Binaural beats: dichotic acoustics, perception and research evidence
Binaural beats arise when two slightly different frequencies are presented separately to the left and right ears. Because no physical amplitude modulation exists in either headphone channel, the wavering percept depends on phase comparison across the central auditory pathway.
Key points
- Binaural beats require presenting two slightly different frequencies separately to the left and right ears; stereo headphones are the standard practical way to keep the two channels from mixing in the air.
- Neither ear receives a physically pulsing waveform: unlike monaural beats, the wavering percept arises when the central auditory system compares phase differences between the two ears.
- Laboratory EEG and MEG recordings can detect auditory steady-state responses at the beat rate under specific protocols, while a systematic review of 14 EEG studies found evidence for broader brainwave entrainment inconclusive (Ingendoh et al., 2023).
- Trials on attention, sleep, and procedure-related anxiety report mixed, context-dependent results, often in small samples or with high heterogeneity across protocols.
1. How the sound works: Dichotic presentation
Binaural beats are an auditory phenomenon that can occur when two coherent pure tones of slightly different frequencies (f1 and f2) are presented dichotically, meaning one tone is delivered exclusively to the left ear and the other tone is delivered exclusively to the right ear (Oster, 1973; Moore, 2012).
Unlike monaural beats, where the two frequencies are physically summed before reaching the cochlea, a dichotic binaural stimulus contains no physical amplitude modulation in either headphone channel:
xL(t) = A·cos(2π·f1·t), xR(t) = A·cos(2π·f2·t), Δf = |f2 − f1|
Because each ear receives a steady, unmodulated sine wave of constant amplitude A, stereo headphones or earphones are required to keep the left and right acoustic channels isolated. If the same stereo signal is played over open-air loudspeakers, sound from both speakers reaches both ears in the room (acoustic crossfeed), partially mixing the two tones in the air and introducing a physical monaural beat component (see Waveforms and playback).
2. How it is perceived: Central binaural interaction and carrier limits
When f1 and f2 are close in frequency, the relative interaural phase difference Δφ(t) = 2π(f2 − f1)t rotates through a full 360° (2π radians) cycle Δf times per second. At low carrier frequencies, auditory nerve fibers preserve timing information via phase locking (synchronizing spike firing to a consistent phase of the basilar membrane waveform).
Phase-locked signals from the left and right cochlear nuclei converge along the ascending binaural auditory pathway, beginning in the brainstem superior olivary complex (notably the medial superior olive, which is sensitive to interaural time and phase differences) and continuing through the inferior colliculus to the auditory cortex (Moore, 2012; cortical beat responses localized by Draganova et al., 2008). Because the percept depends on continuous neural phase comparison across this distributed pathway rather than a single isolated switch, perceived binaural beats have three well-documented psychoacoustic characteristics:
- Shallow perceived modulation depth: Even under optimal laboratory conditions, the subjective fluctuation of a binaural beat matches only a shallow physical amplitude-modulation depth, although listeners can still discriminate beat-rate changes across low modulation rates similarly to salience-matched amplitude-modulated tones (Oster, 1973; Grose et al., 2012).
- Carrier frequency roll-off above ~1,000 to 1,500 Hz: Psychoacoustic experiments (including Licklider, Webster, & Hedlun, 1950; Perrott & Nelson, 1969; Oster, 1973) report that binaural beat perception is clearest at low-to-mid carrier frequencies (roughly 200 to 600 Hz) and declines progressively above roughly 1,000 to 1,500 Hz as fine-structure interaural phase sensitivity weakens, even though peripheral auditory-nerve phase locking in mammals degrades more gradually across higher frequencies (Moore, 2012).
- Individual and protocol variability: Not every listener perceives a clear beat at the same carrier or difference frequency, and masking noise, headphone seal, or attention to a competing task can reduce or abolish the subjective wavering.
3. What research shows: Separating brain signals from behavioral outcomes
Scientific studies on binaural beats investigate two very different questions that are frequently conflated in popular summaries:
- Neurophysiological tracking (laboratory response): Does the auditory brainstem or cortex produce a measurable electrical or magnetic response locked to the carrier or beat frequency during stimulation?
- Behavioral or clinical outcomes (uncertain clinical relevance): Does listening to binaural beats reliably improve attention, memory, sleep quality, or anxiety compared with silence, plain noise, music, or sham audio?
