On a recent Monday livestream I ran a full neurofeedback session on camera, narrating each step while I answered audience questions. This article pulls the teaching out of that session: how hemoencephalography (HEG) training works, how I set up EEG protocols for the left and right hemispheres, and how I think about protocol selection for mood, stroke recovery, and complex presentations. Audience questions are paraphrased and anonymized.
What is HEG neurofeedback, and how does it differ from EEG?
Most people who have heard of neurofeedback picture EEG training, where you shape the brain's electrical rhythms. HEG trains something else: blood flow.
The headset I use carries a set of small infrared lenses pointed at the forehead. They measure heat coming off the head, which tracks vascular dynamics in the underlying cortex. When you concentrate, blood flow surges to the region, and the heat signal rises. I tie that signal to a moving ball, a pitch, and a colored trace, then practice driving it upward.
The signal lags your effort by about one and a half to two seconds. That delay is the blood-oxygen-level-dependent response, the same BOLD signal measured in fMRI. The vascular system takes that long to deliver oxygenated blood to active tissue. HEG reads that underlying physiology with a spatial resolution that is broader than fMRI, reading vascular dynamics across the forehead rather than millimeter-scale slices.
In the session, concentration drove the signal up reliably. Trying to summon a "happy thought" did almost nothing. That tells you something useful: for me, the concentration circuitry responds fast and the affective route is slower or quieter. If I wanted to train emotional access, I would not reach for HEG anyway. HEG tends to be activating, and per session it can push you toward an over-aroused, slightly on-edge state. Over many sessions it might lift mood, but it is the wrong primary tool for that goal.
How do you train HEG without thresholds?
The version I demonstrated is unthresholded and semi-voluntary. The signal moves with my vascular tone, I get audio and visual feedback, and I deliberately concentrate to produce surges. There is no pass-fail criterion.
A few build details if you want to make your own design in Bio Explorer: get a real-time HEG trace on screen first, then route it to a MIDI output with a defined range. I smooth the signal before it hits the audio so it does not swing wildly. I use a moving window of roughly 30 seconds with an update around every half second. That is a deliberately coarse signal, and that is the point. You want the brain learning the slow vascular dynamic, not chasing noise.
How do you set up an EEG neurofeedback protocol for each hemisphere?
After the HEG block, I switched to EEG training. EEG neurofeedback is the more familiar, largely involuntary shaping of brainwaves through operant conditioning below conscious awareness. The game rewards target patterns and dims when the brain drifts off them. Your conscious mind does not steer it. The brain gradually learns.
On the left side of the head (C3), I ran a sensorimotor rhythm reward in the 14.75 to 17.75 Hz band, inhibited slow activity at 4 to 7 Hz (theta), and inhibited high beta at 24 to 36 Hz. Rewarding faster SMR-range activity at a left central site supports focus and self-regulation.
On the right side (C4), I shifted the reward band down to around 12 to 15 Hz SMR and lowered the high inhibit by a couple of hertz. Site and frequency matter together. Rewarding the faster 14.75 Hz band on the right side of the head would tend to drive arousal up and make me anxious quickly. The same protocol that helps on the left can over-activate on the right.
This is why multi-segment sessions demand a rhythm and a discipline. You move the electrode when you change sides. You do not start the next segment with the wire in the old spot. If you train one way and then train the opposite way over the same tissue, you can overtrain. I build protocols to be forgiving of small errors, but forgiving is not the same as ideal. If you are a client and you catch a mistake mid-session, stop and check in rather than immediately training the "right" thing on top of it.
How do you choose a protocol for mood and emotional access?
A viewer asked what I would do for emotional access and happiness. The honest answer is that it depends on why the access is blocked, and the QEEG usually tells you.
Several features point at mood. Left-versus-right frontal alpha asymmetry maps onto approach versus avoidance motivation. Tonic, stuck alpha in the left frontal region is a candidate target. A slowed peak alpha frequency, which is the same measure as the eyes-closed posterior dominant rhythm at PZ, can pinch processing speed and pull both mood and cognition down with it.
The cause drives the protocol. If someone struggles to feel emotion at all, deep internal access protocols like alpha-theta or other flow-state approaches can help open that door. If the block is stress, fatigue, poor sleep, trauma, or attention, you go after that bottleneck instead.
