This piece comes from one of my Monday night livestreams, where I work through new research and take questions from whoever shows up. Two threads caught my attention this week: a large study linking music habits to lower dementia risk, and a new EEG analysis that can separate dementia types with surprising accuracy. I've anonymized the audience questions. Everything here is education, not medical advice. Talk to your own clinician before changing anything.
Does Listening to Music Lower Dementia Risk?
A large cohort study by Jaffa and colleagues (2025, International Journal of Geriatric Psychiatry) followed more than 10,000 Australian adults over age 70 for several years. Everyone was dementia-free at baseline. The researchers grouped people by how much they engaged with music and tracked who developed dementia over time.
People who always listened to music had a 39% lower chance of developing dementia compared to those who listened rarely or never (Jaffa et al., 2025). People who frequently played an instrument had a 35% lower risk. Doing both landed around 33%. For mild cognitive impairment (MCI), often an early stage that sometimes progresses to dementia, the reduction was around 17 to 22%.
My first thought when I saw the abstract was selection bias. Does developing dementia make you less interested in music, which would flip the causal story? The research picture argues against that. People with advanced Alzheimer's still respond to music they know well. Lifelong musicians often retain the ability to play even when other memory has eroded. Music is one of the most preserved capacities in the dementing brain. So the correlation here is more likely pointing at something protective, not something that simply drops off as illness sets in.
This is correlation, not causation. I want to be honest about that. But the effect size is large enough to take seriously.
Why Would Music Build Cognitive Reserve?
Music is one of the very few things that demands both hemispheres at once. Each cochlea wires to both sides of the brain through the brainstem and the superior olivary nuclei, which compare the timing of sound arriving at each ear so you can localize where a sound comes from. Your ears never stop transducing. The auditory system is expensive and bilateral by design.
Layer language on top of that. The meaning of speech leans on left temporal and parietal tissue around Wernicke's area. The prosody, the lilt, the emotional read of a voice leans right. Listening to a nostalgic song recruits the hippocampus for memory, auditory cortex for sound, limbic regions for emotion, and frontal regions for attention and motivation. Playing an instrument adds motor planning, motor patterning, and motor learning to the stack.
That whole-brain recruitment is probably why music shows up as a cognitive reserve factor, similar to what we see with meditation and education. Think of reserve the way you think of metabolic reserve in the body. Muscle mass and bone density laid down earlier protect you against the wear of aging later. Music looks like it deposits something similar into the brain.
The study also found an interaction worth flagging. Among people with 16 or more years of education, those who always listened to music had a 63% risk reduction (Jaffa et al., 2025). That's not just education plus music added together. It's a statistical interaction, meaning the combination did more than the sum of its parts. The mechanism is open. My best guess is that both factors drive plasticity and information flow, and they compound when they overlap.
There's a second reason to start making music: you may keep it. I once watched a frail, elderly West African drummer get helped onto a stage by two men holding his elbows. He was very ill. They strapped a large drum on him and he fidgeted with it. Then he started playing, and he transformed into someone decades younger, filling a resonant hall with a single instrument and no microphone. He walked off with a bounce that wasn't there when he walked on. If you don't already make music, learning is one of the better investments you can make for a brain you want to keep.
Can EEG Detect Dementia Type Early?
The second thread shifts from playlists to brainwaves. How you treat dementia depends on which disease is driving it, and the diseases differ at the level of tissue and neurotransmitter:
- Alzheimer's is medial temporal: hippocampus and parahippocampus, cholinergic tissue. Episodic memory goes first.
- Frontotemporal dementia (FTD) is more vascular and frontal, with more mood and impulsivity in the presentation.
- Lewy body dementia (LBD) is parkinsonian and dopaminergic, with hallucinations and fluctuating cognition.
You modulate acetylcholine for one and dopamine for another. So distinguishing them early matters, and it matters more now than it did fifteen years ago. Back then we had palliative, symptom-modifying treatments and nothing that changed the trajectory of decline. Now we're starting to see trajectory-modifying treatments that flatten the slope. That buys healthy years. Gerontologists call the goal compression of morbidity: pushing loss of function as late as possible so you decline quickly at the very end rather than spending decades fading.
One caution on diagnosis. Elders frequently land in the ER and get mislabeled as demented when they actually have delirium, often from a UTI with fever and secondary inflammation. Delirium causes short-term confusion and sometimes hallucinations, and it resolves with hydration and treating the illness. True dementia must have an episodic memory deficit as a core feature plus at least one other symptom, commonly agitation, suspicion, or confabulation. Word-finding trouble and tip-of-the-tongue moments are timing, speed, and retrieval problems, not the hippocampal encoding failure that defines a dementia.
What Do the EEG Brainwave Fingerprints Look Like?
