What Does Science Say About Neurofeedback?
Below are summaries of the current research related to EEG neurofeedback, including protocols used, findings and more, in seven major areas of brain health: ADHD, Peak Performance, Anxiety, PTSD, Addictive Disorders, Cognitive Decline, and Depression.
Attention-Deficit Hyperactivity Disorder (ADHD)
ADHD is the most well-studied condition in neurofeedback research. Based on meta-analyses and large multicenter randomized controlled trials (RCTs), two frequency neurofeedback protocols researched for more than 40 years have been shown efficacious and specific for ADHD: theta-beta ratio (TBR) and sensorimotor rhythm (SMR) (AAPB Guidelines; La Vaque et al., 2002). Frequency neurofeedback for ADHD received a grade 1 (‘‘best support’’) rating from the American Academy of Pediatrics in 2013.
TBR aims to decrease theta (4–7 Hz) and/or increase beta (12–21 Hz) power in central and frontal locations to reduce the high theta-beta ratios, high theta power, and/or low beta power characteristic of children and adults with ADHD.
Recent RCTs suggest that 30–40 sessions of TBR neurofeedback were as effective as methylphenidate in ameliorating inattentive and hyperactivity symptoms and were even associated with superior post-treatment academic performance (Duric et al., 2012; Meisel et al., 2013).
SMR over the sensorimotor strip (predominantly right-central) is based on the functional association of the sensorimotor rhythm with behavioral inhibition in ADHD. In seminal studies (Lubar & Shouse, 1976; Shouse & Lubar, 1979), it was demonstrated that the beneficial hyperactivity-reducing effects of combined SMR/theta training were maintained even after psychostimulants were withdrawn in hyperactive children. Studies suggest that TBR and SMR reduce inattentive and hyperactive/impulsive symptoms to a similar extent and after a comparable number of training sessions.
A series of meta-analyses have shown that the standard TBR and SMR protocols improve ADHD symptoms, especially inattention (Arns et al., 2009; Micolaud-Franchi et al., 2014; Bussalb et al., 2019; Riesco-Matías et al., 2019). Efficacy is clear for parentally-rated symptoms and less certain for teacher-rated symptoms (Micolaud-Franchi et al., 2014; Cortese et al., 2016; Razoki, 2018; Bussalb et al. 2019). However, parent ratings are associated with candidate gene pathways (Bralten et al., 2013), and teachers may be less sensitive to change (Cortese et al., 2016; Bussalb et al., 2019).
Using objective cognitive outcomes, a recent meta-analysis found neurofeedback to be more efficacious than cognitive training in ameliorating symptoms of inhibition (Lambez et al., 2020). Critically, a meta-analysis focusing on long-term maintenance found that after an average 6 months from completion of neurofeedback, the beneficial effects of neurofeedback were superior to semi-active control groups and methylphenidate (Van Doren et al., 2019). These findings demonstrate that whereas medication efficacy diminishes over time, neurofeedback efficacy increases.
The best evidence for efficacy comes from double-blind placebo-controlled RCTs, though it is challenging to devise a placebo condition that properly controls for psychosocial factors like perceptibility and motivation (Gaume et al., 2016). One of the largest and most comprehensive such trials is currently being carried out (International Collaborative ADHD Neurofeedback; ICAN; Arnold et al., 2013; 2018; 2019), with conclusive results anticipated soon.
EEG neurofeedback for ‘peak’ or ‘optimal’ performance focuses on facilitating brain performance in healthy individuals to achieve maximal brain functioning. Specifically, peak performance protocols aim to control level of arousal, attention and motivation, optimizing level of autonomic control and ability to shift states. A concrete goal of peak performance training is the completion of a specific function or task with fewer errors and greater efficiency, resulting in a more positive outcome (Vernon, 2005).
