What Is Neurofeedback?

A Cat, a Scientist, and the Accident That Launched Brain Training

In 1965, a neuroscientist named Barry Sterman was doing something perfectly ordinary. He was studying sleep in cats at UCLA, recording their brainwaves with EEG electrodes to understand what happened in the feline brain as it drifted from waking to sleeping.

During these recordings, Sterman noticed something odd. When the cats were awake but very still, sitting motionless and alert, waiting for a food reward, their brains produced a distinctive rhythm over the sensorimotor cortex. A consistent 12-15 Hz oscillation. He called it the sensorimotor rhythm, or SMR.

Sterman wondered: could the cats learn to produce this rhythm on purpose? He set up a simple experiment. When a cat's brain produced SMR, a tone would play and the cat would get a food pellet. When it didn't, nothing happened.

The cats figured it out. Within a few sessions, they could voluntarily increase their SMR production. Their brains had learned to regulate their own electrical activity based on external feedback.

Interesting, but not world-changing. Then came the accident.

A few years later, NASA asked Sterman to study the toxic effects of rocket fuel (monomethylhydrazine) on the brain. He exposed cats to the substance, which was known to cause seizures. Most of the cats seized violently. But a handful of them resisted the seizures far better than the rest.

Sterman checked his records. The seizure-resistant cats were the same ones he'd trained to increase their SMR. The neurofeedback training had, without anyone intending it, made their brains more stable and more resistant to disruption.

That accidental discovery is the origin of every neurofeedback protocol in use today. A scientist training cats to control their brainwaves for food pellets stumbled onto the fact that the brain, when shown its own activity and given a reason to change, can learn to regulate itself.

Your Brain Is Always Talking. Neurofeedback Teaches You to Listen.

So what is neurofeedback, exactly? Strip away the jargon and it comes down to a single idea: showing your brain what it's doing so it can learn to do it better.

Here's the basic loop. Your brain constantly produces electrical activity. Billions of neurons firing in patterns that shift depending on whether you're focused, relaxed, anxious, drowsy, or deep in creative thought. An EEG device on your head picks up those electrical patterns through sensors on your scalp. A computer reads the EEG data in real-time, analyzes the patterns, and translates them into something you can perceive: a video that plays smoothly, a tone that gets louder, a game character that moves forward.

When your brain produces the pattern we want, the feedback is positive. The video plays. The tone sounds. The character advances. When your brain drifts away from that pattern, the feedback stops or reverses. The video pauses. The tone fades. The character stalls.

You're not consciously doing anything. You're not "thinking harder" or "trying to relax." You're just sitting there, experiencing the feedback, and your brain is doing what brains do best: learning from consequences.

This is operant conditioning. The same principle B.F. Skinner described in the 1930s, the same principle behind every time you've learned a skill through trial and reward. The only difference is that the "behavior" being conditioned isn't a movement or a choice. It's a brainwave pattern.

And here's the part that surprises most people: you don't need to understand what you're doing for it to work. Your brain figures it out below the level of conscious awareness. Just like you can't explain exactly how you balance on a bicycle, you can't explain exactly how you increase your SMR or reduce your theta-to-beta ratio. But given consistent feedback, your brain learns to do it anyway.

The Brainwave Alphabet: What Neurofeedback Is Actually Training

To understand the different types of neurofeedback, you need to know the basic language your brain speaks in. EEG picks up oscillations at different frequencies, and each frequency band is associated with different mental states.

These aren't separate systems. They're all happening simultaneously, layered on top of each other like instruments in an orchestra. What matters isn't any single frequency in isolation. It's the balance between them.

A brain producing too much theta relative to beta during a task that demands focus? That person will feel spacey, distractible, unable to sustain attention. Too much high beta during a time when calm is needed? That brain is running hot, anxious, unable to settle down.

Neurofeedback protocols target specific frequency ratios at specific brain locations to nudge that balance. Which brings us to the different flavors of neurofeedback.

