Oct 17 2007

Neurofeedback and the Need for Science-Based Medicine

I am frequently asked to give my opinion as a neurologist about neurofeedback techniques – these are therapies based upon using real-time recording of brain waves (EEG) to help train the brain to have a more normal brain wave pattern. These techniques are still not generally accepted, and with good reason, but are none-the-less growing in popularity.

This recent news article is a good indication of why the popularity of this treatment is outstripping the evidence. The article itself is a great example of the standard template journalists use for reporting controversial science stories. Here is the template:

– Begin with a touching anecdote about a person who believes they were helped by the treatment; if it’s a child that’s even better.

– Uncritically report the claims of proponents, emphasizing its growing popularity and the humanity of its practitioners.

– Sprinkle in more anecdotes about how wonderful the treatment is.

– Insert token skepticism (optional).

– Allow proponents to counter the skepticism, without giving the skeptics a chance to counter the proponents.

– Finish by returning to the touching anecdote. Closing sentences usually take the form of – “it may be controversial,” or “skeptics may scoff,” or “who knows if it works or not,” but “this person knows it works because it has saved their life.”

A reporter can follow this template for any story – no matter how worthless or unscientific the treatment. There can always be found anecdotes of people who swear by the efficacy of the treatment. Proponents always have an excuse as to why their treatments are not more generally accepted, and always have ad hoc reasons to dismiss skepticism. You can literally plug the details of any controversial treatment into this template.

What this means is that the story itself tells us absolutely nothing about the legitimacy of the treatment on which it is reporting. But a discerning reader can read between the lines and see the red flags of quackery. The story reveals most of the important ones. For example, the story reports:

The Association for Applied Psychophysiology and Biofeedback says it is most effective for ADHD, anxiety, headaches, hypertension, urinary incontinence, and temporomandibular (jaw) disorders, although it is used to treat conditions ranging from autism to post-traumatic stress disorder.

The more different ailments a treatment is supposedly good for, the more suspicious you should be of the claims for it. To be clear, this does not mean the the treatment necessarily does not work, it should just raise suspicion about the validity of the claims being made for it. This is especially true when the ailments are very different. Headaches and autism are very different types of disorders, and headaches are just a type of problem, not a specific disease or entity. Urinary incontinence is not usually even a brain disorder.

What this tells you is that it is likely that the claims are not rigorously evidence-based. When practitioners and proponents are not basing their therapies on adequate clinical evidence there is a tendency for “indication creep.” (I just made up that term, feel free to use it.) If they are basing their judgments about whether or not the treatment works on anecdotal experience, then the treatment is likely to seem to work for anything and everything, and therefore the alleged indications for the treatment will increase without bounds. The corollary to this is that if a treatment seems to work for everything, it probably works for nothing.

The article also quotes a proponent as saying:

“The biggest problem in our field,” says Kerson, whose organization is based in Richmond, Va., “is the lack of research money. The money is tied up with the pharmaceutical industry, and they are not interested in seeing us grow.”

This is an excuse for lack of research. We are also simultaneously told that the therapy has been used for decades, and that its use is growing, with numerous organizations dedicated to teaching and using it. Excuses for lack of research are often simply lame or imply a conspiracy or bias against the therapy. The reality is that if the treatment had promise, it would get researched. Neuroscientists would be interested in the potential for the therapy and what it tells us about the basic biology. Basic science and clinical science play off each other, and researchers are motivated to pursue fruitful lines of research. There is also a great deal of research money available from the NIH (in the US) and other funding organizations. And if the research were not done in the US then other countries would take the lead. The pharmaceutical industry controls only their own R&D money – they don’t control medical research. Also, the NIH does not fund pharmaceutical research, they feel the industry should pay for their own research, so you cannot argue that research money is being used up by big pharma.

The fact that the treatment has been used for decades is reported to give it the appearance of legitimacy, but actually this is a strike against it. If it has been used for that long, why then are there no credible double-blind studies to demonstrate its efficacy? Also, the fact that use of this modality is outstripping the evidence is just another red flag that its proponents are not adequately science based. They are putting the practice cart before the research horse.