Twenty-three controlled studies have reported neurofeedback learning indices along with beneficial outcomes, including gains in: sustained attention, orienting and executive attention, the P300b event-related potential, memory, spatial rotation, reaction time, complex psychomotor skills, implicit procedural memory, recognition memory, perceptual binding, intelligence, mood and well-being (Gruzelier et al., 2014). Gains have been achieved by a variety of neurofeedback protocols, including: sensorimotor rhythm (SMR), beta and gamma, theta, and alpha power. Indeed peak performance surpasses other neurofeedback domains in that the majority of studies demonstrate evidence of learning.
Neurofeedback may optimize cognitive processing and learning by modifying white matter pathways and gray matter volume resulting in faster conduction velocity in neural networks. With regard to alpha power training, it has been suggested that engaging in a well-practiced task is associated with elevated alpha power, reflecting decreased cortical information processing and a more automatic stage of skill acquisition (Mirifar et al., 2017). In one study, increased SMR power improved accuracy and speed of surgery skills (Ros et al., 2009). In another study, inhibition of theta power reduced radar detection errors (Beatty et al., 1974).
Egner and Gruzelier (2004) reported faster reaction time in an attention task following an inhibit theta/enhance mid-beta protocol, and memory improvement has been reported following upper-alpha training (Escolano et al., 2011; Zoefel et al., 2011). A recent review found that 12 of 14 full studies reported positive effects in athletes, with 7 of 10 showing positive effects on performance, 3 of 6 studies reporting improved affective outcomes, and 3 of 3 reporting better cognitive outcomes (Mirifar et al., 2017). Though the evidence is overwhelmingly encouraging, sample sizes are small, and little is known about how methodological characteristics (e.g., number of training sessions, particular neurofeedback protocol) impact outcomes (Vernon et al., 2009; Mirifar et al., 2017). Thus larger, controlled studies are needed to address these issues and provide a clear understanding of the specific effects of neurofeedback on peak performance.
Alpha-theta (alpha, theta, alpha-theta enhancement) neurofeedback training, which reduces arousal, has been applied to reduce anxiety (as well as addiction) and create a generally relaxed state of well-being (Moore, 2000; Gruzelier, 2009). EEG neurofeedback offers an attractive option, as medication is only mildly more effective than placebo in treating anxiety disorders. Training is typically administered with eyes closed while listening to auditory feedback for a total of 7-12 hours of training.
As applied to generalized anxiety disorder (GAD), 9 of 10 neurofeedback studies reviewed by Moore (2000) and Hammond (2005a,b) produced positive changes in clinical outcome, with evidence for an anxiety reduction that endures even after 18 months (Watson et al., 1978). Indeed, for anxiety disorders, neurofeedback qualifies for the evidence-based designation of an efficacious treatment (Hammond, 2005a,b), with GAD and phobic anxiety disorder (as well as PTSD, summarized separately), demonstrating effects beyond placebo and meeting criteria for “probably efficacious” on the basis of American Psychological Association Clinical Psychology Division (Chambless & Hollon, 1998) and biofeedback specialty criteria (La Vaque et al., 2002). A recent systematic review of biofeedback in anxiety disorders (Tolin et al., 2020) reported a large advantage for EEG neurofeedback over wait list control groups, with higher quality studies showing superior effects; there was no clear benefit relative to active control groups, though few such studies were available to be included.
In a GAD study of high-talent musicians performing under stressful conditions, only musicians who received alpha-theta (enhancement) training yielded enhanced musical performance under stress (Egner & Gruzelier, 2003). In one RCT of test anxiety, neurofeedback participants generated 33% more alpha and showed a significant reduction in anxiety; by comparison, untreated participants and those receiving relaxation training experienced no significant symptom reduction (Garrett & Silver, 1976). A recent study in adolescents with self-reported attention and anxiety (e.g., thoughts of worry) symptoms found enhanced alpha and sensorimotor rhythm (SMR) along with improved symptoms (by visual analogue scales) after neurofeedback training of alpha, theta, and SMR twice a week for five weeks (Tsatali et al., 2019).