Six Types of Neurofeedback (And What Each One Does)

Neurofeedback is not a single technique. It's a family of approaches, each targeting different aspects of brain activity. Here are the major protocols in clinical and research use today.

1. Frequency/Power Neurofeedback (The Classic)

This is the most common type and the direct descendant of Sterman's cat experiments. It trains the brain to increase or decrease the power (amplitude) of specific frequency bands at specific electrode locations.

The workhorse protocol for ADHD brain patterns, for example, trains the brain to increase beta activity (associated with focused attention) and decrease theta activity (associated with daydreaming) over the central and frontal cortex. The theta/beta ratio has become one of the most studied biomarkers in neurofeedback research.

2. Slow Cortical Potential (SCP) Neurofeedback

Instead of training fast oscillations, SCP neurofeedback targets very slow shifts in the brain's electrical potential, shifts that happen over seconds rather than milliseconds. These slow cortical potentials reflect the overall excitability of cortical networks. Learning to control them gives a person the ability to ramp up or ramp down their brain's readiness to respond.

SCP training has some of the strongest research support for ADHD. A landmark 2009 study by Ute Strehl and colleagues showed that SCP neurofeedback produced improvements in attention that were maintained at a six-month follow-up.

3. Low-Resolution Electromagnetic Tomography (LORETA) Neurofeedback

Most neurofeedback works with signals recorded at the scalp surface. LORETA neurofeedback uses mathematical models to estimate where in the three-dimensional brain the signals are coming from, and then trains activity at those deeper sources.

This allows clinicians to target specific brain regions rather than just scalp locations. If a brain map shows abnormal activity in the anterior cingulate cortex (involved in error monitoring and emotional regulation), LORETA neurofeedback can target that region specifically.

4. Infra-Low Frequency (ILF) Neurofeedback

This protocol trains oscillations below 0.1 Hz, frequencies so slow they were long considered noise or artifact rather than meaningful brain activity. Proponents argue these ultra-slow fluctuations reflect fundamental regulatory processes in the brain.

ILF neurofeedback is more controversial than other types. The evidence base is thinner, and the mechanisms are less well understood. But some clinicians report significant results with treatment-resistant cases, particularly for complex trauma and chronic pain.

5. Alpha-Theta Training

Originally developed for treating addiction and PTSD, alpha-theta training encourages the brain to enter a deeply relaxed state where alpha brainwaves (relaxed awareness) transition into theta brainwaves (the drowsy, hypnagogic state). This transition zone between waking and sleeping is associated with increased access to unconscious material and emotional processing.

During sessions, a person lies back with eyes closed, listening to two tones: one for alpha, one for theta. As the brain shifts from alpha dominance to theta dominance, the tones cross over. The goal is to maintain this crossover state without falling asleep.

6. Z-Score Neurofeedback

Instead of training toward a fixed target, z-score neurofeedback compares a person's brain activity in real-time against a normative database. It then trains toward "normal" values, essentially asking the brain: how far are you from the average? Can you move closer?

The appeal is that it's individualized by definition. The risk is that "normal" doesn't necessarily mean "optimal," and population averages may not capture the best state for any particular brain.

Brainwave data, captured at 256Hz across 8 channels, processed on-device. The Crown's open SDKs let developers build brain-responsive applications.

Explore the Crown

What the Evidence Actually Says (The Honest Version)

Here's where things get complicated, and where intellectual honesty matters more than enthusiasm. Neurofeedback has been around for over 50 years. In that time, the evidence has piled up unevenly. Some applications have strong support. Others are promising but premature. And some are frankly overhyped.

Let's go condition by condition.

ADHD: The Strongest Case

This is where the evidence is most convincing. The American Academy of Pediatrics rates EEG biofeedback (neurofeedback) as a Level 1, Best Support intervention for ADHD. That puts it in the same evidence category as medication.

