This piece comes from my weekly Neurofeedback & Chill livestream, where I run neurofeedback on myself, explain what is happening as I go, and take questions. This week's topic was memory: what it actually is, what only looks like it, and what you can do to train it. I am a neuroscientist and biohacker, not your physician. Use this to get oriented and to take to your own providers, not as medical advice.
What is memory, and where does it live in the brain?
Memory is only partly understood. The hippocampus, a seahorse-shaped structure in the medial temporal lobe, does not store your memories. It moves them into storage. The storage itself is distributed across the cortex.
Over a hundred years of experiments have found several signatures of a memory being laid down: changing strengths of activation between neurons, even traces in the prior pattern of synapses a neuron used to hold. The picture that keeps emerging is that a memory is less a file in a folder and more a mathematical signature that gets replicated across the brain. Researchers call this holographic memory. We also have no evidence of a storage ceiling. At the level of the synapse, the bit depth of the system is enormous, and it can be both analog (concentration gradients, firing rates) and digital.
Here is the scale. Run a simplified thought experiment where every neuron connects once to every other neuron. The information density comes out to the number of neurons raised to the power of the number of neurons. That number exceeds the estimated count of atoms in the universe. You have capacity to store more than there will ever be stuff to store. So when someone tells me they are running out of memory, storage is not the problem.
If you want the deeper background on how circuits adapt and rewire, I have written about that in Biohacking Plasticity.
Why most "memory problems" are actually upstream resource failures
Most of the time, when someone says their memory is failing, they are describing some other resource that limits their experience of memory. The architecture of memory is intact. Something upstream is getting in the way.
Is it attention or is it memory?
If you have an attention pattern, you drift into alpha and stop encoding. The information never gets in. That looks like a memory failure. It is an encoding failure driven by inattention. You cannot recall what you never recorded. ADHD-type patterns sit right here, and I cover the attention side in detail in the neurofeedback for ADHD guide.
Is it speed of processing?
The big one that most people feel as "bad memory" is delayed recall. Word-finding trouble, name retrieval, tip-of-the-tongue, the answer that surfaces ten minutes later. That is a speed phenomenon, and it tracks with the timing of your alpha rhythm, specifically how well the alpha stays synchronized within a hemisphere. On a QEEG z-score page you can watch this directly. If your left-hemisphere alpha frequencies spread apart by a z-score or more across locations, you tend to get noticeable word-finding and retrieval lag. Under 0.5 and most people do not notice much.
Is it working memory?
Working memory is the brain's RAM. It holds the active items in your mind, with a finite limit of about seven items, plus or minus two (Miller, 1956). Holding even one more reliable item is worth a meaningful step on a cognitive test. Working memory turns over the moment a new thought or a night's sleep arrives.
Two things limit it. Speed: how fast you can grab words and names and load them in. And protection: whether the next incoming item knocks something out of your seven slots. If you cannot protect the space, you retrieve information and it slides right back out. The fix is usually either raising beta tone for stronger working-memory drive, or, more often, bringing theta down so the inhibitory tone at the frontal midline and the central strip lets you control what enters and leaves your mind. I covered the working-memory side of this in more depth in Biohacking Intelligence.
The practical order of operations: get sleep, stress, attention, and processing speed sorted first. Memory usually takes care of itself when you sleep deeply, run a fast and smooth mind, and can hold what you choose to hold.
What hurts true memory structures?
The hippocampus is fairly rugged, but chronic stress is its biggest enemy. Sustained high cortisol dampens hippocampal plasticity and drives down brain-derived neurotrophic factor (BDNF), which is one of the largest brakes on learning and memory. Studies find a smaller hippocampus in depression alongside elevated chronic cortisol (Sheline et al., 1996; Sapolsky, 2000). Cortisol, like blood sugar and insulin, is supposed to oscillate. When it goes up and stays up, the hippocampus atrophies and both storage and retrieval suffer.
Sleep and stress management are the primary work for most people who think they have a memory problem. Biohacking sleep and managing your stress response move the needle before anything else.
What does true memory loss look like first?
When the memory structures themselves are involved, the earliest and largest changes show up in episodic memory: experiences you have had become harder to access. Forgetting what your partner said after dinner last night is usually fatigue or attention. Forgetting your graduation or your kid's wedding is episodic, and that is true memory.
If you are seeing episodic loss, especially with an aging context, the foundation still comes first, and then you go deeper. Dale Bredesen's research describes a large set of metabolic factors that accelerate brain aging and the dementias, with case series of people improving after multimodal metabolic interventions (Bredesen, 2014). His work is accessible to read, and his company Apollo Health runs the ReCODE and PreCODE programs to screen those factors. If you are outside the US, a functional-medicine workup of MTHFR, homocysteine, oxidative-stress markers, and related labs gets you to similar territory.
My standing advice if you are worried about memory with an aging context: get a QEEG brain map, consider neurofeedback, and look at your metabolic and genetic environment. The critical aging window work is worth knowing here, because the brain starts changing earlier than most people expect.
What does neurofeedback actually train, and how?
On the livestream I ran two protocols on myself. The first was C4 referenced to the right ear (A2), an SMR protocol rewarding roughly 11.75 to 14.75 Hz while inhibiting 4 to 7 Hz theta and high beta. The second was a vertex (CZ) protocol, a slower SMR to support sleep, again inhibiting theta.
The mechanism is operant conditioning. The software reads my brainwaves off the scalp electrode in real time, filters out specific bands, and rewards the brain with small auditory and visual events when the medium-frequency band (sensorimotor rhythm) holds or grows while theta and high beta stay down. Hold criteria for half a second across all three bands and you get a reward event. The brain notices the pattern and trends toward it. Within a session I felt the typical SMR effects: a small lift in energy and mood, and the day's stress softening. SMR often reads as a clearing or settling experience, and people feel it differently depending on site and frequency.
