For performance, heart-rate-variability and respiratory biofeedback remains the best-supported approach, and neurofeedback for sport and motor skills has caught up: three independent meta-analyses of randomised trials now find moderate to large effects on sport performance. Results for general cognitive enhancement in healthy adults are weaker.
Updated :
What the research shows
EEG neurofeedback — standard amplitude training
(SMR, alpha, alpha-theta, beta) — sport and motor performance
Ages
Adolescent (12-17) · Adult (18+)
Techniques
Sensorimotor rhythm (SMR), Alpha training, Alpha-theta, Beta training
AAPB3Probably efficaciousNeuroLogic4Efficacioushigher vs AAPB
Level 4, raised from the AAPB Level 3 — a single pooled rating covering every performance domain, which we disaggregate here. Randomised superiority on sport and motor outcomes now replicates across three independent pooled estimates: Yu 2025 (25 studies in qualitative synthesis, 21 meta-analysed; Hedges' g = 0.78, 95 % CI 0.49-1.07), with larger effects in studies above the CRED-nf median (g = 1.07 vs 0.49); Onagawa 2023 (16 RCTs, 374 participants; SMD 0.85, 0.18-1.51, with publication bias and substantial heterogeneity); de Brito 2022 (7 RCTs, 173 athletes; reaction time SMD -1.08, -1.90 to -0.25; decision-making SMD 1.12). Zhang 2025, a Bayesian meta-analysis of 41 RCTs in athletes, gives athletic performance SMD 0.88 (95 % CrI 0.69-1.05). The AAPB's base — Gong 2020 (SMR above alpha and control on shooting), Maszczyk 2018 (double-blind against sham, judoka balance), Christie 2020, Domingos 2021 — is reinforced by Chen 2025, a 4-week sham-controlled trial in 30 professional shooters with choice reaction time gains maintained at 4 weeks. Not 5: no sham superiority replicated in two independent settings. Adolescents by extrapolation: Carvalho 2026 (7 studies, 133 football players from youth to elite level) is the only source including minors, with small samples and limited randomisation.
Alpha, theta, SMR) — cognitive performance in healthy adults (attention, memory, executive function
Ages
Adolescent (12-17) · Adult (18+)
Techniques
Alpha training, Sensorimotor rhythm (SMR), Other neurofeedback
AAPB3Probably efficaciousNeuroLogic3Probably efficacioussame vs AAPB
Level 3 held. The AAPB Level 3 is a single pooled neurofeedback rating: the cognitive sub-literature is where it still belongs. Positive pooled estimates exist: Yeh 2022 (11 studies) reports theta neurofeedback effects on working memory (Hedges' g = 0.56, 0.10-1.02, I² = 62.9 %) and episodic memory (g = 0.62, 0.13-1.10, I² = 42.0 %); Yeh 2025, a network meta-analysis of 60 studies, ranks alpha training first for both. But the better-controlled subsets do not hold: Kimura 2024 pooled 15 RCTs (569 participants) for an overall attention effect of SMD 0.27 (0.10-0.44), with no significant pooled effect within the eight sham-controlled trials; Chiasson 2023 found only a minority of studies reporting statistically meaningful task effects, with no evidence of reporting bias. Rêgo 2022 (n = 80, nine sessions of SMR neurofeedback or tDCS, single-blind) found no between-group behavioural or ERP difference, while subjective improvement was reported in active and sham arms alike. Under the Parsons (2026) framework a sham-null is not a reason to lower — but it is not a reason to raise either. Adolescents by extrapolation.
Other neurofeedback methods
fNIRS / HEG neurofeedback
Ages
Adolescent (12-17) · Adult (18+)
Techniques
HEG / fNIRS
AAPB0Not ratedNeuroLogic3Probably efficacious
Level 3. Rated on the sweep alone: AAPB does not treat fNIRS. Zeng 2025 is the best-controlled trial in this indication: randomised, double-blind, prefrontal fNIRS-decoded neurofeedback (n = 20) against sham (n = 25), three daily 25-minute sessions, with the Stroop reaction-time effect reduced relative to sham at post-test (t = 3.056, p = 0.004) and at one-week follow-up (t = 2.180, p = 0.035); multivariate classification accuracy was higher in the trained group and correlated with the behavioural change (r = -0.36, p = 0.015). Kohl 2023, a bidirectional-regulation controlled study of the right temporoparietal junction (n = 50, four training days), found a significant group by time interaction on reorienting-of-attention reaction times. Matsuzaki 2023, a multilevel meta-analysis of three NIRS cognitive-training-with-neurofeedback studies (166 participants), reports benefit on episodic, long-term and working memory. Not 4: one randomised superiority trial over three sessions, no independent replication of the behavioural effect, and Matsuzaki's included studies used consumer NIRS devices.