A. Subcortical and cortical tracking vs. whole-brain "state locking"
In laboratory EEG and MEG recordings, researchers distinguish between high-frequency frequency-following responses (FFR), which track the carrier pitch or fundamental frequency f0, and slower auditory steady-state responses (ASSR) or ongoing EEG band-power changes at the beat rate Δf:
- Cortical and subcortical contributions to FFR (Coffey et al., 2016): Using magnetoencephalography (MEG) with a speech syllable at a fundamental frequency of f0 = 98 Hz in 20 listeners, Coffey et al. demonstrated a right-asymmetric auditory cortical contribution alongside subcortical nuclei to the 98 Hz FFR. That study examined pitch-rate FFR to a 98 Hz fundamental frequency; it did not test binaural beats and does not establish 98 Hz as a universal upper limit for cortical phase locking across all stimuli.
- Smaller binaural ASSR compared with monaural beats (Schwarz & Taylor, 2005; Orozco Perez et al., 2020): Comparative EEG and MEG studies examining 40 Hz and theta-rate beats report that dichotic binaural beats evoke smaller cortical steady-state amplitudes than physical monaural beats (Schwarz & Taylor, 2005), while MEG source modeling localizes both central (binaural) and peripheral (monaural) beat responses to auditory cortex (Draganova et al., 2008). In Orozco Perez et al. (2020) (N = 16, comparing 7 Hz theta and 40 Hz gamma beats around ~400 Hz carriers), both beat types elicited subcortical carrier-frequency FFRs and cortical beat-rate ASSRs (with larger ASSR amplitudes for monaural beats in the tested conditions), while functional connectivity patterns differed with beat frequency and stimulus type without supporting broad state entrainment; no statistically significant change in self-reported mood on the Profile of Mood States (POMS) was detected, and behavioral vigilance was not tested.
- Systematic review of EEG entrainment (Ingendoh et al., 2023): A systematic review of 14 peer-reviewed EEG binaural-beat studies concluded that evidence for reliable brainwave entrainment remains heterogeneous and inconclusive: 5 studies reported results consistent with entrainment, 8 reported negative/contradictory results, and 1 reported mixed results, with widespread differences in carrier frequencies, exposure durations, and EEG analysis pipelines (see also López-Caballero & Escera, 2017; Jirakittayakorn & Wongsawat, 2017).
B. Meta-analyses and narrative reviews on cognition, anxiety, and mood
- Garcia-Argibay et al. (2019) (meta-analysis, 22 studies, 35 effect sizes): Reported a modest pooled overall effect across memory, attention, anxiety, and pain (g = 0.45), while emphasizing substantial between-study heterogeneity and noting that exposure duration and masking conditions varied widely.
- Basu & Banerjee (2023) (meta-analysis on cognition, 15 studies, 31 effect sizes): Found mixed, task- and frequency-dependent effects across attention and memory domains (overall g = 0.40) with high heterogeneity across protocols.
- Bășeanu et al. (2024) (systematic review with narrative synthesis, 12 studies, N = 1,349): Synthesized 12 studies (10 procedural-anxiety studies, 1 anxiety study in Parkinson's disease, and 1 depression study, without a pooled meta-analysis), noting short-term state-anxiety reductions primarily vs. passive care while several trials showed no significant advantage over ordinary music.
C. Recent experimental trials on fluid intelligence, attention, and framing
- Expectancy and home cognitive testing (Klichowski et al., 2023): Across two two-part home experiments totaling 1,000 tested participants (920 analyzed: 369 in Study 1 and 551 in Study 2, where each participant completed a single two-part session on either Raven's Progressive Matrices [RPM] or the Matrix Reasoning Item Bank [MaRs-IB]), Study 1 tested 15 Hz binaural beats after participants self-selected between Neutral Sounds and Stimulating Sounds ("sounds enhancing brain activity") descriptions, while Study 2 randomized participants across silence, classical music, a 240 Hz pure carrier tone, 5 Hz binaural beats, and 15 Hz binaural beats under undefined "some sounds" instructions. Home binaural-beat listening did not improve cognitive scores; instead, within-condition fluid-reasoning accuracy declined significantly from part 1 to part 2 across the binaural-beat sessions regardless of expectancy framing or beat frequency, whereas the silence, classical music, and 240 Hz pure-tone control groups showed no statistically significant pre-to-post score change.