Post-Traumatic Stress Disorder (PTSD)
Evidence-based practice guidelines for PTSD recommend trauma-focused cognitive behavioral therapy (CBT) and eye movement desensitization and reprocessing (EMDR) as effective treatment modalities. However, the dropout rate for these therapies is high (Bisson et al., 2013; National Institute of Clinical Excellence (NICE), 2005). Pharmacological treatment (e.g., selective serotonin reuptake inhibitors; SSRIs) may also be effective, but the evidence is weaker. Further, treatment with pharmacological and psychotherapy-based therapies may last several years and are ineffectual for ~40% of patients (Bradley et al., 2005; NICE, 2005; Stein et al., 2006).
EEG neurofeedback is a non-pharmacologic alternative that meets “probably efficacious” criteria for PTSD (Hammond, 2005a,b; Reiter et al., 2016) on the basis of American Psychological Association Clinical Psychology Division (Chambless & Hollon, 1998) and biofeedback specialty criteria (La Vaque et al., 2002). A recent systematic review and meta-analysis pooled data across four randomized controlled trials (RCTs) in PTSD (n=123) and revealed a very large effect size (standard mean difference of -2.30; 95% CI: -4.37 to -0.24) for improvement in PTSD symptoms that exceeded effect sizes for internet-based cognitive therapy and meditation-related exercises (Steingrimsson et al., 2020). The studies consistently favored neurofeedback in terms of symptom severity and number of patients achieving remission. Specifically, PTSD symptoms were reduced by 34-66% in the neurofeedback group, but ranged from a reduction of 15% to an increase of 13% in the control groups (3 passive, 1 active). The one study with follow-up (van der Kolk et al., 2016) reported 46% symptom reduction posttreatment and 51% symptom reduction at 1-month follow-up (compared with reductions of 13% posttreatment and 14% at 1-month follow-up in controls). At 1-month follow-up, 58% (11/19) of neurofeedback patients achieved remission as compared with 11% (2/19) of controls. In one study (Noohi et al., 2017), neurofeedback significantly improved performance on cognitive tests of executive function. In another (Peniston & Kulkosky, 1991), all neurofeedback patients (14/14) reduced psychotropic medication use as compared with one patient (1/13) in the control group.Though the extant evidence is encouraging (see also reviews by Reiter et al. 2016; Panisch & Hai, 2018), additional controlled studies are desirable for greater confidence and clarity regarding the efficacy of neurofeedback in PTSD. Indeed small, heterogeneous samples and different study designs preclude specific recommendations for the optimal neurofeedback protocol. Enhance alpha/inhibit theta protocols are often used for PTSD (e.g., Pensiston & Kulkosky, 1991; Noohi et al., 2017), but there is considerable variation in the frequency bands trained (e.g., Pop-Jordanova & Zorcec, 2004 used SMR enhancement), session duration (e.g., Kluetsch et al., 2013: single session; Peniston & Kulkosky, 1991: 30 sessions), inter-session interval and duration of treatment. Also, only one RCT included an active control group (van der Kolk et al., 2016; standard treatment), and no studies have incorporated a sham control.
EEG neurofeedback has been applied to addictive disorders for over 30 years, demonstrating promising results in well-controlled intervention studies, good adherence, reduced addiction severity, and psychosocial benefits even in patients with severe substance abuse. Consequently, EEG neurofeedback has been classified as “probably efficacious” as an adjunctive treatment for substance abuse (AAPB Guidelines; La Vaque et al., 2002; Sokhadze et al., 2008).
Known as the Peniston protocol (or alpha-theta training), the classical neurofeedback protocol for addictive disorders was originally applied in the treatment of alcoholism (Peniston & Kulkosky, 1989; Peniston & Kulkosky, 1990). The Peniston protocol assesses EEG activity in an eyes-closed resting condition while clients aim to increase parietal alpha (8-12 Hz) and theta (4-7 Hz) associated with a relaxed state, reducing EEG hyperarousal and augmenting coping skills (Gruzelier, 2009). Due to commonalities between substance use and ADHD, the Peniston protocol was later supplemented with initial sessions that aim to enhance central sensorimotor rhythm (SMR; 12-15 Hz) as is done for ADHD. Called the Scott-Kaiser modification, this composite protocol has been efficacious in individuals with polydrug abuse and high levels of impulsivity (Scott et al., 1998; Scott et al., 2005); other ADHD-based protocols (e.g., enhance SMR, inhibit theta and high-beta; Fielenbach et al., 2019) have also been applied. Given variation in type, duration, and severity of substance use, a neurofeedback protocol personalized for the observed brain activity has been advocated (Sokhadze et al., 2008).