A few technical points that matter for memory work:
- Site changes the frequency. The left hemisphere runs faster than the right, the front faster than the back. The same SMR is a different number depending on where you place the electrode. CZ runs a bit slower than C4.
- The reference matters. C4 minus A2 narrows the focus onto the tissue under the electrode and sits on the supervisory attention circuit, which sharpens focus. C4 minus A1 gives a broader, more global SMR.
- Beta versus alpha aging. Beta frequency is roughly set early in life. Alpha keeps speeding up through the mid-teens with myelination and cell density, and it slows with age. Alpha is the speed to watch.
For memory specifically, the hot spots are the foundational ones. C4 SMR regulates sleep broadly. C3 stabilizes vigilance, which you need for acquisition. The SMR neurofeedback literature is most relevant here, and on the cognitive side, controlled studies of SMR training report working-memory and other cognitive gains (Vernon et al., 2003). In healthy participants, the classic SMR and beta training protocols have been associated with improvements in attention and memory performance. Train executive function, sleep, and anxiety, and the IQ and the memory tend to come along.
One of my standing protocols is "left, right, and center": a left-side beta protocol in the upper teens, then a right-side SMR protocol lower, then a slightly softer vertex protocol to act as a center of gravity and soften any activation the left-beta caused. Each site produces a homotopic resonance at its mirror partner, so you can balance the supervisory and stabilizing roles of the central strip.
A note of caution that ran through the whole session. I work as a coach rather than in a treatment role, so my tolerance for side effects is very low. I aim for gently positive effects and move things slowly over months rather than plowing through discomfort. Two specific warnings:
- Alpha responds to training with rebound. Train it one direction and it pushes back. The cleaner move is to get high theta or high beta out of the way and let a low alpha rise on its own. If you want to raise alpha speed directly, the move is eyes-closed posterior training at PZ, the big posterior alpha generator, sometimes paired with frontal beta if the goal is word-finding or mood.
- Delta is off-limits for band training. Manipulating below 4 Hz puts you near seizure and ictal-spike territory and near resting-mode and metabolic systems. Reward coherent delta and you risk triggering a seizure in a susceptible person. Keep band training between roughly 3 to 4 Hz and 38 Hz. Below that, weirdness abounds. Above 38 Hz with passive electrodes you are mostly measuring noise.
If you are new to this and wondering whether the method holds up, I have written a research overview at Is Neurofeedback Legitimate?.
Which nootropics actually help memory?
A few compounds have a reasonable case.
- Citicoline (CDP-choline). Evidence supports improved myelination and remyelination after oxidative stress, which speeds up the neurons themselves. It has been used for years as an aging medication and now runs off-label as a study aid. Mildly stimulating, brings alpha speed up, improves short-term recall and retrieval, easy to find.
- Creatine. Growing evidence of improved brain function, and a low-risk option (Rae et al., 2003).
- Vasopressin (desmopressin/DDAVP). I mention this for completeness, not as a recommendation. It is a neuropeptide, the antidiuretic hormone, available synthetically and not a controlled substance. Sales reps in the 1970s and 80s used the old inhaler form to boost short-term memory, retrieval, and focus, and the effect is notable. The cost is bloating and water retention, and mucking about with your neuropeptides is not something I would do casually. Skip it.
- Racetams can help but have become hard to source.
Meditate and do neurofeedback first, then add a benign nootropic like citicoline or creatine if you want to layer something on top.
What behavioral and metabolic biohacks move memory?
Meditation. Concentration practices can shift both processing speed and cortical thickness with aging. Regular practice is associated with thicker prefrontal cortex and stronger regulation (Lazar et al., 2005). The mechanism overlaps heavily with attention training. More on that in Biohacking Meditation.
Photobiomodulation. Red and near-infrared light helmets around 1070 nm appear to feed mitochondria. The full mechanism is still being worked out, and the research describes lifts in alertness and short-term memory after transcranial near-infrared exposure (Barrett & Gonzalez-Lima, 2013). I cover the details in brain biohacking with photobiomodulation.
Metabolic interventions for injury and fog. For traumatic brain injury, post-COVID, post-mold, post-Lyme fog, or chronic stress, I like to layer a metabolic tool on top of EEG neurofeedback: hyperbaric oxygen, photobiomodulation helmets, infrared blood-flow training (passive HEG), or hormetic stressors like sauna and cold. Many mild-TBI cases involve an accumulation of small concussions over a career of sport or clumsiness, so the picture is accumulated wear and tear rather than a single focal deficit. That kind of diffuse load responds well to a stacked approach.
A note on the gadget that gets the most questions: binaural beats do not work as advertised. Dozens of papers have failed to find a driving effect, and my own double-blind placebo-controlled work at UCLA found no brain signature. The human brainstem lacks the frequency-following response that auditory entrainment would require. Photic (light) driving is a different story and can entrain cells, and 40 Hz visual flicker has real research behind it (Iaccarino et al., 2016). Treat binaural beats as a structured anchor for meditation, not as a box that installs frequencies in your head.
How do you build a memory strategy that actually works for you?
Iterate and find your own lever. Sleep deeply, run a fast and smooth mind, control what you load into working memory, and most memory complaints resolve without ever touching the memory structures. If the limit is true episodic memory, the foundation still comes first, then metabolic and targeted work.
Get individualized. Use a brain map to read your alpha speed and band power, use HRV during sleep, use methylation and blood testing if aging is on the table, and narrow in on what moves the needle for your brain. A QEEG is the first concrete step, and you can build everything from there.