The Use of Neurofeedback as a Clinical Intervention: A Real-Time Solution

By Mollie Marko, PhD

Changing how a person thinks can be quite a challenge. Although seminars to improve gratitude, mindfulness meditation retreats and the like may be trendy, there is also real science to back up the idea that they may elevate mood,1 reduce stress2 and even help treat depression.3 With an estimated 4.45% of the US population suffering from depression,4 it is no surprise that these non-invasive, non-pharmaceutical interventions are growing in popularity. 

Recently, researchers have begun to develop a new method for the treatment of depression. The approach utilizes real-time neurofeedback, a feedback signal based on the person’s own brain activity in response to a given stimulus or task, which is recorded using functional magnetic resonance imaging (fMRI) scanning.  Researchers can use this neurofeedback to determine how a person is responding to a task and subsequently train them to respond differently.

The key aspect of this approach is that it occurs in real time: the participant’s brain activity is recorded and analyzed within seconds, allowing the researchers to modify the stimuli or task during the actual study. If the participant is doing well, (i.e., their brain activity matches what is desired by the researchers), then the participant can be rewarded (i.e., the task provides positive feedback), and vice-versa.

In this way, the process reinforces “good” brain behavior and punishes “bad” brain behavior. As Anne Mennen, a Princeton University graduate student currently studying real-time fMRI and depression, explained, “We are trying to change the brain by tapping into the brain itself, instead of relying on [the participant’s] own attention and participation in the task.”

Training with Neurofeedback

We have all been there. You are reading an article and suddenly you realize that you can’t remember anything about the paragraph you just read (hopefully not the case as you read this article!).  Your mind has wandered and you must consciously pull back your focus to the task at hand.

Is there anything we can do to improve our attention? In 2015, a Princeton University study by Megan deBettencourt and colleagues used real-time fMRI to examine healthy participants during an attention task.5 Participants were shown photos of a neutral face overlaid with a scene in a 50/50 ratio, and told to focus on either the face or scene.  At the same time, participants’ whole-brain activity was monitored using fMRI.

After some initial trials, a computer program, called a classifier, used machine learning to read the fMRI results and determine if the participant was focusing on the face or scene.  If, for example, the classifier saw evidence that the participant was focusing on the face when instructed to focus on the scene, the next photo would be more difficult: the face would be more pronounced, and the scene would be harder to see.  This has the effect of externally amplifying the (internal) lapse in attention, making it perceptually salient, and encouraging the participant to focus.

Conversely, if the classifier detected that the participant was attending to the scene as instructed, the scene in the next photo shown was made more visible.  This had the effect of making the task easier, rewarding the participant for having attended properly. On completion of the study, deBettencourt and colleagues were able to determine the parts of the brain involved in the task, as a traditional fMRI experiment would. Most importantly, they observed that neurofeedback training caused behavioral performance to improve.

Mennen and colleagues are now using this approach for patients with depression, who show a bias towards negative stimuli.6 For this study, rather than a neutral face and scene, the patients are shown a face with a negative expression overlaid with a scene, and told to focus on the scene.  In other words, the study is training patients to focus less on negative stimuli. 

Mennen explained, “The neurofeedback is posed as a window into [the patient’s] own attention.  So, if they're doing well, they're going to make the negative face go away.”  Although the study is ongoing, Mennen and colleagues hypothesize that over time, this training will lead patients to focus less on negative stimuli, and ultimately reduce global measures of depression.

A Computational Solution

Despite the promise of real-time fMRI research, the path forward has logistical constraints.  Stephen LaConte, Ph.D., an Associate Professor in the Department of Biomedical Engineering and Mechanics at Virginia Tech explained, “The number of groups working towards this is relatively small. One of the major hurdles that labs and fMRI facilities face is that setting up a real-time system is not turnkey. It requires an investment in time, effort, and expertise.”

Consider the computations required for any real-time fMRI study: in brief, the computer program analyzing the fMRI image needs to determine which parts are brain and which parts are background. It needs to smooth and filter the image to remove any noise or irrelevant signals. It must subtract the baseline brain activity.  It must perform statistical analyses on the resulting brain activity.  And it must do all of this in seconds, so the results can be accessed in real time.

This kind of computation requires large, expensive and high-powered machines, managed by experienced programmers.  For many research and medical institutions, the equipment and expertise necessary are simply not available.

However, Kenneth Norman, Ph.D., of Princeton University, has a potential solution. Through a collaboration between the Princeton Neuroscience Institute (PNI) and Intel, a team of researchers is creating an open-source, cloud-based software system for fMRI analyses.  By using cloud computing, the advanced computational power needed for these analyses is off-site, but accessible to all.  Ideally, anyone who wishes to complete a real-time fMRI study could log into the cloud through the internet, upload their fMRI image, have the cloud complete the analysis, and receive the result – all in a matter of seconds.

In fact, during a presentation this past March at the NIH, Norman delivered proof of concept by showing fMRI data being sent to the cloud from Princeton University to be analyzed and returned with feedback within seconds.  The results were displayed in real time on a laptop at the NIH, demonstrating just how accessible the cloud can be.

An additional component of the collaboration between PNI and Intel comes in the form of BrainIAK (, a “Brain Image Analysis Kit” that provides free, open-source software for all types of fMRI analyses. This enables anyone to download and access well established computational programs for real-time fMRI analyses.  Eventually, these programs will be accessible on the cloud, creating a complete and open-source toolkit for fMRI analyses.

LaConte, who is not involved in the project, is enthusiastic about the work by Norman and colleagues. “No one knows where the cloud computing efforts will lead,” he explained, “but there is certainly promise for increasing collaborative efforts around the world and decreasing equipment and expertise burdens to individual labs.”  Through their collaboration, PNI and Intel are making both the software and the hardware required for real-time and other fMRI analyses accessible to all.  This collaboration has the potential to open doors to novel research and invite creative minds to take part in this exciting new field.


1. Emmons RA, Stern R. Gratitude as a psychotherapeutic intervention. J Clin Psychol. 2013;69(8):846-855.

2. Goyal M, Singh S, Sibinga EM, et al. Meditation programs for psychological stress and well-being: a systematic review and meta-analysis. JAMA Intern Med. 2014;174(3):357-368.

3. Kuyken W, Hayes R, Barrett B, et al. Effectiveness and cost-effectiveness of mindfulness-based cognitive therapy compared with maintenance antidepressant treatment in the prevention of depressive relapse or recurrence (PREVENT): a randomised controlled trial. Lancet. 2015;386(9988):63-73.

4. Smith K. Mental health: a world of depression. Nature. 2014;515(7526):181.

5. deBettencourt MT, Cohen JD, Lee RF, Norman KA, Turk-Browne NB. Closed-loop training of attention with real-time brain imaging. Nat Neurosci. 2015;18(3):470-475.

6. Mogg K, Bradley BP. Attentional bias in generalized anxiety disorder versus depressive disorder. Cognit Ther Res. 2005;29(1):29-45.