This article comes from one of my Monday night livestreams, where I work through neurofeedback practice in real time and answer questions from the chat. I have anonymized the questions and folded them into the teaching. If you train your own brain or work with clients, this is the kind of decision-making I do every session.
What Is the Difference Between One-Channel and Two-Channel Neurofeedback?
The field of neurofeedback is about 60 years old, and our ability to train the brain has run ahead of our ability to fully explain what we are doing. So I keep the framework simple and built on how the brain is actually organized.
One-channel training puts a single active electrode on the scalp and trains one or more frequency bands at that site. You might run a C4 SMR protocol for 30 minutes, rewarding 12 to 15 Hz while inhibiting theta and high beta. Or you might split the session into segments, alpha first, then SMR.
Two-channel training uses two active sites at once. That comes in two flavors. A sum protocol adds the two channels together before the software measures the bands. A dual protocol trains both sites with their own band targets, and you have to move six or more parameters at the same time to earn a reward.
The key point about two-channel work: you are training the relationship between the two sites, not just doubling the training. You are shaping their tendency to co-activate, co-regulate, and build coherence with each other. That is a different effect than running each site alone.
How Should You Decide What to Train First?
Brain maps inform training. They do not dictate it. The map is the brain's pattern, not the person, and average is not the same as optimal. Training someone toward the middle of the bell curve because that is where the database norms sit is a mistake. I use the QEEG to understand the brain, then train individual resources based on what that person needs.
Think about the three legs of the stool: sleep, stress, and attention. As you stabilize those, the whole system tends to regulate. So I start there before going after higher-order goals.
Sequence matters for safety. Deep creativity or flow-state work on top of unaddressed anxiety can drop someone into their worst material. Strong executive-function training can do the same. When anxiety is loud, calm it first or interleave it with the harder work.
How Do You Sequence Segments Within a Session?
Sequencing means running sets of one-channel protocols inside a single session. A classic move is to split 30 minutes into two segments: 15 minutes of an alpha reward, then 15 minutes of SMR or beta.
Short segments let you swap pieces in and out to find what works. If your C4 SMR (a classic ADHD protocol) is producing good focus but sleep is still rough, hold that segment constant and test a C3 beta against a CZ SMR to see how sleep responds. Watch how executive function and sleep shift, then keep the segment that helps.
The framework I use here is the old arousal model, also called the activation model. Alpha sits as a neutral idle, beta as active, fast beta as faster and more active, delta as inactive, theta as transitional. The model also accounts for developmental change, because the brain speeds up over time. You can read that speed directly from the peak alpha frequency. Aging, developmental difficulty, brain injury, deep fatigue, and illness all drag those speeds down, and you have to account for that when you pick reward bands. That is where the idea of an individualized optimal reward frequency came from, long before infra-low training existed. The arousal model has worked across sleep, ADHD, substance use, autism, and anxiety. I lay a laterality model on top of it, because I trained in a laterality lab and think about hemispheric dominance running across the arousal framework.
Two practical sequencing tricks:
- Whatever you end on tends to produce the stronger subjective after-effect. If you want the effect of a protocol, run it last.
- You can use the second segment to buffer the first. SMR or C4 training acts like gravity. Run an experimental alpha or beta protocol, then close with C4 SMR, and it will tamp down stray after-effects.
I also pair sites the way you pair muscle groups in the gym. Biceps and triceps work well together because they oppose each other. C3 beta with C4 SMR works for the same reason. So does FZ theta-down with PZ alpha-up. You are training opposing systems that balance.
When Should You Use Sum vs Dual Two-Channel Training?
Once your segments are dialed in, some protocols can combine into two channels. Be cautious about putting a major hub like a mid-temporal site or PZ back midline into a two-channel protocol. Two-channel training builds information flow and coherence between the locations, especially with a sum. That can be useful, but only if you intend it.
Homotopic sites, the mirror-image areas across the two hemispheres, combine well in a difference protocol. A T5 minus T6 difference does something similar to running T5 minus A1 and T6 minus A2 as a dual. You may want the reward frequencies set slightly differently, but the logic holds.