A recent review (Schmidt et al., 2017) identified 7 EEG neurofeedback clinical intervention trials in substance use since 2010, including 4 randomized controlled trials (RCTs). Disorders included misuse of: opiates (2 studies; Dehghani-Arani et al., 2010; Dehghani-Arani et al., 2013), stimulants like cocaine and methamphetamine (3 studies; Hashemian et al., 2015; Horrell et al., 2010; Rostami & Dehghani-Arani, 2015), alcohol (1 study; Lackner et al., 2015), and mixed substance and polydrugs (1 study; Keith et al., 2015). Sample sizes ranged from 10-100, and the number of neurofeedback sessions varied from 10-30. Neurofeedback protocols were mainly the Peniston protocol (some with adjustments; see also Dalkner et al., 2017) and Scott-Kaiser modification. In all studies, neurofeedback supplemented other interventions (e.g., pharmacotherapy, psychosocial like cognitive behavioral therapy [CBT]). Except for the alcohol dependence study, all studies reported positive addiction-related outcomes, especially reductions of addiction severity and craving. There were also global psychological and health improvements in most studies.
Two studies reported objective measures, showing substance use abstinence in a urine test (Horrell et al., 2010) and improved scores on neuropsychological tests of attention and impulsivity (Keith et al., 2015). Changes in baseline alpha and theta activity were found in alcohol dependence, as well as changes in the overall EEG, SMR and (reduced) gamma in opiate dependence. The one sham-controlled study revealed superiority of alpha-theta neurofeedback in clients with methamphetamine misuse compared with sham (Hashemian, 2015). Critically, one study showed the superiority of neurofeedback to psychotherapy, with equivalent efficacy for clinician- and computer-guided neurofeedback (Keith et al., 2015). In sum, recent studies show promising short-term effects of EEG neurofeedback in reducing craving and modifying dysfunctional brain activity. Additional RCTs are needed that aim to control for nonspecific effects by comparison with other psychophysiological treatments (e.g., electrodermal/HRV biofeedback); RCTs with long-term follow-up are needed to evaluate the occurrence of relapse.
Neurofeedback has been applied to improve cognitive function in a variety of conditions, most prominently attention-deficit hyperactivity disorder (ADHD), associated with impaired attention and executive function (see separate research summary). There is now an emerging body of research on neurofeedback for improving cognitive function in such conditions as stroke (Kober et al., 2015; 2017) and multiple sclerosis (Kober et al., 2019; Keune et al., 2019), with a particular focus on Alzheimer’s disease (AD), the most common form of dementia, as well as mild cognitive impairment (MCI), a pre-dementia condition (Petersen et al., 2004; Albert et al., 2011), in the hopes of delaying the insidious cognitive decline and dementia onset.
Memory impairment is the hallmark of early AD and its precursor amnestic MCI (aMCI); other cognitive domains may also be impaired. In the EEG, MCI and AD are generally characterized by an increase in slow frequencies (delta: 2-4 Hz; theta: 4-8 Hz) and a decrease in faster frequencies (alpha: 8-12 Hz; beta: 13-20 Hz) (Vigil & Tataryn, 2017). These EEG features have been linked to poor cognitive performance (Klimesch, 1999), atrophy of thalamus, hippocampus and basal ganglia (Moretti et al., 2012; Wolf et al., 2004), and the formation of amyloid-beta plaques (Sharma & Nadkarni, 2020). Notably, a smaller change in alpha between eyes-open and eyes-closed states has been tied to psychomotor and cognitive slowing, as well as memory impairment in MCI (Van der Hiele et al., 2007) and AD (Pritchard et al., 1991). Neurofeedback protocols in healthy and mildly impaired older adults have mainly targeted enhancing alpha, inhibiting theta, or increasing the alpha-theta ratio at posterior sites (e.g., Chapin & Russel-Chapin, 2014). Some have used attention training to enhance sensorimotor rhythm (SMR; low beta) or reduce theta-beta ratio (TBR) at central sites (Jiang et al., 2017; Jang et al., 2019), given that enhancing attention improves encoding, maintenance and retrieval of items held in working memory.
