What Is fMRI Neurofeedback? History, How It Works and the Evidence

fMRI neurofeedback is a form of brain training in which you lie in an MRI scanner, watch a live display drawn from activity in one part of your brain, and practice mental strategies to move it. EEG neurofeedback reads electrical rhythms at the scalp; fMRI can follow specific regions deep inside the brain, at the cost of a scanner and a research team. This guide to fMRI neurofeedback is for anyone curious about how it works, where it came from and what studies honestly show.

What fMRI neurofeedback is and how it works

Neurofeedback rests on a simple idea: if you can see a signal from your own brain, you may be able to learn to change it, the way a mirror helps you correct your posture. Researchers describe it as operant learning, in which the feedback rewards whatever mental approach moves the signal in the intended direction. What sets the fMRI version apart is the measurement. Compared with EEG, functional MRI covers the whole brain and locates activity to within several millimeters.

A study usually begins by choosing a target region linked to the question being asked, such as an area involved in emotion, pain or attention. A 2020 systematic review of 146 studies found the most common design was practice at raising activity in a single region, shown as a continually updating thermometer. Most of those studies were run in healthy volunteers.

You are then given a goal, such as raising or lowering the thermometer, and you try different strategies, such as mental imagery, shifting your attention or a mindfulness-style approach. Learning varies a great deal. A 2021 analysis of 608 participants from 28 experiments found that a considerable share of people do not learn to control the signal, and that a practice run without feedback before training was linked to better performance.

Where it comes from

Neurofeedback as a research tool dates to the late 1960s. In 1970, D. P. Nowlis and Joe Kamiya published work on controlling the EEG alpha rhythm through sound feedback. The same year, M. Barry Sterman and colleagues reported in Science that cats could be conditioned to produce more of a sensorimotor rhythm, and in 1972 Sterman and L. Friar described reduced seizures in a woman with epilepsy after sensorimotor EEG feedback training.

In 2003, a team led by Nikolaus Weiskopf at the University of Tübingen, with co-authors including Niels Birbaumer and Rainer Goebel, described a system that fed local brain signals back to a person in the scanner with a delay of under two seconds. In 2005, a study led by R. Christopher deCharms, who was affiliated with the company Omneuron, reported that people could learn to control the rostral anterior cingulate cortex, with matching changes in how painful a heat stimulus felt.

How a session works, from blood flow to feedback

fMRI does not record brain cells firing. It measures the blood oxygen level dependent signal, usually shortened to BOLD, a change in local blood oxygen that is coupled to nearby neural activity.

In a typical setup the scanner captures a full image of the brain about every two seconds. Software processes each image as it arrives, correcting for head movement and slow signal drift, pulls out the signal from the target region and updates the display. Turbo-BrainVoyager is one program used for this step, and OpenNFT, published in 2017, is an open-source framework that supports feedback based on a region's activity, on connectivity between regions or on activity patterns.

Because the blood flow response is slow, the feedback trails your mental effort, and teams usually explain this delay first. A 2013 review described a common structure: blocks of 15 to 30 seconds spent trying to regulate the signal, alternating with rest, in runs of five to 15 minutes repeated two to five times in a session.

What people use it for

Because fMRI neurofeedback is mostly a research method, people usually meet it by joining a study. Researchers have explored it for depression and repetitive negative thinking, PTSD and anxiety, chronic pain, distressing voices in schizophrenia, ADHD, and alcohol use and craving. These are reasons it has been studied, not established outcomes. You can browse the conditions and symptoms hubs to see other approaches for the same concerns.

You may see claims that musicians, elite athletes or high-pressure professionals use fMRI neurofeedback to sharpen performance. Published research does not document this; most studies in healthy people test whether self-regulation can be learned at all. If you want neurofeedback you can book, EEG neurofeedback and other forms of biofeedback are far easier to find, with their own evidence to weigh. Any of these sits alongside usual care, not in place of it.

What the research says

fMRI neurofeedback has early, inconsistent evidence as a therapy: people often learn to shift the targeted signal, but well-controlled trials have not shown reliable symptom benefits. A 2018 critical review of 99 experiments in NeuroImage concluded that self-regulation seems viable but that matching behavioral changes are rarely replicated. A 2022 meta-analysis of 31 clinical trials found that only eight of 22 controlled trials showed significant improvement over control, with very high variation between studies.

The largest sham-controlled tests are negative. A 2022 double-blind trial in the American Journal of Psychiatry trained 88 boys with ADHD and found no difference from sham feedback on ADHD symptoms. A 2025 meta-analysis of PTSD trials found that the two fMRI trials with sham controls showed no improvement. For depression, a 2022 review of four randomized trials found a trend favoring training that was not statistically significant, and rated the evidence low certainty.

Smaller studies keep the question open. A 2023 sham-controlled trial in 39 adults with depression reported less brooding after active feedback, and a 2018 proof-of-concept study found 12 people with schizophrenia learned to lower activity in a hearing-related region.

Safety and who should check first

fMRI neurofeedback sends nothing into the brain and uses no ionizing radiation; the scanner only measures. A study of 114 people with chronic pain, run by the company developing the method, found no rise in adverse events during scanning or training. The issues most often seen are mental fatigue, physical discomfort, claustrophobia and loud scanner noise that can cause ringing in the ears.

The scanner's strong magnet means pacemakers, some neurostimulators, cochlear implants, some aneurysm clips and metal fragments can be unsafe, so every site screens first. Talk with the team if you are pregnant, have a seizure disorder or have a mental health condition that is not stable. Researchers also flag a theoretical risk of practicing unhelpful strategies, so ask how support works if the signal will not move.

Finding it and what to expect from availability

Most people find fMRI neurofeedback through university or hospital research studies, often listed on clinical trial registries. If a private provider offers it, ask about the protocol, supervision, the evidence for your goal and fees up front. Newer approaches are still experimental. Decoded neurofeedback, first reported in 2011, trains activity patterns without participants knowing the target. You can read the fMRI neurofeedback modality page, browse the Gyfts directory or explore related approaches.