The strong example I keep coming back to: C3 and C4. They are referential, tied-together tissues. If you train C4 minus A2 with SMR, you get a small resonant SMR effect at C3 even though you are not training there. So running C3 beta with C4 SMR as a dual makes the left-side beta more interdependent with the right-side SMR. That stabilizes executive function powerfully. But dial in the beta on the left and the SMR on the right as separate segments first. Find the frequencies that feel right and that help sleep onset, sleep maintenance, and focus before you commit them to a dual. The wrong frequencies in a dual create side effects much faster.
Why Does the Tissue Under the Electrode Matter More Than the Math?
When you subtract two sites, C4 minus PZ for example, it is tempting to think you are training only the subtracted signal. Each electrode location has its own behavior and its own failure modes, and each contributes to the measured signal differently.
PZ, over posterior cingulate territory, mostly wants to make robust alpha. It does not shift theta or beta much. So if you pair PZ minus A1 alpha with C4 minus A2 SMR in a dual, PZ does not adopt good alpha behavior. It picks up the other half of the protocol and starts producing beta, which is one of its failure modes: stuck in beta, threat-sensitive, ruminative, fearful. You do not want to risk that. Yet C4 minus PZ rewarding low beta, roughly 11.5 to 14.5 Hz, works fine, because C4 is far more willing to make low-beta SMR than PZ is to make uncomfortable fast alpha at that same frequency.
The anterior cingulate has two big failure modes: too much theta or too much beta. When you see low alpha and low neutral tone there, look underneath and you will usually find relative theta power or beta coherence driving it. The direction you train is back toward good regulation. Alpha wants to be made at FC minus PZ, not at FC alone. Know how each tissue falls over, and train it back toward balance.
What EEG Patterns Show Up With Anger and Rage?
Always read these as likely or plausible, never as certainties. A pattern on your map does not mean the experience is yours.
Anger, frustration, and a strong negativity bias tend to show up as activation on the right front of the head: F8, FT8, sometimes F4. There is an avoidance system there that ramps up when you feel negative or low, and it lateralizes toward FT8. I call that site the problem child. When it is too active and stays active, you get a strong avoid response and a sense that everything is hard, often with elevated beta.
I think about the front of the brain as characters. The stabilizer on the left and supervisor on the right at the precentral areas. The CEO or project manager in front, the lifeguard in the cingulate, the princess and the pea at right temporoparietal. The left-front approach system is the happy kid on the porch saying hi, what's up, I want to do it. The right-front avoid system is the grumpy old man saying go away. They are supposed to sit out front playing checkers, balancing each other so you meet the world appropriately, neither blind nor too fearful.
When the right front is too active or disconnected, the left struggles to manage positivity and you get irritability and negativity. Phineas Gage is the original case: a railroad spike through the front of the head turned the sweetest man irascible and angry. Anger is often a fear response too, and that flavor sits more in the back, in posterior cingulate rumination. Exhaustion and anxiety make people brittle and short the same way. For more on the threat side of this, see biohacking fight or flight.
How Does Right-Front Theta Differ From Left-Front Theta?
Mood does not have a thermostat. It is hard to see mood directly in an EEG, but you can see what drives the experience of it: slowed alpha speed and frontal asymmetry, usually with a left-sided driver showing left-front alpha or theta excess.
Left-front theta (F3, F5, F7, AF3 and nearby) relates to mood, but also to motivation. It looks like inattentive ADHD of the mood. The happy kid on the porch is inside saying it is too hard, leave me alone. Effort and initiation are where the difficulty lands.
Right-front theta is different. This is a feeling of dread, of overwhelm, of being frozen, bracing against things that feel nearly impossible. I call it dread pirate theta. Almost everyone I describe it to recognizes the word dread immediately. Fifty years ago we would have called this agitated depression. I read it as anxiety driving depression, and I target it like atypical depression: treat the theta as the culprit, more than the alpha or beta asymmetries.
A note on medication. If someone is on buspirone, the drug blunts that right-front theta and it resists change. Treat it like ADHD work, attack the theta hard, and recognize buspirone is a blood-level drug with a long washout, so you train through it rather than around it. Benzodiazepines produce beta excess that will not shift while the drug is on board.
Are Some EEG Patterns Resistant to Neurofeedback?
Neurofeedback taps basic associative learning, and almost everything at the cell level learns that way. What I see instead are drugs that blunt or block the target, benzodiazepine beta and buspirone theta being the clear examples. The signal is there; the pharmacology is sitting on top of it.