Several recent studies have reported better memory performance in MCI following neurofeedback. Lavy and colleagues (2019) found improved verbal memory after ten 30-minute sessions in which MCI participants enhanced individual central-parietal upper-alpha; improvement was maintained at 30-day follow-up. Jirayucharoensak and colleagues (2019) used alpha- and beta-enhancement neurofeedback (twenty 30-minute sessions) as an add-on to usual care in healthy or aMCI women and found improved rapid visual processing and spatial working memory. A small MCI study that enhanced beta over dorsolateral prefrontal cortex found improved memory, cognitive flexibility, complex attention, reaction time, and executive function (Jang et al., 2019). In AD, studies using individualized neurofeedback protocols have reported improved cognitive screener performance (Surmeli et al., 2016) and memory/executive function as compared with wait list control (Berman & Frederick, 2009). To summarize, initial evidence suggests that EEG neurofeedback is a promising methodology for timely, effective intervention for cognitive decline. Large-scale controlled trials with follow-up are needed to identify/validate the optimal protocols to delay MCI onset and conversion to dementia, as well as elucidate the relationship between neurofeedback and particular cognitive functions.
Neurofeedback for depression is based on well-established EEG research indicating that the left frontal area is more associated with positive affect, while the right frontal area is more involved with negative emotion (see, e.g., Davidson, Philos. Trans. R. Soc. Lond. B, 2004). A biologic predisposition for depression exists when there is an asymmetry in brain wave activity, such that there is excessive left frontal alpha (8-12 Hz) reflecting less activation and failure to suppress the subcortical structures that mediate depression (Walker et al., 2007). Indeed research has shown that when the left frontal region is “stuck” in an alpha idling rhythm, there is both reduced positive affect and more withdrawal behavior. Conversely, when there is increased left frontal beta (15-18 Hz), there is more activation and a greater sense of wellbeing. One neurofeedback protocol for modifying this suboptimal brain state involves modifying the left-right alpha balance at electrodes F3 and F4 (with a Cz reference). Research supports the efficacy of this ALAY (“alpha asymmetry”) protocol (Choi et al., 2011; Peeters et al., 2014), including evidence indicating changes in the asymmetry and depressive symptoms endure 1 and 5 years after the end of treatment (Baehr et al., 2001). In a recent study, major depressive disorder (MDD) most participants who received 1-hour/week ALAY intervention for 6 weeks regulated their asymmetry and showed improvement in depressive symptoms, though 43% were non-responders (Wang et al., 2016). Notably, although pharmacologic intervention yields remission of depression, it does not affect the frontal alpha asymmetry, suggesting that individuals who receive such intervention continue to have this biomarker for future depression.
Another neurofeedback protocol directly targets reducing left frontal alpha rather than modifying the left-right alpha balance (Walker et al., 2007). This protocol involves enhancing left-frontal beta (typically 15-18 Hz) and inhibiting left-frontal theta or alpha to yield greater activation, which, in turn, generally triggers improved mood. Studies have shown that enhancing beta and inhibiting theta or alpha at C3 reduced depressive symptoms in most patients (Walker et al., 2007). In a recent controlled trial, Liu (2017) applied an enhance beta/inhibit alpha protocol at F3 in 32 college students with MDD. In addition to regulating brainwaves, the neurofeedback intervention was protective, significantly reducing recurrence and intensity of depressive symptoms for 3 weeks post-intervention; in contrast, depressive symptoms increased in active control participants.