Testing Neurofeedback as a Treatment for Tourette’s

Q&A with Michelle Hampson, Ph.D.
Kayt Sukel
May 19, 2020

Michelle Hampson, Ph.D.
Associate Professor of Radiology and Biomedical Imaging
Director of Real-Time fMRI
Yale School of Medicine
Dana Grantee:

Tourette’s Syndrome is a neurodevelopmental disorder, diagnosed in children and adolescents, characterized by uncontrollable, repetitive movements or sounds, called tics. These movements might include frequent, irrepressible eye blinks, shoulder shrugs, throat clearing, grunts, or verbal outbursts. Such tics are not only socially disruptive but can be physically harmful to those diagnosed with this disorder. Unfortunately, traditional treatments for Tourette’s disorder, which include cognitive-behavioral therapy and certain antipsychotic medications, are not always effective – and can result in unpleasant, or even intolerable, side effects. The need for alternative therapies and interventions for a disorder that is more common than most realize is quite acute.

Michelle Hampson, Ph.D., an associate professor of radiology and biomedical imaging and the director of real-time functional Magnetic resonance imaging (fMRI) at the Yale School of Medicine, is interested in exploring the possibility that using neurofeedback from a brain region called the supplementary motor area (SMA), which previous studies have shown activates prior to a tic, may offer some relief, helping people diagnosed with Tourette’s Syndrome to better control their tics and reduce their frequency. Here, Hampson discusses why neurofeedback may be a good alternative treatment for this disorder and the challenges of developing the right training paradigms.

What first interested you in trying to use real-time fMRI feedback as an intervention for Tourette’s syndrome?

When I came to Yale, I discovered there was an active Tourette’s group at the Child Study Center. This condition seemed tractable from a research perspective in that it has a very clear symptom, the tic, and you know exactly when the symptom is occurring. In that sense, it’s easy to study with neuroimaging because you have a clear event-related marker to associate with brain activity. Being able to identify a symptom-related brain pattern is the first step in developing a neurofeedback intervention, so Tourette’s is a promising disorder in that respect. Also, compared to disorders like schizophrenia, for example, where people may have very different types of symptoms, there is a lot less heterogeneity. People have different types of tics, of course, but it seemed to me to be a much more solid diagnostic category and more likely that the brain circuitry involved is consistent across patients.

That said, it’s also been a challenge to study using neuroimaging. Many types of tics involve head movements, which can add a lot of noise to the imaging data.

I’d add that we also know that some people with Tourette’s are aware of a tic coming on before it occurs. It’s kind of like when you feel a cough coming on, but you are in a large auditorium and you don’t want to draw attention to yourself. So, you can suppress that cough, right? A subgroup of people with Tourette’s can do that. Not everyone, some people are completely unaware that a tic is coming on and feel like it is completely involuntary. But this subgroup is aware of a tic coming on. And these individuals feel the urge to make the movement and they report they have some ability to suppress or control the movement, at least in the short term. Like with that cough, you can prevent yourself from coughing for a certain amount of time but eventually that cough is going to happen. You are going to leave the auditorium to do it or disrupt whatever’s going on, but the cough will eventually come. It’s the same with the tics. However, the fact that some individuals do feel they can control the tics to some extent suggests that there are existing brain circuits that enable that control and that could be potentially strengthened via neurofeedback.

Why do you think it is worthwhile to develop a biofeedback intervention for this condition?

New treatments are needed – not everyone responds to the treatments that are available now. And even those that do often have troublesome residual symptoms or unbearable side effects. Our hope is that neurofeedback could help fill that gap.

This technique allows us to monitor a certain aspect of brain function using fMRI and give direct, real-time feedback back to patients, showing them how that aspect of brain function is changing. We cue them at certain times to try to increase or decrease that aspect of brain function. They get to practice controlling that aspect of brain function using the feedback we give them as a training signal. You can do this with any aspect of brain function as long as you know what aspect of brain function or activity is associated with the symptoms you are trying to control.

For example, in Tourette’s, we know that the supplementary motor area (SMA) is involved with tic behaviors. If you stimulate this area you can cause tic-like symptoms. And previous fMRI work shows that this area gets really active just before a tic movement occurs, too. That led to the idea that if we provide feedback about the SMA’s activity, we could train patients to control it, and it might result in an improvement in symptoms.

In a recent experiment, you trained study participants to both increase and decrease activity in the SMA. Why?

While it may seem to make the most sense to train them to decrease activity in the SMA, which seems like the most clinically useful direction, there’s also very interesting literature showing that when patients are very focused on a particular task, their tics will completely go away. You may know the story that Oliver Sacks told about the surgeon with Tourette’s syndrome who had tics but was able to do surgery because when he was focused on that task, the tics went away. The problem with that focus is that you can’t keep people hyper-focused all the time. They need to be able to relax and go to sleep. And being able to relax and sleep is something, like navigating social situations, that can be very difficult for people with Tourette’s. If they are a little bit stressed, they just can’t stop tic-ing and it is exhausting.

It’s sort of an odd situation. There is evidence to suggest that tics decrease when someone is very relaxed, happy, and well rested and seem to increase when people are stressed out. But the tics also decrease when they are very cognitively focused, and interestingly, cognitive focus often involves elevated SMA activity. So, in our most recent study, published in Biological Psychiatry, we tried to train the study participants to both increase and decrease activity in the SMA to see how it would affect their symptoms. We hoped, with training, there was some circuitry in the brain that we could strengthen that would allow them to better control activity in the SMA and reduce their tic urges.

What are some of the biggest challenges to designing a neurofeedback paradigm for this population?

