← Research library
ResearchWeak evidence

Digit span went up after 10 Hz beats, and there was no group that heard nothing

Mujib and colleagues (2021) · Experimental Brain Research · Summary by Sentiosonics research agent
Year
2021
Stimulus
Alpha
Design
Three parallel active arms, pre/post within each group, no control or sham arm
Sample
n = 60

Summary

This study is widely cited as evidence that binaural beats improve memory, and the thing it is missing is the thing that would make that reading possible. Sixty people took a digit span test, listened to fifteen minutes of beats, and took the test again. Those who heard 10 Hz beats scored better the second time. So might anyone who took the same test twice.

The design put 20 participants each into three groups receiving alpha (10 Hz), beta (14 Hz) or gamma (30 Hz) binaural beats for fifteen minutes, with EEG recorded before, during and after. Digit span, which asks you to hold and repeat back a sequence of digits, was tested immediately before and after the session. In the alpha group the mean score rose with a large effect size (p = 0.0000235, d = 1.15), and the time taken to finish fell (p = 0.0000157, d = 0.846). The gamma group also sped up, more modestly (p = 0.014). The beta group did not change on either measure. On the hardest items, the eight-digit sequences, only the alpha group improved on score (p = 0.0002) and time (p = 0.008).

The EEG results are the paper’s real contribution and the part the title is about. In the alpha group, theta power fell and alpha power rose, while theta imaginary coherence increased and gamma coherence decreased after stimulation. In the gamma group, both theta and gamma power rose along with theta and gamma coherence. The authors propose that the behavioural change runs through increased theta connectivity in parieto-frontal and interhemispheric frontal networks, and argue that because the beta group showed neither a cognitive nor a neurological effect, what they are seeing is frequency-specific rather than general arousal or attention.

That argument is the best available defence of the result, and it is not enough. A digit span test taken twice, twenty minutes apart, improves on its own; this is one of the most reliably demonstrated practice effects in cognitive testing. Without an arm that sat in the same room for the same fifteen minutes hearing silence or a plain tone, there is no way to know how much of the alpha group’s gain is the beat and how much is the second attempt. The beta group’s flat result constrains that explanation without removing it, since three groups of twenty can easily differ by chance on a measure this noisy.

We grade this weak, at a score of 2.9. On our rubric weak means exactly this situation, small samples or no control group, and here it is the control rather than the sample that does the damage. The EEG findings stand on their own; the behavioural claim in the title is not established by this design.

Limitations. No control, sham or no-intervention arm, which the authors acknowledge. Twenty people per group. A single fifteen-minute session, so nothing is known about repeated use. Digit span measures short-term verbal span rather than the everyday memory people are trying to improve, and it is highly susceptible to practice. The authors note a small number of EEG electrodes limited what they could say about deeper structures. The mechanism offered is plausible and correlational: the paper shows connectivity changed and scores changed in the same group, not that one caused the other.

Our analysis

How we graded this study

Each criterion is scored 1–5. Study design and replication count double in the overall score (how we grade). Graded from: full text read.

Study designUncontrolled pre/post within each of three groups. The only between-group comparisons are active frequency against active frequency, so no arm in the study answers whether a beat beats no beat. The authors name the problem themselves, calling it a relatively short duration of a single training without a separate control group.
Sample size60 participants in total, which falls in the rubric's 50-99 band, scored one point below it because the three groups are 20 each and every comparison in the paper is either within one group of 20 or between two groups of 20. The same documented departure as Ahmadi 2021 and Amd 2026.
Control conditionNone. There is no silence arm, no pure-tone arm, no sham and no no-intervention group. The beta (14 Hz) group functions loosely as a comparison because it did not improve, but it still received an active stimulus.
BlindingNo blinding reported. Outcomes are a computer-presented digit span test and EEG, both instrument-recorded rather than self-rated, which is why this is scored 2 rather than 1.
Outcome measureDigit span score and time to complete the task, plus EEG band power and imaginary coherence recorded before, during and after stimulation. All objective. The authors also tightened their significance threshold to p < 0.025 to account for multiple comparisons, which is better practice than most of this literature.
ReplicationSingle study, no replication found. It also sits against the nulls, and the authors themselves note that most published work on theta beat stimulation has reported no effect, citing Garcia-Argibay's meta-analysis.
Conflicts of interestFunded by the MoST Endowment at NED University of Engineering and Technology, Karachi, with support from the National Center of Artificial Intelligence. The corresponding author states on behalf of all authors that there is no conflict of interest.
Peer review & venuePeer-reviewed and PubMed-indexed (PMID 34232346), Springer.
Overall grade · 2.9 / 5Weak evidence

Graded by our research agent on Nov 16, 2026.

Get updates from the field

A monthly letter: new research, graded, and the claims worth checking. No hype.

Monthly, one click to unsubscribe. See our privacy policy.