Setting Up a Live Two-Channel Session in EEGer
For the live demo I ran a C3 minus A1 and C4 minus A2 dual: theta inhibit at 4 to 7, a beta reward, and high-beta inhibit at 22 to 36. Knowing my own brain, I drop the reward bands by half a hertz, training 14.5 on the left and 11.5 on the right rather than the round numbers. You learn this by feeling frequencies. Some people are fine half a hertz off. Others feel a quarter hertz strongly. EEGer will let you set eighth-of-a-hertz gradations, though I do not go that fine.
A few practical points from the setup:
- A dual needs clean signal on both channels, and it is harder to get going. The brain has to move all six parameters at once, so you earn maybe half to two-thirds as many rewards.
- Always double-check the setup and look for the mistake. Manual setup in EEGer is error-prone.
- Make your artifact reject roughly twice the signal size with eyes open, three times with eyes closed.
- High-frequency, high-amplitude noise at the scalp is a bad scalp connection, where the amplifier bridges the gap and picks up 60 Hz line noise. Reseat the wire, spin it in the socket, or move the amp. When I cleaned up my C3 channel, all that noise dropped away.
For more on the SMR protocol itself, see SMR neurofeedback. If you are new to the whole approach, start with is neurofeedback legitimate and the ADHD guide.
Does Neurofeedback Need More Regulation?
My answer is more rigor, not more regulation. Traditional neurofeedback is closer to fitness than to medicine. You can train wrong and cause trouble, but it takes time. Effects come on slowly, and adverse effects like disrupted sleep or feeling on edge are self-limiting, because people back off when a session feels bad. That self-correction is a feature. It tells you that you are close, and you use changes in sleep, stress, and attention to steer.
The active technologies are a different story. Sending electricity or magnetism into the brain, TDCS, TACS, off-label TMS, carries real risk for consumer use, because we know too little. TMS is medically approved for depression with a narrow set of protocols for exactly that reason. Systems that automate or self-regulate, and infra-low and micro-current approaches, can produce side effects much faster even within otherwise gentle methods.
The failure points in traditional neurofeedback are about communication. The loop breaks when the provider works from a recipe book, does not adjust, or stops listening, and the client does not report adverse effects. That argues for better training and better tracking, not legislation.
I am also skeptical that pushing neurofeedback toward insurance coverage is the right direction. I would rather see the cost floor keep dropping. About 80% of our clients train from home with remote brain mapping and remote support, and that is plenty of supervision for most goals. See remote neurofeedback and, if you are weighing cost, is neurofeedback covered by insurance.
Where Is the Hardware and Software Headed?
The field is running on tools from around the turn of the century. EEGer is excellent and still my main software, but a clinical license runs about seven thousand dollars, and there are few vendors. Pocket Aerobics, which made the rugged Whiz and Q-Wiz amps, appears to have folded. Neurobit out of Poland makes good amplifiers, the Optima two-channel being a solid option, though tariffs and electrode supply out of China are pushing costs up. A small tip for the Optima: it is sensitive to mechanical noise, so wrap a silicone band around each end to keep it off a hard surface, the way the Q-Wiz uses its silicone case.
There is a clear gap in the market for a rugged sub-thousand-dollar two- or four-channel amp paired with friendly software. The manufacturing cost for a four-channel amp is closer to a hundred or two hundred dollars than four hundred, especially if you already hold the IP and code.
I am building toward this. Practice-management software for tracking effects and planning protocols is coming in a month or two, with HIPAA, GDPR, and SOAP-note compliance for people running their own practices. After that, AI trained on brain maps and clinical decision-making can get us maybe 80% of the way to suggesting safe, likely-effective protocols, with a skilled human vetting the plan. The two largest costs in neurofeedback, beyond equipment, are the time spent reviewing and planning protocols and the time spent getting reliable after-effect data back from clients. Better software and a smart assistant trained to make decisions the way I do can drop both.
If you want to get your own brain mapped, we are running a half-price brain map across the office locations, now including Nova Scotia, West Palm Beach, New York City, St. Louis, Orange County, Los Angeles, London, and Stockholm, with more locations opening this year. In the US you can ask for remote and we will send equipment to your door. Outside the US, come to a London, Stockholm, or US office to get mapped and pick up gear.