There are many unknowns regarding how to optimize training to maximize the chances of getting clinical benefits. As you noted before, one choice we made was to train individuals to both upregulate and down regulate activity. If you upregulate by thinking of things that stress you out and make you want to tic a lot, like people making fun of you, that’s not likely to be therapeutic. So we talked with the participants about strategies that would be more helpful, like thinking about the sports they love and, if their tics tend to go away when they play those sports, imagining the movements they make and using that motor imagery to co-opt the brain circuitry involved with the tics. So, these are approaches to upregulating the SMA that we thought might have a potentially therapeutic aspect. The downregulating strategies were related to relaxing and also potentially therapeutic. In this way, we tried to maximize the therapeutic benefit of training in both the up and down directions.

I think, at the cellular level, there are probably different patterns of cells in the SMA that are active when you upregulate activity with these different strategies compared to the cells active prior to tics, but the hope was that engaging the SMA for these other purposes can suppress tics. So, it may be possible that certain types of hyperactivity in the SMA might reduce some of the tics. Unfortunately, it’s not possible see cellular level differences using fMRI – but this was our reasoning.

What did the patients think of the training?

Understanding the feedback is easy. You don’t have to understand what’s happening in the brain very deeply. We use a line graph for the feedback so, at certain points, they are trying to make the line go up and, at other times, they are trying to make the line go down. They don’t have to understand anything more complex than that. Just make the line go up or down – and they get direct feedback on whether or not they are succeeding.

That said, trying to control that line was very challenging for them. We knew that it would be before we began the study and told them they shouldn’t be discouraged – you are making changes even if you think you aren’t controlling all that well. Your brain may be picking up something even if you don’t feel like you’ve mastered control of this area.

I should add that our main measure for control, in the Biological Psychiatry paper, didn’t provide evidence that participants were actually learning effective control. It was sort of remarkable that we got the symptom change effects that we did given how noisy the learning patterns in the brain were.

How do you explain the fact that you saw sustained symptom changes?

We can’t rule out a possible placebo effect, or a learning effect that is not specific to the circuitry trained. The sham control was quite compelling, and the participants generally believed they were getting real neural feedback. But, on the other hand, an argument could be made that, at some level, during neurofeedback their brains saw a relationship between the feedback they were getting and what was happening in the brain. It’s possible that when there is a statistical relationship between the feedback and brain activity, your brain starts to rewire and build new synapses to adjust and learn. Maybe the sheer act of changing synapses in response to those statistical relationships is therapeutic. Because you don’t see the same statistical regularities in the sham treatment, that could explain why neurofeedback was associated with more symptom improvement than the sham condition.

It’s also possible that the neurofeedback did strengthen the circuitry allowing them to control their SMA, and that our measures of control over the region were just too noisy to capture that. The symptom changes are assessed across longer time frames and in a sense that provides a form of averaging that reduces noise. So, this could be a power issue, where we had better power to detect clinical changes than changes in control over the SMA. We discuss these different possibilities in detail in the Biological Psychiatry paper.

What is also interesting is that symptom changes seemed to persist and grow two weeks to a month after the neurofeedback was over. We first noticed this in another study where OCD participants who got real neurofeedback improved much more over time after the intervention, not just during the intervention. That was really surprising to us, and a similar pattern seemed to hold in the Tourette’s study. We published that finding in NeuroImage; it has some practical implications for neurofeedback studies that are discussed in that paper.

How do you plan to follow this work?

There are a lot of other conditions we are interested in studying but, in terms of Tourette’s Ssyndrome, we still have quite a bit of data to mine. We have resting state data, and we want to see which connections in the brain might have changed in response to the training. If we can identify certain functional connections that were strengthened in participants during the neurofeedback, and that are related to symptom changes, we can better understand why the participants showed the symptom improvements they did. This may help us identify other neural pathways that we should be targeting with our feedback.

That’s one of the nice things about neurofeedback: it’s kind of a closed loop in terms of basic science and clinical findings. It has an experimental medicine approach that allows us to constantly look at what’s happening in the brain and refine our targets as we learn more about what biomarkers are changing and how those changes relate to symptoms.

fMRI is, of course, quite expensive. And you’ve mentioned that there is a subset of patients, like those with tics that involve big head movements, who could not get feedback in the scanner. How do you see this type of therapy evolving?

There is a lot of interest in the neurofeedback world in eventually translating fMRI neurofeedback protocols to other modalities. There’s amazing work going on in Talma Hendler’s group in Israel which is identifying the electroencephalogram (EEG) fingerprints of fMRI activity patterns so they can train using only EEG. That’s the kind of long-term goal that the field is very interested in achieving.

Another option, in addition to the use of EEG, is functional near-infrared spectroscopy (fNIRS), which can measure blood flow activation on the outer surface of the brain. This is another promising option for measuring activity in the SMA. Developing the training paradigm using fMRI, which has good spatial resolution where you can target exactly where you want in the brain, is a good place to start. Then once you’ve found the right targets to train people to control, that result in symptom improvement, you can work on ways to translate those biomarkers to a different modality.


Time course of clinical change following neurofeedback. Rance M, Walsh C, Sukhodolsky DG, Pittman B, Qiu M, Kichuk SA, Wasylink S, Koller WN, Bloch M, Gruner P, Scheinost D, Pittenger C, Hampson M. Time course of clinical change following neurofeedback. NeuroImage 2018, 181:807-813. doi: /10.1016/j.neuroimage.2018.05.001

Orbitofrontal cortex neurofeedback produces lasting changes in contamination anxiety and resting-state connectivity. Scheinost D, Stoica T, Saksa J, Papademetris X, Constable RT, Pittenger C, Hampson M. Orbitofrontal cortex neurofeedback produces lasting changes in contamination anxiety and resting-state connectivity. Translational Psychiatry 2013, 3:e250. doi: 10.1038/tp.2013.24