On my Monday biohacking livestream I set up a neurofeedback session on my own head and walked through every step on camera. This article captures what I taught: how the hardware works, what the brain-wave bands mean, why the training is mostly involuntary, and how SMR training connects daytime focus to nighttime sleep. Audience questions are folded in anonymously.
What does the hardware actually measure off your head?
The amp I used is a two-channel pocket EEG device that runs into software the field has leaned on for decades. The software was built by an engineer who used to write real-time operating systems for F-16 fighters before moving into signal processing. That history matters, because the lowest-latency, least-jittery brain-training software tends to keep the cleanest, fewest layers between the signal coming off your head and the feedback hitting your eyes and ears. For some protocols, that latency makes the difference between training the brain and training noise.
A single channel of EEG is a subtraction. Red minus black gives you one channel of data. You hang a reference electrode off one ear and a ground off the other. The ground can go anywhere, so I do not bother specifying its location in a protocol. When you see a montage written as C4 minus A1, that names the two locations being subtracted to produce the waveform.
One practical detail people miss: coat both sides of the electrode with conductive paste. Bare metal against skin creates a bridge and an antenna, and you pick up 60 Hz line noise across the whole signal in the US. A clean white trace stays thin. A trace picking up 60 Hz looks fuzzy.
If you accidentally flip the active and reference wires on a single channel, the only thing that happens is the waveform flips upside down. We measure the amount of each brain-wave band in that waveform, not the polarity, so a flip does not change the training.
Why is it bald-head friendly, and do you need scalp prep?
I skipped scalp prep that night because I had a freshly showered, bald head and soft skin. You can often get away with it, though I do not recommend skipping it as a habit. When you place a wire on bare skin, smear a little paste on the spot first. Paste will find its way onto your dog, your kids, and the backyard regardless. Paste handling is one of those skills you only learn by doing it a thousand times.
I measured the vertex (Cz) by finding the top of my head, going up an inch from the right preauricular notch (T4), and splitting the difference to land C4 over the right sensorimotor cortex. Later in the session I moved the active wire to Cz at the vertex.
What do the brain-wave bands mean during training?
A neurofeedback display filters the raw signal into bands, and each band has a job.
Theta (roughly 4 to 7 Hz) behaves like a release signal in a circuit. The cortex is built from millions of mini-columns, each a tight cluster of neurons firing together and producing electricity. When a chunk of that tissue wants to let its behavior run automatically, or take itself partly offline, theta goes up. In most clinical pictures, excess theta is the thing getting in the way. Across ADHD, some anxiety flavors, tics, certain sleep problems, and mood difficulty, excess theta shows up as a common driver. The point is never one brain wave in isolation. It is which set of bands, doing what, in which part of the brain. I cover the broader picture in Biohacking with EEG Phenotypes.
SMR, sensorimotor rhythm (12 to 15 Hz for most people) is the band I rewarded. SMR only occurs on the sensorimotor strip. The same frequency elsewhere is regular beta or fast alpha. SMR functions like a calming, regulatory rhythm despite its beta-like frequency. I backed my reward band off slightly that day (down to about 11.75 to 14.75 Hz) because I was a little stressed, the way you take weight off the bar to make a lift easier. More on this band in SMR Neurofeedback.
High beta (I used 22 to 36 Hz) spikes when you are stressed; bursts of anxiety produce beta spindles. The deeper reason to inhibit fast beta is that muscle activity lives in the same fast range and is louder than the brain because it sits outside the skull. If you reward without inhibiting high beta, a person can learn to clench and tense to drive the feedback. At least one paper shows you can involuntarily recruit muscle fibers this way, which teaches the wrong thing.
So a standard setup inhibits a slow band, rewards a middle band, and inhibits a fast band. When all three trend the right way at once, theta down, SMR up, high beta down, the game advances and you get a point.
Is neurofeedback voluntary? Do you have to concentrate?
Neurofeedback runs through operant conditioning below conscious awareness. You cannot feel your brain waves, and you do not directly control the feedback. The game runs when your brain produces the target pattern and stalls when activity drifts. The brain notices the change in the outside world and gradually shapes itself toward the rewarded state. Good job, brain. Nope. Good job. The conscious mind mostly gets in the way by moving too much.
It was discovered on cats, and cats are terrible instruction-followers. That is the point. A nonverbal child, a teenager who does not want to be there, even an unconscious person can train. You do not focus on anything specific.
