This article comes from my weekly Neurofeedback and Chill livestream, where I run a session on myself, narrate what I'm doing, and answer brain questions from the chat. The questions below come from viewers; I've removed names and kept the substance. If you're new to the format, I'm Andrew Hill, a clinical neuroscientist, and most of what follows is the kind of thing I talk through while electrodes are drying on my scalp.
Why does alpha push back when you train it?
Alpha (8 to 12 Hz) sits in the middle of the frequency spectrum, between the slow waves (delta and theta) and the fast waves (beta and gamma). It functions as the brain's idling rhythm: a resting-ready, quiescent state that also gates sensory input by inhibiting task-irrelevant cortex. You can read more on that dual role in Decoding Alpha Waves: Your Brain's Idle and Its Brakes.
Because alpha is an idling rhythm, it holds the frequency it idles at. It resists dysregulation. When you train it with neurofeedback and push too linearly, you usually get rebound: alpha surging back against the training. That rebound is sometimes the point. Certain trauma protocols (the Sabourin and Fisher style work) use an alpha rebound deliberately to produce the effect they want. Most of the time, though, alpha does not move the way you expect, while theta and beta generally do.
That property shapes how I plan a session. Instead of training alpha directly, I often leave a gap in the alpha band and let alpha do what it needs to do on its own. My recent favorite session is an FZ minus PZ down-training: I'm giving the brain a learning signal to reduce excess activity in the cingulate regions. The cingulates respond well to down-training.
Does alpha asymmetry mean depression?
Alpha asymmetry shows up in QEEG, and it gets read as a depression marker. It can be that. In my clinical experience, an alpha asymmetry is valid for the specific person and complaint maybe half the time, a bit more. Some EEG phenotypes are more trustworthy, where the pattern usually means the thing. Alpha asymmetry is a weaker signal.
If someone presents with mood as the primary complaint and a clear asymmetry shows up on the map, plus a meaningful slowing of alpha speed, there's a good role for asymmetry training (down on the left, up on the right). Both of those signatures can also show up with simple fatigue or a sleep complaint. So I want the QEEG signature and the matching complaint before I train an asymmetry. Without that, the training tends not to move the way you'd hope. For the broader logic of reading patterns off the map, see Biohacking with EEG Phenotypes and the QEEG Brain Mapping guide.
Is high alpha at FZ resistant to down-training?
It depends what "high alpha" means. Slow, high-amplitude alpha (alpha 1, roughly 7 to 10 Hz) and fast alpha (10 to 12 or 10 to 13 Hz) behave differently. At PZ in the back, where you make a lot of alpha anyway, alpha is resistant to being trained down for obvious reasons. At FZ it's generally not that resistant, though all alpha carries some baseline resistance you have to plan around.
A site is never trained alone. You're always working a montage, a combination of sites. I use FZ minus PZ because the two cingulate regions often work together. I also use FZ minus A1, and I use FZ minus CZ in the back as a short throw, which becomes a variant of sensorimotor rhythm training. FZ can be touchy, so I don't usually train down regular or slow alpha there. If I see excess fast alpha (10 to 12 or 10 to 13 Hz) at FZ, that's often tied to procrastination and some forms of anxiety, and it tends to get in the way. In that case I inhibit, often 12 to 20 Hz because beta bursts are more of a problem on the cingulates, and I sometimes widen to 10 to 20 to capture the surging fast alpha. For more on the FZ-and-procrastination link, see Procrastination: Biohacking Your Brain for Action and Biohacking Anxiety.
Why shouldn't you reward beta at FZ?
Rewarding beta at FZ is a reliable way to produce side effects. A regular beta reward (say 15 to 18 Hz at FZ minus A1) generates a lot of trouble in most people: OCD-level intrusiveness, enhanced or newly created ticks, perseveration, anxiety. This holds even when the QEEG shows low beta at FZ. You would not know this from the map alone. You learn it from knowing the tissue.
The cingulates do most of their work in very low beta and by inhibiting theta and shifting in and out of alpha. Their failure modes are excess theta or excess beta, not a beta deficit. So FZ doesn't really have a cramp mode that calls for rewarding beta up. If someone has high theta at FZ plus tic-like features (songs stuck in the head, nail biting without the classic intrusive-thought structure of OCD), the problem is the theta. You can sometimes train up SMR to address it, but the placement matters.
What counts as SMR, and where does it live?
SMR (sensorimotor rhythm) only occurs on the sensorimotor strip, the C strip. Low beta at FZ minus A1 (12 to 15 Hz) sits in the right frequency range but is not genuine sensorimotor rhythm because the electrode placement is wrong, and that montage will likely cause trouble. FZ minus CZ at 12 to 15 Hz is real SMR training and is safer, though even that produces side effects in some people. Tonight's second protocol was a C4 minus A2 SMR reward (11.75 to 14.75 Hz), which is genuine sensorimotor training. For the full picture on why SMR is worth training and what it does for sleep, focus, and self-control, see SMR Neurofeedback.
A useful aside: a weighted blanket creates SMR. Deep pressure and tight holding produce sensorimotor rhythm and deep stillness. That's the mechanism behind Temple Grandin's cow-crusher discovery, the calm that comes from strong, even pressure.
Is LENS the same thing as neurofeedback?
LENS stands for Low Energy Neurofeedback System, and it's also named for Len Ochs, who built it. It's a form of microcurrent stimulation that sends a small current into the brain rather than running the operant-conditioning loop that traditional neurofeedback uses.
The origin story, which I believe is true and not apocryphal: during a depression study, people kept standing up from one particular amplifier feeling dramatically different, asking what had just happened. The team eventually found a short in the amplifier that was sending current back down (probably the ground wire), producing a microcurrent stimulation at a specific frequency. Ochs engineered that into a device with an assessment-plus-stimulation process built in. That process is a bit of a black box.