A 2014 meta-analysis by Micoulaud-Franchi and colleagues, published in Journal of Clinical EEG and Neuroscience, pooled data from randomized controlled trials and found significant improvements in inattention and impulsivity. Crucially, several studies have shown that these improvements persist after training ends, some for more than six months. This is important because it suggests neurofeedback isn't just a temporary crutch. It's actually changing how the brain functions.

The theta/beta ratio is elevated in roughly 80% of children with ADHD. Neurofeedback protocols that normalize this ratio produce corresponding improvements in attention and behavioral control. The brain isn't being forced into a new pattern. It's being taught a pattern it can then maintain on its own.

Anxiety and Stress: Promising

Alpha and alpha-theta neurofeedback protocols show consistent positive results for anxiety reduction in controlled studies. A 2019 systematic review found that neurofeedback produced significant reductions in self-reported anxiety across multiple protocols. The mechanism makes sense: training increased alpha power in frontal regions promotes a state of relaxed alertness that is, almost by definition, incompatible with anxiety.

The caveat: many of these studies are small, and the field needs more large-scale randomized trials with active control conditions.

Depression: Active Research

Several studies have targeted frontal alpha asymmetry in depression, based on the well-established finding that people with depression often show greater right-frontal activation relative to left-frontal. Training to shift this balance leftward has shown promise in multiple studies, but the evidence isn't yet definitive.

A 2020 randomized controlled trial by Mehler and colleagues in Biological Psychiatry found that neurofeedback targeting frontal alpha asymmetry did not outperform a sham (placebo) condition for depression. This was a well-designed study and it tempered some of the earlier optimism. The field is now refining protocols and exploring whether other targets might work better.

PTSD: Emerging

A 2016 study by van der Kolk and colleagues in Frontiers in Psychology found that neurofeedback produced significant reductions in PTSD symptoms, with improvements comparable to the best available psychotherapies. This was particularly notable because the study included treatment-resistant patients who had not responded to other interventions.

But one study, however well-designed, isn't proof. More replication is needed.

Peak Performance: Intriguing

Outside the clinical world, neurofeedback has attracted attention from athletes, musicians, and executives looking to optimize cognitive performance. Studies on expert athletes have shown that neurofeedback can improve reaction time, reduce performance anxiety, and enhance the ability to enter flow states.

A study on Olympic athletes in Italy found that SMR neurofeedback improved performance accuracy and decision-making under pressure. Similar results have been reported with elite golfers, archers, and surgeons.

The peak performance application is where neurofeedback intersects most naturally with consumer interest. You don't need a diagnosis to want better focus, calmer composure under stress, or easier access to flow states.

Clinical vs. At-Home Neurofeedback: What Changed

For most of its history, neurofeedback required a clinic. You'd sit in a practitioner's office, hooked up to a medical-grade EEG system that cost $10,000 or more, while a trained technician monitored your session and adjusted protocols based on your brain map. Sessions ran 30-60 minutes, two to three times a week, for 20-40 sessions. At $100 to $200 per session, a full course of treatment could cost $4,000 to $8,000.

This model worked. It also locked neurofeedback behind a cost barrier that excluded most people.

Three things changed.

Consumer EEG hardware got dramatically better. Ten years ago, consumer EEG meant a single-channel headband that could barely distinguish between "brain on" and "brain off." Today, devices like the Neurosity Crown provide 8 channels of EEG data at 256Hz, covering frontal, central, and parietal regions. That's enough electrode coverage and sampling resolution to run meaningful neurofeedback protocols.

Processing moved to the edge. The Crown's N3 chipset runs signal processing on the device itself, not on a remote server. This means real-time feedback with minimal latency, which is critical for neurofeedback. Even a few hundred milliseconds of delay between a brain state change and the feedback signal degrades the conditioning effect. On-device processing keeps the loop tight.

Open SDKs made custom protocols possible. This is the change that matters most for the future of neurofeedback. The Neurosity Crown exposes raw EEG data, power spectral density, FFT frequency data, and computed metrics like focus and calm scores through JavaScript and Python SDKs. A developer, researcher, or tinkerer can build a neurofeedback application in an afternoon.