One client this same week ran a compound protocol and reported being remarkably blissed out afterward. It stacked two pieces: a frontal dual at F7 and F8, rewarding beta and inhibiting alpha on the left while doing the opposite on the right, followed by a midline FZ-to-PZ down-training that inhibited excess theta. He is also a heavy meditator, so I would not promise that result for everyone. It shows how much you can move state with the right combination.
What precautions apply to neurofeedback after stroke?
A practitioner described a client three to five years out from two strokes, with 3 to 4 Hz dominant activity and spikes across several posterior and frontal sites. Here is how I would approach it.
First, look at more than one montage. A Laplacian or a local bipolar pair plus a linked global reference will show you where the slow activity actually lives. Multiple strokes producing excess 3 to 4 Hz usually mean hypoperfusion in the damaged regions. Expect hypercoherence in theta between affected sites and hypocoherence in alpha or beta. Map which regions are yoked together, then look at the cross-frequency coupling. If two sites are locked together in 3 to 4 Hz with high amplitude and hypercoherence, you often see disinhibited sensory filtering, which shows up as sensory flooding, and sometimes anxiety or trauma symptoms riding along.
Read the raw EEG, not just the processed map. Spikes and spindles do not always appear in the QEEG. Spikes are more seizure-prone; spindles are bursty over-revving. You may see beta spindles where the brain is trying to cut through the slow bursts. Examine the one-hertz-wide bins to find the specific frequencies worth targeting.
Then work slowly. Assume the brain fatigues easily, so use shorter sessions and slightly gentler reward thresholds. HEG works here too, in short blocks of a few minutes, using the unthresholded version. Complementary moves are worth considering: a lower-carb approach or exogenous ketones, careful attention to sleep so untreated apnea is not driving secondary inflammation, and, with neurologist clearance, hyperbaric oxygen in a hard chamber at two atmospheres for around 20 sessions across a month. Strokes this old carry enough stability to train without acute medical oversight on-site, though recent injuries always require the neurologist on board first.
How do you work with complex cases like schizophrenia, pain, and dysautonomia together?
Another viewer asked about someone presenting with pain, schizophrenia, and dysautonomia at once. Pain and dysautonomia often travel together. The schizophrenia label deserves caution. Strong OCD or anxiety can mimic a thought disorder, and so can infections or fevers. Confirm before you accept the label.
When it is genuinely a schizophrenia-spectrum presentation, the neurofeedback approach is basic, with one caveat: change comes slowly. In the people I have worked with who had symptomatic schizophrenia, the training produced anxiety, sleep, and executive-function changes on measures like the IVA continuous performance test, but the dysregulation in clarity of thought persisted at four to six months. The interesting part came later. Each of them stopped training around that point, and when I saw them several months afterward, they showed much stronger integration, better hygiene, changed handwriting, clearer organization. That long-tail effect after stopping is something I have seen specifically in these cases. This is clinical observation across a small number of people, not controlled trial data.
The broader principle: you can name a dysregulated anterior cingulate and the phenomena that follow from it without deciding whether the chart says OCD or bipolar. That keeps you ethical and inside your scope, as long as the person's acuity does not exceed what you can safely provide. If it does, build a referral network around yourself.
Most neurodivergent presentations carry dysregulated EEG. A meaningful fraction show spikes or spindles. You will not always catch them in the processed map, so scroll the raw traces, especially with atypical anxiety or confused thought. Look for big slow swells, phase reversals, and regions that seem to disconnect and go offline.
What is the hard part of neurofeedback?
Sticking wires to your head, setting filters, getting a clean signal: none of that is the difficult part. The difficulty is knowing what to do next. After a session leaves you wired, focused, or tired, do you continue the same protocol or change it? Protocol adjustment over a course of training is a complex decision, and it still depends on experience rather than any automated rule.
A practical reminder from the session itself: I spent several minutes troubleshooting a "dead" amplifier before realizing I had not clipped on the ear references and ground. No reference means no signal. Nine times out of ten the answer is on the screen; the tenth time something is not plugged in.
What comes next
I am close to releasing simpler neurofeedback software that handles this style of HEG and basic EEG training, since there is no good reason to be stuck with clunky legacy tools. Two other projects come first: a directory of neurofeedback providers across the field, and a brain fitness tracker that lets you log protocols against sleep, stress, and goals to see retrospectively which interventions move which traits.
If you want to look at your own brain, I run offices in New York, St. Louis, Los Angeles, and Orange County, with a discounted QEEG available, and you can book a free call to talk through what neurofeedback and a brain map can show you.