I don't love the technology. It produces side effects too easily and needs a very skilled provider, and I don't know many skilled LENS providers. I see people most months coming in because of side effects another system provoked, sometimes strong anxiety. After Ochs Labs fell apart, several principals each took a version of the software and made their own, which is why you now have IASIS, HPN, Clarity, and others that are essentially the same approach. Dave Siever's re-engineered platform got Ochs's blessing and is a later iteration.
There is some real signal here. I saw the HPN version presented at an ISNR conference showing meaningful change in NFL players after CTE damage, so for repetitive brain injury it may be worth finding. But the things people credit LENS for, like brain injury recovery, traditional neurofeedback also handles fine, with much less risk of side effects when you move gently and iterate.
What kind of neurofeedback do I actually do?
What I run on the stream is phenotype training, also called traditional neurofeedback: amplitude (band) training. My version is QEEG-informed. I map the brain, read the phenotypes to understand the person, then iteratively work through the phenotype manipulation, taking feedback from the person's lived experience, adjusting, and remapping. At least half the field still works this way, and I think it gives the most reliable results across people with the least side-effect risk. It blends the client-focused, intuitive side with the rigorous, map-based side. If you're weighing options, Is Neurofeedback Legitimate? and the neurofeedback topic hub cover the evidence base.
Why does within-session change not predict results?
A viewer worried that an SMR theta-inhibit dropped from 31 to 25 microvolts one week and bounced to 30 the next. Don't over-focus on raw amplitude trends across weeks. The same person trained at the same location shows variable band levels day to day. Fatigue and not eating both push slow-wave amplitude around.
When I looked at large populations doing neurofeedback and examined the signature of change within sessions, only about half of people showed a within-session change on the trained bands. Yet all of them produced the evoked potential that shows the brain registering that neurofeedback is happening and reacting to the activity under the wire. The people who don't show the in-session trend still make changes weeks and months later. So focus on repeating the protocol and getting the subjective effect you're after, not on the session-to-session microvolts. This is operant conditioning operating below conscious tracking, which connects to how habits and plasticity get built; see Biohacking Plasticity.
Can a kid fidget, do homework, or look at a picture during a session?
Yes to all of it, with caveats. Sitting relatively still matters because movement tugs the wires and changes the signal, but it doesn't have to be perfect. A fidget toy, a plushy, a weighted blanket, or art books are fine. What you avoid is anything social or with competing audio and visuals.
Neurofeedback is implicit learning. The brain figures out the rules by associating its own activity with the beeps; nobody is thinking about brain waves or trying to feel them. Social engagement floods that channel. Arguing on social media, or a screen full of explosions and crashes, masks the simple discrete signal the brain is learning from. So no social media, and movies only when a kid needs them for engagement.
If a kid moves too much, show them that clenching the jaw or wiggling the head changes the signal and stalls the game. The games become self-defeating with motion, which corrects the behavior on its own. You can also unveil picture grids of their favorite things to hold attention, which works well for kids with hyperfocused interests.
Does task-based training help or hurt?
For some flavors of ADHD and anxiety, concentrating makes the brain brown out: you get a surge of alpha everywhere, a near-dissociative alpha under stress. Having that kid do homework during a session elicits the phenomenon, so the software catches the excessive slow-wave surge faster. That can make training go faster. Watch ergonomics, because a kid curled over a book gives you poor signal, and wait until six to eight sessions in, once you know the protocols are working. For the parenting context around ADHD, see Does Neurofeedback Work for ADHD? and the ADHD topic hub.
What about ISF, ILF, and slow-wave training for ADHD?
ISF and ILF train below 1 Hz, the sub-hertz range. You can probably get ADHD effects training slow, given that slow cortical potential (SCP) training, another sub-hertz signal, has good ADHD evidence. But I don't have a clean way to do ILF or ISF for ADHD, and I don't see the reason to reach for it. SMR and beta training already work for most people. Getting fancy with the frequency range is unnecessary when the standard approach produces the change.
How does the snacker protocol blunt late-night cravings?
The protocol I call the snacker is an FP1 minus M2 low-beta reward (12 to 15 Hz) with a theta inhibit. The theory, and I mean theory: the FP1 minus M2 montage may engage the right insula along with frontal cortex. The insula is heavily involved in the sensation of feeding and craving. TMS to the insula can help people stop smoking, so this is the addictive, appetitive, oral craving family.
I don't know exactly why it works, but it reliably reduces the urge to graze, and the pull toward sugar, alcohol, even late-night cannabis. With three different clients, all women with dysregulated gut after gastric surgery and long histories of sugar craving, the same striking thing happened: after the session, they decided they no longer liked sweet foods. One mom bit into her usual cake pop and could still taste sweet, but the food registered as a cloying chemical bomb she didn't want to keep eating. The protocol shifted the appetitive relationship to hyperpalatable food. For the broader frame on retraining vices, see Biohacking Bad Habits.
A note on the live session
Tonight I flipped the order of my two inhibits on the FZ minus PZ down-training, which changed the emphasis between the theta band and the 12-to-20 band by way of different auto-goals. The last two weeks, training down 12 to 20 gave me a mood lift. With the order flipped, I expected a surge of focus from leaning harder on theta and instead felt a little woozy and more distractable. That subjective read is the data. It tells me how to arrange the protocol for my own brain next time, which is exactly the iterative, response-driven loop that QEEG-informed neurofeedback runs on.
If you want to map your own brain and build a protocol from real data rather than guesswork, that's what we do at Peak Brain. Drop questions in the comments on the stream, and I'll work through them next Monday.