Promises, Placebos, and New Directions: Neurofeedback for anxiety shows some promising results, placebo effects and potential for new discovery

Anxiety, according to the NIMH, affects at least 18.1% of American adults, 4% severely, and 60% more women than men. Although anxiety is an adaptive response to stress, prolonged anxiety can have devastating effects on health and life, and because of the cost, unpredictability and side effects of anxiety medication, patients and care-givers alike are seeking alternative treatments. Can EEG based neurofeedback help?

The studies thus far challenge us with many positive study results that are nevertheless contradicted by some high quality placebo-controlled studies

A wide range of target frequencies and protocols

Enhancement of the widely studied alpha activity in the EEG was one of the first studied neurofeedback applications (for an intro on neurofeedback see here). However, anxiety neurofeedback has since used a wide enough range of EEG target frequency-bands and protocols.  This includes frequencies in the alpha, beta and theta ranges, comprising almost half the typically measured spectrum of frequencies. Other less common but promising targets of neurofeedback for anxiety include alpha asymmetry in the frontal cortex, and pre-frontal and limbic ‘connectivity,’ perhaps treatable through fMRI feedback. This may reflect the many variations of anxiety, but these studies also point to aspects of neurodiversity, identifying aspects of individuality that affect treatment, including gender, cognitive traits (such as attentional bias), and confidence in treatment.

Types of Anxiety

The ambiguities surrounding anxiety neurofeedback probably have a fair bit to do with the diversity of elements involved. “Anxiety” like depression is diagnosed by questionnaires and can refer to Generalized Anxiety Disorder, Panic Disorder, OCD, PTSD, phobias, and some less common disorders—which could each be quite different. The most studied disorders are GAD and PD, but some interesting results in neurofeedback also pertain to OCD, PTSD, and phobias, as we’ll discuss below. Theory today suggests that anxiety involves deficits in cognitive skills, such as the control of attention and these cognitive aspects of the disorders are the most likely targets for EEG neurofeedback, whose effects are thought to be mediated most by cognitive skill enhancement.

Evidence of Success

According to Hammond (2005) and Moore (2000), seven of eight studies reviewed for Generalized Anxiety Disorder using alpha and theta enhancement showed positive changes in symptoms. Two controlled studies not included in Moore (2000) claimed to show long-term neurofeedback benefits in the treatment of 25 alcoholics with anxiety. Ten hours of training over three weeks produced significant (p<0.001) changes in anxiety and alpha production compared with controls, with identical results in the 18-month follow-up study; patients receiving real treatment increased alpha production from 33 to 58% while controls lowered alpha production slightly. A 2003 randomized, controlled, blinded study (Egner & Gruzelier in Hammond, 2005) aimed at enhancing musical performance among high-level musicians under stress. In comparison to a variety of alternative treatments (exercise, mental skills training, and SMR neurofeedback), only the alpha/theta enhancement group experienced improved performance.

Hammond also highlights three randomized controlled studies of phobic anxiety (Garrett & Silver, 1976) with alternative treatment control groups (relaxation training and EMG biofeedback) and an untreated control group. In one of these studies, alpha EEG enhancement led to 33% more alpha activity post-treatment, in another a change from 64 to 78% in alpha EEG; all such feedback groups showed significant drops (P< 0.001) in anxiety scores. None of the control groups showed significant reductions in test anxiety.

Two studies of PTSD got similar results. In one randomized controlled study (Peniston and Kulkosky, 1991), all fourteen of the veterans receiving traditional treatments relapsed after 30 months, compared to three of the fifteen receiving neurofeedback (about fifteen hours), all of whom decreased medication and showed significant improvements in anxiety scores, in contrast to the controls.

Frontal Alpha Asymmetry

Some studies seem to indicate that some forms of anxiety and negative affect (such as a aggression) correspond to asymmetrical alpha activity in the frontal lobes—the right more active than the left. In a randomized controlled test of neurofeedback for anxiety based on alpha asymmetry, reduction of right frontal alpha activity was correlated with a strong reduction in anxiety; BAI (anxiety test) scores dropped by approx. 50% for neurofeedback subjects compared to active controls (Menella et al. 2017).

Placebo-controlled Doubts

On the other hand, a small nonclinical sample showed a placebo effect for anxiety neurofeedback Plotkin and Rice (1981). Alpha enhancement and suppression feedback both produced a significant decrease in trait anxiety correlated with patient expectations. In a follow-up, however, Rice et al. (1993) showed that although perceived success plays a role, it does not account for the entire effect of feedback. All subjects still showed reduced anxiety six weeks after treatment but only alpha-increase subjects showed a reduction in heart-rate response to stressors at separate later test sessions.

One Randomized Control Trial that compared an anxiety-neurofeedback video-game (Mindlight) to a sham-neuroeedback game showed the same results for both groups (Schonevald et al., 2016). Similarly, a randomized double-blind placebo-controlled exploratory study of the “Low Energy Neurofeedback System” — currently promoted by many clinics — showed no difference in effect between groups (Gillham et al., 2012). Finally, in a study using SMR beta-training for anxious cardiac patients, with an un-treated control group, participants experienced reduced anxiety symptoms compared to controls but little change in this EEG metric, which has been interpreted as demonstrating a placebo-effect (Michael et al., 2005).

Future Directions

Despite contradictions between studies, it is safe to say that so far several quite different neurofeedback treatments can have some degree of positive long-lasting non-placebo benefit for the treatment of anxiety, while at the same time the placebo effect, or mere confidence in treatment, can play a significant role as well. And unlike drug treatment, there are no side effects.

That said, the constellation of results points to the need to look at changes in a wider set of EEG metrics that are perhaps missed by simply looking at broad frequency bands (see The Blue Frog in the EEG.). It also highlights the need to understand the placebo effect and the relevance of neurological and cognitive individuality to outcomes.