Think about what this means. The core neurofeedback loop, read brainwaves, analyze patterns, present feedback, reward desired states, no longer requires specialized clinical hardware. It requires a good EEG device, a computer, and code.

The Future: Where Neurofeedback Is Heading

The first generation of neurofeedback was clinical. A practitioner, a patient, a protocol, an office.

The second generation is personal. A device, an app, your living room.

The third generation, the one being built right now, is intelligent. And this is where things get genuinely interesting.

Closed-Loop AI Neurofeedback

Current neurofeedback protocols are relatively static. You pick a target (say, increase SMR at C3/C4), you set thresholds, and you train. The protocol doesn't adapt in real-time to what your brain is doing across sessions.

But combine real-time EEG data with AI, and something new becomes possible. An AI system that watches your brain activity over time, learns your individual patterns, identifies when you're drifting before you notice, and adjusts the training protocol dynamically.

The Neurosity Crown already integrates with AI tools through the Model Context Protocol (MCP). This means your brain data can flow directly into AI systems like Claude, which can analyze patterns, suggest protocol adjustments, and create feedback experiences that adapt to your unique neurology.

This isn't a distant-future scenario. The hardware exists. The data pipeline exists. The AI capabilities exist. What's being built right now is the software layer that connects them.

Neuroadaptive Environments

Take the neurofeedback loop and extend it beyond a training session. What if your environment responded to your brainwave data throughout the day?

The Neurosity Crown provides real-time EEG data and open SDKs, which means developers can build applications that respond to brainwave patterns. A text editor that dims notifications based on your brainwave state. A meditation app that adapts its guidance based on real-time EEG data rather than following a fixed script. An ambient audio system that shifts based on the brainwave patterns your Crown is capturing.

These are all potential neurofeedback applications that developers can build using the Crown's EEG data and SDKs. They're all closed loops between brain activity and environmental response. They just don't look like the clinical neurofeedback session of the past.

Open-Source Brain Training

Perhaps the most consequential shift is that neurofeedback is becoming hackable. With open SDKs, raw data access, and developer communities, the design of neurofeedback protocols is no longer confined to a small group of clinicians and researchers.

The Neurosity developer program and its JavaScript and Python SDKs mean that anyone who can write code can build a neurofeedback system. Want to train your brain to increase gamma coherence while you're writing? You can build that. Want to create a neurofeedback game that trains attention in children? The data is there. Want to combine neurofeedback with heart rate variability for a multi-modal biofeedback experience? The Crown's raw data streams make it possible.

This is how technology matures. It goes from laboratory to clinic to living room to platform. Neurofeedback is entering the platform phase. And platforms explode with innovation when developers get access to the building blocks.

What Barry Sterman's Cats Started

Let's zoom back out.

In 1965, a neuroscientist watched cats learn to control their own brainwaves for food pellets. A few years later, he discovered that this training had fundamentally changed how those cats' brains responded to stress.

Sixty years later, we have thousands of studies, millions of clinical sessions, and an increasingly clear picture of what neurofeedback can and can't do. The core principle hasn't changed since those cats: show the brain what it's doing, give it a reason to change, and it will change.

What has changed is access. You no longer need a clinic, a specialist, or a $10,000 EEG system. You need an 8-channel EEG device, a computer, and curiosity.

The brain has always been capable of self-regulation. It does it constantly, adjusting arousal levels, managing attention, modulating emotional responses. Neurofeedback doesn't give the brain a new ability. It gives the brain information it's never had before: a real-time view of its own activity. And with that information, the brain does what it's been doing for millions of years of evolution. It adapts.

Your brain is already training itself, every minute of every day, based on the feedback it receives from the world. Neurofeedback just makes that process visible, intentional, and precise.

The question isn't whether your brain can learn from its own electrical activity. Sterman's cats answered that in 1965. The question is what you'll teach it.