The real reason the research has not been done is because those using the technique don’t have to do research. Drug companies spend millions of dollars on pharmaceutical research because they have to (by FDA regulations). If you don’t have to spend the money or time to do the research then you probably won’t – unless there is built into the profession a tradition of dedication to science-based practices. So what the lack of research is really telling us is that the regulations are lax and the practitioners lack a culture of science-based practice.

Look for use of this standard journalist template and for these red flags when evaluating any story about a new or controversial health claim. But what about neurofeedback itself? Does it work, despite that lack of scientific rigor by its proponents? Well, the lack of research means that we cannot say definitively, but I do not think the treatment is very plausible.

First, neurofeedback is based upon the notion that certain brainwave patterns are “normal” and others are “abnormal.” This is true, but not for the patterns neurofeedback proponents are talking about. Various disease states in the brain will result in abnormal EEG patterns – the whole brain can be slow, regions can be slow, or there can be epileptiform activity. But there is also a range of normal brain wave activity, mostly reflecting various states of attention vs relaxation. Neurofeedback distinguishes between various EEG patterns that neuroscientists would all deem normal and claims without a good basis that some are normal or healthy and others are not. They then use feedback to train patients to have a “normal” EEG pattern, when actually they are just using feedback to train patients either to relax or to focus their attention.

As far as that goes, neurofeedback can be basically a form of cognitive-behaviorial therapy. We can learn, and our brains to adapt and adjust (so called “plasticity”). So it is not surprising that if people practice relaxing they will get better at relaxing. This is the concept of biofeedback, and therefore I see neurofeedback as a form of biofeedback. But the claims made for it are couched in very different terms – not training the person how to relax or focus, but training the brain to have better brain waves. That is where the nonsense comes in. Treating brain waves is more sexy than just behavioral therapy; it makes it seem like something magical is happening.

I liken this to an athlete training for a sport by, say, shooting baskets over and over. The neurofeedback equivalent of this is to hook up electrodes to the athlete’s scalp and have them shoot baskets until their brainwaves look different and claiming that they are training the brain to be better at shooting baskets. Sure, they will get better at shooting baskets, but the brainwave angle is completely superfluous.

What we need are double-blind controlled trials that compare simple relaxation or cognitive-behavioral therapy without the neurofeedback to the same treatment with the addition of neurofeedback, to see if that element adds anything to the outcome. I suspect that it won’t, but I can be convinced otherwise by good evidence.

Neurofeedback proponents need to do some basic research to validate that the EEG patterns they are treating actually mean anything, and that their treatments add something measurable to the outcome. Until then they cannot justify continuing to use, let alone expanding, their treatments.

12 responses so far

12 thoughts on “Neurofeedback and the Need for Science-Based Medicine”

  1. Nevar says:

    This brings to mind those brain wave entrainment techniques using binaural beats or what not. I assume that listening to music (or rhythms) probably does create certain responses in the brain, but can one really entrain one’s brainwaves to a certain frequency using these methods? I’m rather sceptical towards the whole idea.

    I mean; if I need to concentrate on something, I’m sure many physiological changes start occuring including changes in brainwave activity. So how do we know their concentration enhancement CD does anything without proper tests? This all sounds like a do-nothing-but-still-get-the-benefit thing to me.

  2. Skeptyk says:

    On the AAPB’s homepage, they advertise a conference on Heart Rate Variability, another therapy for which popularity and hype outstrips evidence.

    That said, HRV and NF practices, AFAIK, are not as invasive as, say, “acu-“puncture, and have some elaborate ritual elements which can contribute to enhanced placebo effect. As Steven said, retraininng your own brain waves is sexier than behavioral therapy. “Behave yourself” vs “Rewire yourself”? If presented that way, who would not prefer the cybercool of NF?

  3. mas says:

    Not sure i agree with the ‘indication creep’ idea – seems a bit limiting.
    If new discoveries in psychoneuroimmunology are anything to go by, then things are a lot more complex than that!
    There is some evidence now, for instance, on the benefits (and possible underlying processes) of meditation. Does it’s credibility become reduced because it has the potential to impact on all bodily systems?

  4. The fact that some treatments have broad legitimate applications does not invalidate the notion that practitioners that are not evidence-based will tend to apply their treatments to an ever broader list of indications. The question is, how to tell the difference.