- Parametric EEG and sustained-attention factorial trial (Melnichuk et al., 2025): Tested whether beat rate (16 Hz beta vs. 40 Hz gamma), carrier frequency (340 Hz vs. 400 Hz), stimulation onset timing, and background white noise altered EEG tracking or sustained attention against a pure-tone control across 64 analyzed adults in a 16-cell factorial design. Separate analyses reported higher average target detection on the Identical Pairs Continuous Performance Task for 40 Hz beats paired with the 340 Hz carrier and when background white noise was present, whereas overall vigilance decrement across task blocks was not attenuated. In this experiment, background white noise attenuated the measured beat-frequency EEG response, particularly for gamma beats. Because each between-subject cell contained only 2 to 6 analyzed participants and the behavioral effects emerged in separate interaction and subgroup analyses, these results do not establish a single optimal parameter combination (comparison in What Research Shows; reported interactions and subgroup comparisons in References).
D. Sleep and preoperative / acute procedural anxiety
- Overnight and nap laboratory sleep-staging studies (Jirakittayakorn & Wongsawat, 2018; Fan et al., 2024): In a 3-night laboratory study using EEG, EOG, and EMG for sleep staging (Night 1 adaptation, Night 2 baseline, Night 3 experimental; N = 24 enrolled, 23 analyzed), Jirakittayakorn & Wongsawat (2018) initiated a 3 Hz delta binaural beat (250 Hz carrier) at the first epoch of Stage N2 sleep and stopped it at the first epoch of Stage N3. Compared with baseline Night 2 and the unstimulated control group, the experimental night showed longer Stage N3 duration, shorter N3 latency, and higher whole-night sleep delta-band EEG power (within-group p = 0.0293; between-group change vs. control p = 0.0261), alongside elevated Stage N2 delta spectra whose increase correlated with N3 latency reduction. In a 4-condition afternoon nap polysomnography crossover trial in N = 12 pre-screened adults who could discriminate the beat stimuli, Fan et al. (2024) compared sham (silence) and 0, 0.25, or 1 Hz binaural beats on a 250 Hz carrier, reporting shorter N2 latency (p = 0.023) and N3 latency (p = 0.045) at 0.25 Hz versus sham (uncorrected Wilcoxon comparisons across sleep endpoints), without significant neural entrainment or changes in N3 duration or delta/sigma power. The study disclosed industry funding, two co-authors employed by the funding company, and a joint patent application (sponsor details in References). Both studies used small laboratory cohorts and stage-monitored or pre-screened protocols that differ from unmonitored home playback.
- Home evening pre-sleep trial (Shumov et al., 2023): In a counterbalanced home study (221 enrolled / 202 analyzed) where each participant completed three evening bedtime attempts (21:00 to 24:00) listening to monotonous background sound with binaural beats, monaural beats, or sham audio without beats while performing a smartphone behavioral psychomotor response-omission test, sleep-onset latency did not differ significantly across the three conditions (repeated-measures ANOVA p = 0.21; pairwise adjusted p = 1.0).
- Preoperative and dental anxiety (Xiong et al., 2025; Fatima et al., 2026; O. Akkurt & Ekici, 2026): In a systematic review and meta-analysis of 15 surgical RCTs, Xiong et al. (2025) reported that binaural beats reduced perioperative anxiety compared with blank audio controls across 14 trials (n = 1,047, SMD = −1.38, 95% CI −1.89 to −0.87, I2 = 91.6%) and vs. non-binaural audio across 8 trials (n = 598, SMD = −1.01, 95% CI −1.99 to −0.03, I2 = 95.4%), and reduced postoperative pain vs. blank controls across 5 trials (n = 433, SMD = −0.61, I2 = 75.8%) and vs. non-binaural audio across 3 trials (n = 265, SMD = −0.52, I2 = 0%), with extreme heterogeneity across anxiety endpoints. In a review of 13 adult theta (4 to 8 Hz) binaural-beat RCTs (n = 630), Fatima et al. (2026) found that quantitative meta-analysis was feasible only for pain (4 RCTs, n = 173, SMD = −0.53, 95% CI −0.84 to −0.23, p = 0.001, I2 = 0%, GRADE low certainty), while anxiety and cognitive outcomes could only be synthesized narratively due to methodological heterogeneity. Conversely, in a three-arm randomized clinical trial during impacted lower third-molar surgery (N = 90, 30 per group: binaural beats, isochronic tones, and no-audio control), O. Akkurt & Ekici (2026) found no statistically significant between-group differences in preoperative or postoperative anxiety across MDAS, DFS, STAI-S, or VAS scales; isolated physiological differences appeared only at single intraoperative timepoints (lower mean systolic blood pressure during incision in the binaural-beat group, p = 0.041; lower mean respiratory rate during local anesthesia in the isochronic-tone group, p = 0.015), although 78.3% of audio-group patients indicated they would prefer audio stimulation in future dental visits.