Slow cortical potentials (SCP)
Ages
Adolescent (12-17) · Adult (18+)
Techniques
Slow cortical potentials (SCP)
AAPB0Not ratedNeuroLogic2Possibly efficacious
Level 2. Rated on the sweep alone: AAPB does not treat slow cortical potentials, so this sits among the other neurofeedback methods rather than as an amplitude-training sub-row. Lizama 2026 is an exploratory parallel-group randomised controlled trial in 42 amateur and semi-professional golfers (21 per group, mean age 43.0 years, SD 13.13) stratified by handicap, comparing an 8-week SCP programme with a wait-list: putting performance did not improve significantly, and only exploratory in-game heart-rate-variability and RMSSD measures favoured the trained group. Lizama 2025 describes the same cohort's electrophysiology after 16 sessions (2,560 trials): more positive SCP trials, higher beta peak frequency at C3 and O2, lower absolute power at F8 and T5, with no performance outcome. A randomised design with an adequately powered sample, but a null primary performance result and physiological findings only; the authors themselves flag the absence of an active sham-control condition.
Level 1. Rated on the sweep alone: AAPB does not treat infra-low frequency training. Schmidt 2024 is the only adequately powered test: double-blind, placebo-controlled, ten 30-minute sessions over five weeks in 42 healthy psychology students aged 18 to 35, with outcomes at post-test and two months — no significant differences between active and placebo groups on any dependent variable after correction for multiple comparisons (heart-rate variability, digit span, n-back, trail making, go/no-go, SCL-90-R, WHOQOL-Bref and a peak-performance questionnaire); the authors note a possible type-2 error. de Matos 2026, a randomised double-blind crossover mechanism study (39 analysed datasets, one session), found higher Beta1 global network efficiency during verum than sham, which did not survive Bonferroni correction across five bands. Under the Parsons (2026) framework a sham-null is not disqualifying, but there is no positive controlled performance finding here to rate on.
Biofeedback
Heart rate variability and respiratory biofeedback (sport, cognitive and academic performance, stress management)
AAPB4EfficaciousNeuroLogic4Efficacioussame vs AAPB
Level 4 maintained. AAPB rates heart rate variability and respiratory biofeedback separately from neurofeedback, and states that it found no data for any other biofeedback modality used on its own. The base is randomised and replicated across independent groups: Paul 2012 (N = 30, three arms — HRV biofeedback, no-training control and a motivational-video placebo; ten sessions improved physiological measures and basketball shooting more than both comparators), Choudhary 2016 (N = 24 track athletes, ten weeks; only the HRV arm improved VO2 max, heart rate variability and 5 km time), Dziembowska 2016 (N = 41 male athletes, randomised, reduced anxiety and increased self-esteem against a no-biofeedback control), Aritzeta 2017 (n = 152 vs n = 81; state anxiety Hedges' g = 2.48, trait anxiety g = 2.50) and Zafar 2017 (gamified respiratory feedback improving autonomic and cognitive measures under a Stroop stressor), with the systematic reviews of Jiménez Morgan 2017 and Pagaduan 2020. Zhang 2025 adds a pre-registered Bayesian meta-analysis of 41 randomised trials in athletes: athletic performance SMD 0.88 (95 % CrI 0.69-1.05), cognitive 0.81 (0.48-1.14), mental health 0.76 (0.44-1.09). Not 5: only Paul 2012 has a credible attention-placebo arm and no bona fide active comparator has been replicated. Adolescents by extrapolation.
EMG biofeedback — musicians and musculoskeletal performance
Ages
Adolescent (12-17) · Adult (18+)
Techniques
EMG biofeedback
AAPB0Not ratedNeuroLogic2Possibly efficacious
Level 2, provisional. Rated on the sweep alone. AAPB states that it found no data for any biofeedback modality other than heart rate variability and respiration, so this is an absence of assessment rather than a contrary finding. Olsen 2027 is a systematic review with meta-analysis of EMG biofeedback in musicians covering muscle activity, musculoskeletal health and performance; the archive holds the index row, but the full text has not been obtained and no abstract appears in the supplied sources. No effect estimate, sample size or description of the controls is therefore quoted here. The level is set at what one unread synthesis can support and should not be raised above 2 before the full text is read. To verify.
Multimodal biofeedback (EMG, EDA, thermal, heart rate and respiration, with or without neurofeedback)
Ages
Adolescent (12-17) · Adult (18+)
Techniques
Other biofeedback
AAPB0Not ratedNeuroLogic3Probably efficacious
Level 3. The AAPB's Level 4 applies explicitly to heart rate variability and respiration; multimodal work was not rated and is scored here on its own evidence. Christie 2020 stratified then randomised 19 ice-hockey athletes to 15 sessions of SMR neurofeedback combined with multimodal biofeedback or to no training: shooting performance improved in both groups, more so in the trained group. Zhang 2025 pools 41 randomised trials of biofeedback and neurofeedback in athletes, many of them multimodal protocols, with athletic performance at SMD 0.88 (95 % CrI 0.69-1.05) and cognitive performance at 0.81 (0.48-1.14), GRADE-appraised certainty and substantial heterogeneity. The AAPB's remaining records are perception studies: Dupee 2016 (N = 5 Olympic athletes) and Ferguson and Hall 2021 (N = 5, qualitative). The AAPB's own conclusion is that combining modalities may produce a more robust effect but that it is unclear which modality, or combination, is most efficacious. Not 4: one small randomised trial against no training, with no independent replication at a comparable protocol.