    One criterion is to consider the potential for the mechanism of action – can it reasonably be applied to a range of disorders? The other is to consider how much the target disorders have in common – do they have a common underlying contributor that can be the target of the common treatment?

    Meditation causes relaxation, so it could theoretically benefit any symptom or disorder that relaxation would help, or that is worsened by stress.

    The list of disorders for neurofeedback given above, however, do not share a common mechanism. Headaches, ADHD, urinary incontinence and hypertension? Not much of a common theme here.

    What they do have in common is that they are common problems for which people frequently seek treatment, many people may not be responding to conventional treatment, or may simply desire a non-drug solution. Similar lists crop up for most dubious treatments. It’s not a coincidence.

  5. Rgevirtz says:

    While I agree with your skepticism and actually teach my graduate students to adopt this kind of attiude, I think you need to look with a little more depth at the research that some of us are doing to determine not only efficacy but mechanism for the techniques mentioned. You mention HRV biofeedback and dismiss it as not having any empirical basis. A number of us have published several well controlled studies using heart rate variability feedback for disorders that have a likely autonomic pathway (IBS, Asthma, Headache, etc. ) I would call you attention to the study published in Chest by my colleage Paul Lehrer, Ph.D. showing efficacy for the procedure against a credible placebo condition. If interested, I can supply more citations of our work as well.

    Just because clinicians make irresponsible claims, it doesn’t mean that the baby needs to tossed with the bathwater.

    Lehrer, Vashillo, and Vashillo (2004) Biofeedback as a treatment for asthma, Chest, 126,352-361

  6. krissncleo says:

    The other day I asked about a company called Brain State Technologies that uses Brain State Conditioning-their “unique form of biofeedback-to treat troops returning home from the war. Said company claims that biofeedback can cure/heal everything from adhd, depression, migraines to anger management problems and seemingly, and host of other ailments. My question, does boifeedback work in these situations, was anwsered in your blog “Neurofeedback and the Need For Science Based Medicine.” Veterns and Families.org and The warrior Transition Project use anecdotal testimony and media pieces to promote their product. They claim that because neurofeedback has been around for decades and that it can cure a wide variety of ailments, but they do not say how measuring beta and alpha waves will do this. I’m still in college so I didn’t know if this was a new treatment or pseudoscience.

    There is a very nice lady that comes into the cafe where I work and she told me her son is going through the program so she asked me what I thought about it. Now I have something to tell her. For the sake of the veterns comming home I hope that their other treatments are based in science.

    Thank you so much.


  7. alec_s says:

    Actually there are some fairly rigorous studies involving treatment of ADHD with neurofeedback, for example:


    Although this was a clinical study, it looks as though they were reasonably strict in testing for and statistically controlling any systematic differences between groups. Their results look impressive, although I wish the article had also included some measure of statistical significance.

    However, many more studies are needed. An article I read:


    points out there as of yet there are no theories as to why neurofeedback actually works. It seems the field has a promising start, but needs more rigorous (non-case) studies to form a clearer picture of what’s going on.

  8. adriancmvd says:

    Hi there,

    I’m here to provide you with a complete and well organized resource that studies Neurofeedback.

    1- http://isnr.org/ComprehensiveBibliography.cfm
    Compiled by D. Corydon Hammond, PhD Professor, Physical Medicine & Rehabilitation University of Utah School of Medicine

    Frank H. Duffy, M.D., Professor and Pediatric Neurologist at Harvard Medical School, stated in an editorial in the January, 2000 issue of the journal Clinical Electroencephalography that the scholarly literature suggests that neurofeedback should play a major therapeutic role in many difficult areas. “In my opinion, if any medication had demonstrated such a wide spectrum of efficacy it would be universally accepted and widely used” (p. v). “It is a field to be taken seriously by all” (p. vii).

    My comment:
    The suggestion by Dr. Novella for a DOUBLE BLIND research design applied to this Behavioural learning procedure not adequate, because THE PERSON IN TRAINING PERCEIVES HIMSELF- there is no way to blind it.
    When false feedback is provided, a subject recognizes this and so the Blind unfolds.

