circuits that subsequently sway the autonomic nervous system, thereby adjusting heart rate. Conversely, post-sound heart rate alterations might adjust brain activity, rendering the EEG alterations we detect reactive. Another conceivable postulate posits that sounds might independently and directly act on the brain and heart rate without a unidirectional cause-effect dynamic. Alternatively, during silence, anticipatory factors, mindfulness, or latent cognitive activities might modulate heart rate.

Our research augments the growing literature probing the intricate relationship between auditory stimuli and neurocardiac dynamics. We comprehensively view auditory influences on neuro-cardiac intersections by assiduously investigating HRV and EEG coherence under varied sound stimuli through wavelet coherence frequency analysis. Nonetheless, additional studies are imperative to clarify these dynamics and validate or refute our discussion’s postulates and their potential real-world applications.

In our experiments, we deployed Bose QuietComfort 15 Acoustic Noise Cancelling Headphones for sound relay. Recognized for their superior noise-cancellation, it’s paramount to note that such a feature might subtly alter the frequency response characteristics, potentially influencing the integrity of the sound stimuli. Future investigations should consider utilizing headphones with meticulously documented response profiles and contemplate the potential ramifications of active noise-canceling attributes on their results.

In summation, our use of wavelet coherence frequency analysis between HRV and EEG envelopes showcases the capability of different sound frequencies to alter heart-brain synchronization. Fundamentally, our results emphasize that varied auditory stimuli can induce diverse modifications in this interplay. These revelations beckon additional research, especially given the possible therapeutic or adverse implications for various patients or scenarios. As biomedical engineering continues to evolve and integrate more deeply with neuroscience and cardiology, studies like ours highlight the promise and complexities inherent in understanding and potentially leveraging the nuanced interactions of the human body’s systems.

Author contributions

All authors contributed to the study conception and design. Data collection was performed by Camila Bomfim vonJakitsch. Material preparation was performed by Osmar Pinto Neto, Camila Bomfim vonJakitsch, Tatiana Okubo Rocha Pinho, and Rafael Pereira. The first draft of the manuscript was written by Osmar Pinto Neto, Rafael Pereira and Ovidiu Constantin Baltatu; all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The research conducted by authors Osmar Pinto Neto and Ovidiu Constantin Baltatu was supported by scholarships provided by the Anima Institute.

Declarations

Competing interests

The authors have no relevant financial or non-financial interests to disclose.