Neurofeedback (EEG Biofeedback)

Efficacy (3.8/5.0): The strongest signal is ADHD. Arns 2009 meta-analysis found effect sizes of 0.81 for inattention and 0.40 for hyperactivity/impulsivity, which are clinically meaningful. Cortese 2016 in Lancet Psychiatry showed these effects became non-significant versus sham in probably-blinded assessments, but this finding has been meaningfully challenged: Arns and Strehl published detailed methodological critiques of Cortese's blinding classifications and inclusion criteria, and Van Doren 2019 demonstrated NF gains persist and even improve at 6-12 month follow-up, which is strong counter-evidence against a pure placebo explanation since placebo effects typically decay over time. The d=0.81 from Arns carries substantial weight given the durability data backing it. For epilepsy, Tan 2009 meta-analysis found 74% of patients achieved significant seizure reduction (AAN Level B: probably effective). Van der Kolk 2016 RCT for PTSD (n=52) showed 72.7% no longer meeting criteria versus 27.3% waitlist (d>0.9). In healthy adults seeking peak performance, effects are small: Enriquez-Geppert 2017 meta-analysis found d=0.23-0.30, though Gruzelier's music performance work showed 15-20% improvement. The overall picture: moderate-to-large effects for clinical conditions, small effects for optimization.

Breadth of Benefits (4.2/5.0): This is one of neurofeedback's genuine strengths. Studied applications span ADHD, epilepsy, PTSD, generalized anxiety (d=0.5-0.8 in pre-post designs), depression (fMRI-NF amygdala regulation: 33% vs 17% sham for >50% symptom reduction per Young 2017; EEG-NF uncontrolled d=0.5-0.7), sleep (SMR training increased spindle density and reduced onset latency ~30% per Hoedlmoser 2008 and Cortoos 2010), TBI rehabilitation (QEEG-guided protocols improved attention and processing speed), and athletic performance. Few interventions touch this many distinct neurological and psychological domains.

Evidence Quality (3.8/5.0): Multiple meta-analyses exist for ADHD and epilepsy. Cortese 2016 is the most-cited skeptical meta-analysis, but it has been substantively challenged by published methodological critiques from Arns and Strehl regarding its blinding classifications and study inclusion criteria, and Van Doren's follow-up durability data provides a strong empirical counter to the "just placebo" interpretation. The PTSD data from van der Kolk is impressive but relies on a single RCT with a waitlist (not sham) control. The sham debate remains a methodological challenge, though the specific EEG conditioning effect is likely larger than the d=0.2-0.4 range previously estimated when accounting for the critiques of Cortese's analysis. For anxiety and depression, most positive studies are pre-post without sham controls. fMRI neurofeedback adds rigor with sham-controlled designs but is a different modality than clinical EEG-NF. Overall: a large body of literature where the balance of evidence, including critiques of skeptical analyses, supports real efficacy beyond placebo.

Speed of Onset (2.2/5.0): This is slow. Most protocols require a minimum of 20 sessions before measurable changes emerge, with 30-40 sessions being typical and 40-60+ needed for complex conditions like chronic PTSD or TBI. Sessions run 30-60 minutes, 2-3 times per week. That translates to 2-5 months of consistent training before a meaningful assessment of response. Some users report subjective shifts (calmer, better focus) within 5-10 sessions, but objective EEG changes and sustained clinical improvement take longer.

Durability (4.2/5.0): This is neurofeedback's strongest selling point and the best argument against pure placebo. Van Doren 2019 showed ADHD gains maintained or improved at 6-12 months without ongoing treatment. Peniston's addiction work demonstrated 30-month maintenance of sobriety improvements. The theoretical basis is sound: neurofeedback drives structural neuroplastic changes (learned self-regulation skills), not transient chemical modulation. Unlike pharmacotherapy, which requires continuous dosing, completed neurofeedback training appears to produce lasting brain state changes. Some clinicians recommend periodic "booster" sessions, but the core gains persist.

Bioindividuality Upside (3.9/5.0): QEEG-guided protocol selection dramatically improves outcomes. ADHD patients with excess theta activity respond at 75-85% rates with targeted protocols, versus 40-50% with generic approaches. Athletes with previous TBIs appear especially responsive. The importance of proper assessment and protocol matching cannot be overstated. This means neurofeedback works very well when properly personalized, but the "average" response rate includes many people who received suboptimal protocols. Consumer devices, which use one-size-fits-all approaches, consistently underperform clinical NF. A 2025 JMIR meta-analysis concluded that consumer devices' ability to enable meaningful brain modulation is "not currently supported."