    The AAPB, in it’s 40th year, publishes an efficacy guide that is quite skeptical and also critical. It is called Evidence Based Practice in Biofeedback and Neurofeedback. The 2008 edition is released but being sold at present. I can thus merely offer you the 2004 version.
    2- http://aapb.org/tl_files/AAPB/files/Yucha-Gilbert_EvidenceBased2004.pdf

    The results of these Efficacy ratings is possible because the resource used to train, THE EEG, is as exacting and specific as any hospital EEG. So when a person progresses in “real life”, the concurrent phenomenological manifestation can be tracked and studied.
    I ask you if there is a simple, cost effective and real-time accurate method to do say the same for pharmacological interventions…? Other than, of course, EEG readings.

    May this message be read along with the above to INFORM the reader.

    All the best, And Thanks to Steven,

    Adrian Machado Van Deusen

  9. r34498 says:

    It’s easy enough to double blind the “Brain State” neurofeedback, because the software essentially “records” the session in order to show “results” to unsuspecting clients. When trainers select the protocol to use (PP alpha up, or FF theta down, etc.) the study designer could randomly assign subjects to a treatment group or a control group and the program protocol called out for their session could simply be a previously recorded session of another person, rather than a so called live feedback session. The musical tones being played to the client would not be discernable as the “correct” “encouragers” or “discouragers”. The company claims that the brain will only change if it “sees itself”, so conveniently, clients are reassured, there is no way to harm a client by accidentally running a protocol with sensors that fell off or something. It’s brilliant marketing. I’d bet cash that responsive clients would report feeling the effects. The whole operation is classic placebo effect….warm blanket, cold bottled water, soothing ambiance and a person paid to listen to their problems for two hours at a time! The company employs a reputation monitoring company that electronically tails ex-employees and threatens to sue them if they report their experience to the Better Business Bureau or the state board of mental health examiners and FLOODS the search engines with positive “testimonials”. The prisoner “research” involved FOUR inmates, selected not at random, but by the warden who knew their temperaments (and gullibility). It’s a crime what they get away with.

  10. Onica says:

    Hi, I´m actually training as a neurotherapist at a psychiatric foundation. Im a recently graduated psychologist and conditioning seems to make sense at the very least for focusing attention and reducing anxiety or relaxing. However, I have not received satisfactory answers on a couple of instances and the foundation administrators seem to go on a tangent whenever I ask these seemingly inconvenient but in my view, essential questions. I wonder if anyone can help me figure this out, I would really appreciate it because Im starting to feel like santa doesn´t exist. The way neurofeedback is practiced where Im training goes as follows; We work with five different brain waves and around 20 different sites on the scalp where electrodes are placed. The electrodes read brainwave activity, that is all, and this is observed on a monitor. The patient engages in an activity where a brainwave which needs to be strengthened and two other brainwaves which have to be reduced, are represented on a monitor. The more you focus and the more relaxed you are the better your performance and the more effective the brain-exercise. The patient essentially feels their way toward the correct brainwave range because when they are in the right state, they start hearing a beep via earphones. Also, if the brainwaves are represented in a boat race video game, they focus so the boat in the middle wins the race, which is the wave being exercised. Now, the thing that baffles me and which I have tried to ask in about 10 different ways, is why do you need so many electrode sites when you are merely training yourself to be within a brainwave range and we are only working with 5 brainwave ranges (HB, Beta, SMR, Alfa, Theta). Lets say the SMR wave is too fast, arent you, by focusing and receiving the reinforcement, affecting the SMR wave present allover the brain? You cant direct your concentration to different brain regions, you´re just sitting there and focusing. Is this a way to benefit the electrode industry? Also, how does placing electrodes atop language areas affect language? The electrode doesnt stimulate and all you are doing is concentrating and trying to get in the correct wave range, you are not hearing new vocabulary words. There is another protocol which involves guided visualizations with your eyes shut. This makes more sense, as a therapy, and only one electrode is used on a central location. Please, clever people, I need answers!!

  11. wikibla says:

    I am very sorry, Mr. Novella, but I do not believe that you are sufficiently educated in neurofeedback to write about it.

    Quote, “The corollary to this is that if a treatment seems to work for everything, it probably works for nothing.”

    You are clumping all of neurofeedback into one small box and putting a stamp on it. Imagine how silly I would look if I took a bottle of Seroquel and said, “doctors prescribe drugs to cure just about anything, but I can’t cure my blood pressure with this drug, so all drugs must be a hoax.” By clumping neurofeedback into one small box, this is the error you are making. If I were to clump all of drugs into one box, then I might say “drugs work for nothing because they are prescribed for practically everything.”