During the session I demonstrated one place where conscious focus helps: I went quiet, stopped moving, and dropped my theta from the mid-40s in microvolts down below 30, pulling my theta-to-beta ratio toward two-to-one, which is healthy for an adult. The software then adjusted, the threshold moved next to where my brain landed, and the rewards got sparse again until it readjusted. That adaptive threshold is the central trick. Every 30 seconds or so the software resets the criterion just above or below where your brain is, so as you tire and theta surges, the task loosens rather than becoming impossible.
I also believe, though the literature has not borne this out yet, that heavy social engagement breaks the process. There is good evidence that implicit learning, picking up rules in the background without being told, requires a state that social cueing floods. Sitting on Facebook during a session likely interferes. A quiet, passive state works best.
Why does SMR training help both focus and sleep?
SMR and sleep spindles are the same thalamocortical circuit activity in two states of consciousness. Awake, this rhythm enables physical stillness and calm focus. Asleep, those same circuits fire spindles that protect sleep architecture by gating out external sensory intrusions. You hear a car outside and your brain bursts a spindle to keep you under instead of waking you.
Training C4 over the right sensorimotor strip supports executive function, impulse control, and the supervision of attention. Training the vertex at Cz pulls in body calmness and helps with both sleep onset and the production of vertex spindles that keep you asleep. Weak SMR at Cz can also block onset, the way you struggle to power down the machine you have been driving all day. That night my raw Cz signal showed visible SMR spindles, probably because I was tired. I write more about this overlap in Biohacking Sleep.
Why start with a brain map, and what is the common misread?
You do not pick random targets. You start with a QEEG brain map to understand a person from their resting patterns, build models about resources and performance, and decide which protocols matter. Full detail on that process lives in the QEEG Brain Mapping guide.
Here is the misread I want to correct. Brain mapping is not about being abnormal, and you do not train toward the middle of a bell curve. We use an age-matched sample only as a yardstick. People are weird, and the weirdness is fine. We look at the outliers, the features sticking out, and ask whether they matter for how you tend to act and feel. Resting traits are stable day to day. They suggest tendencies, not moment-to-moment states. If something does matter, that is good news, because it is not fixed. Tools like neurofeedback can push the brain around. That malleability is what neuroplasticity gives you.
How do brain waves get generated in the first place?
A question came up about whether specific neurons own specific brain waves. They do not. Picture the cortex as a dense city of skyscrapers. Each mini-column is one building holding about 30,000 neurons plus up to 100,000 support cells, all firing together as one rhythm. What defines a column is that it shares a single electrical environment. Some layers connect to neighboring buildings like a clothesline across the alley; others run long-distance lines across town.
A brain wave is the synchronized firing of thousands of cells, measured through scalp, skull, meninges, and cerebrospinal fluid. Local tissue tunes groups of columns to fire faster, slower, more flexibly, or to talk to distant regions differently. Block parties everywhere, each playing its own song, sometimes pulling the neighbors into the same tune.
Can you measure gamma, and is it a marker for intelligence?
Gamma is one of the field's running misunderstandings. As frequency climbs, amplitude shrinks, and every layer of tissue acts as a filter that dampens the signal. Theta at 10 microvolts is four big waves a second and easy to see. The same notional energy at 40 Hz is tiny and largely gone by the time it reaches the scalp. Much of the older EEG literature reporting 40 Hz gamma was later retracted once it became clear that eye-movement muscle activity mimics a 40 Hz burst.
The gamma that seems to matter for advanced meditators, from work by researchers on long-term contemplative practice, sits far higher than 40 Hz, in ranges you cannot touch without tens of thousands of dollars in active electrodes and tightly controlled environments. Gamma is tied to conscious awareness. Around 40 Hz gamma couples in time with roughly 4 Hz theta; many gamma cycles nest inside one theta cycle. Break that cross-frequency coupling and you get unconsciousness, which is what anesthetics do. Treat this as the frontier theory it is rather than established practice. For a grounded look at trained brain states, see Biohacking Meditation.
What I want you to take from this
A neurofeedback session looks almost boring from the outside. I sit still, talk, and a Pac-Man moves while three bars track theta, SMR, and high beta. The work is happening underneath, through adaptive thresholds and operant shaping the conscious mind never touches. If you want to try it, the order is fixed: get a QEEG brain map first, let the resting data point to targets that matter for you, then train. Start there, and you will know what you are changing and why.