In short
Clinical reading
NeuroLogic rates HRV and respiratory biofeedback at Level 4, as the AAPB does. Neurofeedback is split: Level 4 for sport and motor performance, raised above the AAPB's pooled Level 3 on three independent randomised meta-analyses, and Level 3 held for cognitive performance in healthy adults, where the sham-controlled subsets show no pooled effect. Among the other methods: fNIRS at Level 3, slow cortical potentials at Level 2, infra-low frequency at Level 1.
Protocols
Resonance-frequency HRV embedded in mental preparation, with home practice; in neurofeedback, SMR, alpha or alpha-theta at central and parietal sites, 10 to 20 lab-based sessions with clinician-adjusted thresholds, protocol chosen by discipline.
Limits
Protocol inconsistency remains the field's major limitation — endless combinations of duration, site and band, small samples, few sham groups — and there is no clear dose-response rule. Nothing is known in children, and adolescent results are extrapolated from adults and university-age athletes. Performing arts and music have no controlled trial at all: the literature is limited to expert-versus-novice EEG comparisons and case reports, so no level is given. Rehabilitation of the injured athlete is explicitly outside the AAPB rating's scope and is not rated here.
Study base
31 publications from 2022-2026 filed in the archive for this indication, 29 kept after triage — 26 neurofeedback, 3 biofeedback. No new primary HRV-biofeedback trial appeared in the window: the Level 4 rests on 2012-2020 randomised trials, confirmed by recent syntheses.
Brendan's perspective
We raise sport and motor performance to 4 and hold cognitive performance at 3, where the AAPB gives a single pooled 3 for all of it. Disaggregating is the point. Three independent pooled estimates — Yu 2025, Onagawa 2023, de Brito 2022 — plus Zhang 2025 converge on moderate to large effects in athletes, and Yu 2025 finds the larger effects in the methodologically stronger studies, the opposite of the usual bias pattern. Cognitive enhancement in healthy adults does not survive the same scrutiny: Kimura 2024 finds no pooled effect within the sham-controlled trials. In practice I start with resonance-frequency HRV, a deployable skill that keeps its 4, then SMR or individual-alpha work over 10 to 20 supervised sessions, with transfer into the actual task rather than into a screen. This is an area I have worked in a great deal, and one of the few where I believe in both biofeedback and neurofeedback without much reservation. The difficulty is measurement, and it is not a small one. Defining a proper outcome in high performers is genuinely hard: on standardised tests these people sit at ceiling, so a null gain on the instrument is not a null gain in the person — it is an instrument that has run out of room. Anecdotally, the subjective gains reported by athletes and high performers are often substantial, and the gap between that and what the tests register is the real methodological problem in this indication. The honest limits: nobody controls coaching, practice time or opponents here, performing arts have no controlled trial at all, and the one adequately powered infra-low frequency study (Schmidt 2024) is null.
Jiménez Morgan & Molina Mora (2017) Effect of heart rate variability biofeedback on sport performance, a systematic review doi:10.1007/s10484-017-9364-2
Pagaduan et al. (2020) Can heart rate variability biofeedback improve athletic performance? A systematic review doi:10.2478/hukin-2020-0004
Gong et al. (2020) Efficacy, trainability, and neuroplasticity of SMR versus alpha rhythm neurofeedback doi:10.3389/fnhum.2020.00094
Yu et al. (2025) The Effect of EEG Neurofeedback Training on Sport Performance: A Systematic Review and Meta-Analysis doi:10.1111/sms.70055
Zhang et al. (2025) The effects of biofeedback training on athletes' mental health and performance: a systematic review and Bayesian meta-analysis doi:10.3389/fpsyg.2025.1662868
Zeng et al. (2025) Improving interference control without conflict exposure: prefrontal fNIRS-decoded neurofeedback training doi:10.1117/1.NPh.12.4.045009
Kimura et al. (2024) Efficacy of neurofeedback training for improving attentional performance in healthy adults: A systematic review and meta-analysis doi:10.1162/imag_a_00053
Schmidt et al. (2024) The potential of infra-low frequency neurofeedback training in peak performance: The first double-blinded placebo-controlled longitudinal study in healthy adults doi:10.1016/j.jpsychires.2024.04.035