    You must understand that there are many different protocols in neurofeedback that do different things, with different reward mechanisms, different frequencies, and different sites, and the sequencing of different treatments will have an effect as well. These different protocols are like different classes of drugs. There is much controversy as to which ones are effective and for what disorders, just like drug classes, but this fog will eventually clear away with the decades. I speculate that it will clear when it becomes more regulated.

    If you want a little secret, if you ever see anybody making the error of clumping up all of neurofeedback as a whole without reference to the different protocols, you can assume that they don’t know what they are talking about.

    The fact that so many graduated doctors are lacking of such fundamental knowledge is infuriating. I can only imagine how impressionable the the average person might be on the topic given the ignorance of the well-intended educated.

    Here is another error you made in your blog post:

    Quote, “If it has been used for that long, why then are there no credible double-blind studies to demonstrate its efficacy?”

    This one has two answers. Firstly, you are once again clumping up all of neurofeedback. Which type of neruofeedback are you referring to? But you are right, in all types of neurofeedback there are no double blind studies. This is because it is literally impossible to have one. All neurofeedback protocols, except for perhaps the very old, archaic ones, require the skill of a practitioner to actively shape and reshape the protocol after the initial session to fit the patient’s unique brain. If they are not doing this because of a blind then the treatment is much more likely to fail. This is not the case with pharmacology, where there is no need to “reshape” the molecule of the drug that is being administered.

    This is a good reason to be skeptical of studies on neurofeedback. Studies that try to exercise too much internal control may compromise the validity of the study by limiting a clinicians ability to be flexible with their treatment.

    Also, the difficulty of having a double blind study in neurofeedback is akin those of psychotherapy, where the skill of the therapist is vital to the outcome. This is just one of many examples of how, if you have a background in prescribing medicine, you have to forget about most of the things they taught you in order to learn about neurofeedback, which is a different beast entirely.

    I will at least admit that a double blind is possible with enough manipulation of the software to deceive both the clinician and the client. However, since neurofeedback takes, on average, 20 sessions before the effectiveness is fully realized, it would be extremely unethical to have a mentally ill patient waiting for nearly a half of a year for a treatment that isn’t being administered. The alternative would be to rush out the 20 sessions in a short time frame, but this is not viable for most protocols.

    I have a third criticism of your article. You said you mentioned that pharmaceutical companies spend millions of dollars to fund their research, and that this is mandated by the FDA. Would you believe me if I told you that the developers of most modern-day protocols have already done the same thing? (Perhaps the most thoroughly documented protocol is the Othmer method, with extensive records of clients and studies going back around 30 years.) But, of course, you may write them off as having a biased intention, which is a valid claim. However, just because they are not being regulated it does not mean their research is invalid. If you would like to look into their research you have to get your hands dirty and read the research articles yourself, which is a pain in the butt, but something we should be doing anyways.

    I would very much welcome a strong, authoritative confirmation from a governmental organization about the effectiveness of certain protocols, like how the drug companies have their FDA, but until then this is what we got. You can take it, or you can leave it, and I’m guessing that you still think it’s best to leave it. But you can bet that it’s not going away, and I’m betting that it will only get better over-time.

  12. wikibla says:

    Onica, I am aware that your comment is three years old, but I will still give a short answer.

    If you are questioning whether Santa exists then it may be best to go to the North Pole to find him. I am, of course, suggesting that you do neurofeedback on yourself (with supervision, of course.) You will know after about 10-20 sessions whether he exists.

    Keep in mind that this is not akin to taking a drug and seeing what happens. While the effects of caffeine are fairly predictable from person to person, the effects of, say, a T3-T4 bipolar montage at “x” frequency are entirely dependent on the particulars of the persons unique brain. Also, if you do not have an optimal setup, i.e. if you’re targeting the wrong frequency, then you fight feel no effect or negative effects, even after 20 sessions.

    This is why it’s important to have good training from experts. It sounds like this place you’re going to may not be so great. If you are still interested in Neurofeedback, I would do some research to find out who’s legit and who’s quack.

Leave a Reply