EEG has a long history in epilepsy, sleep, and QEEG-based neurofeedback. The newer work uses it to flag and separate dementia types early. Research on the posterior dominant rhythm, the eyes-closed alpha rhythm, finds it behaves differently by disease:
- Healthy brains and early Alzheimer's keep a well-preserved background alpha frequency.
- Lewy body dementia can lose that alpha rhythm almost entirely. The brain's idling, synchronizing tone drops into the static.
- Alzheimer's and FTD both show a slowed dominant rhythm, but alpha power is weaker in Alzheimer's and better preserved in FTD. That fits the physiology: Alzheimer's involves heavy medial temporal tissue loss by the time it's symptomatic, while FTD is more vascular and frontal.
A 2025 study by Jungrungrueang and colleagues in Scientific Reports pushed this further with EEG connectome analysis, looking at connectivity and phase relationships rather than power alone. Alpha-band connectivity around 8 to 12 Hz was weaker across all dementias, but each type carried its own signature (Jungrungrueang et al., 2025). Alzheimer's showed an odd surge in delta-band connectivity: low power, high connectivity at the same time. That's a clean example of power and coherence moving independently. They also saw hypersynchronization at slow frequencies and more complex, chaotic activity at fast ones.
Fed into a convolutional neural network, these patterns classified the dementia types with 93.6% overall accuracy, and head-to-head comparisons like Alzheimer's versus FTD ran over 97% (Jungrungrueang et al., 2025). You can already see those distinctions on a PET scan, but a PET scan means tracers and an enclosure, and the magnet alone can cost over a thousand dollars an hour to run. EEG means a cap on the head for twenty minutes. That's the practical promise here.
Where Does Neurofeedback Fit, and Where Doesn't It?
A reasonable question from the chat: can neurofeedback prevent dementia? Prevention is too strong a claim. Offsetting and supporting is closer to what the research suggests and what I've seen in the brain maps.
A related finding gives some texture. Work out of NYU's memory clinic followed elders with memory complaints over several years. People who presented with high global theta/beta ratios were more likely to progress from MCI into dementia. People without that signature were less likely (Prichep et al., 2006). Could you train the theta/beta ratio down and shift the risk? Maybe. I won't oversell it. It's a reasonable hypothesis, not a settled result.
Here's how I think about the layers:
- Before symptoms, training plasticity makes sense. Blood-flow training in someone with a lot of concussion history may raise vascular tone and support plasticity, which I'd expect to reduce vulnerability over time. This is research-based reasoning, not a proven prevention protocol.
- Once symptomatic, neurofeedback may help with specific features: anger, irritability, anxiety, sleep, executive function. I've seen the brain maps and heard reports of some relief in parkinsonian and Alzheimer's-type presentations as part of a larger strategy. What it won't do is reverse aggressive, ongoing tissue loss.
- The first move I'd point to is metabolic. Pull the fuel off the fire. A program like Apollo Health's ReCODE (or PreCODE for younger people) screens the many factors that drive metabolic aging, then targets them with B vitamins, hormones, omega-3s, and diet changes tuned to your genes and metabolism. Neurofeedback runs alongside that, not instead of it.
One more metabolic note. Alzheimer's involves something like insulin resistance in neurons, sometimes called type 3 diabetes: the cells become resistant to using glucose and undergo oxidative stress while starving amid plenty of fuel (de la Monte & Wands, 2008). The ketone pathway tends to stay spared. For an elder who sundowns and gets disoriented in the late afternoon, exogenous ketones (ketone esters) instead of the craved carb snack can sometimes produce a cognitive brightening. If it works, that's a signal to do a full metabolic screen with a good functional medicine doctor. Ask the geriatrician first. In younger people, I'd rather you train your body to make its own ketones through strategic fasting than rush to exogenous supplements, since throwing in fuel you don't need can stack with cortisol surges and feel terrible.
What Should You Actually Do?
The longevity advice is the same, just earlier. Build reserve in your 50s and 60s, because rebuilding it in your 70s and 80s is much harder. Put on muscle mass and keep bone density high. Protect heart, lung, and oral health. Meditate, where the accumulated daily dose correlates with less cortical thinning in frontotemporal tissue and the insula (Lazar et al., 2005). And learn to make music, then go move to it.
If you want a read on your own rhythms, a QEEG won't tell you why your brain is doing what it's doing, and it won't diagnose Alzheimer's. It will show you whether your alpha is slowing, whether your theta/beta ratios are running high, and where your tissue looks fatigued. Those features can be worked on regardless of cause. If you're a clinician who wants help interpreting an EEG at a high level, I'm glad to support that from the research and mapping side, though true diagnostic work on spikes or ictal events belongs with an epileptologist or EEG-based neurologist.
The encouraging part of this week's research is timing. We now have ways to spot early decline while there's still room to act, and the actions that help are ones you can start tonight. Pick an instrument, set a meditation timer, and book a metabolic screen if aging is on your mind.