My father spent twenty-eight years optimizing chemical plant processes down to the flow-rate decimal. He kept every adjustment in composition notebooks still stacked in his basement. When I studied Chemical Engineering, he quizzed me over dinner and once corrected a textbook's heat transfer coefficient at the table. I followed him into the discipline, spent years in pharma manufacturing before I started writing about this stuff, and the family habit of running numbers at the dinner table never quite left me. The man does not estimate. He calculates.
Last month at the assisted living facility outside Detroit, I watched him reach for a word. A precise technical word, the kind he used to retrieve the way a vending machine drops a can. It did not come. He laughed it off and moved on. I sat in my car afterward and did not move for a while.
That night I opened PubMed. Not a brain-health listicle. The actual papers. Carl Cotman on BDNF. The Nedergaard lab on glymphatic clearance. SPRINT MIND. FINGER. Watching someone who thinks in process variables lose a word is a different problem than watching someone who never tracked variables lose one. I wanted the spec sheet.
The Spec Sheet Nobody Gives You
Most brain-health writing is a list of behaviors: exercise, sleep, eat well, stay social. The behaviors are correct. The problem is that nobody explains the underlying mechanism, which means you cannot tell which inputs actually matter and which are noise. In process engineering you would call that running a control loop without understanding the chemistry.
The 2024 Lancet Commission on dementia estimated that 45% of dementia cases are potentially preventable through modifiable risk factors. That is the calibration number — it tells you the system is modifiable. My colleague Eleanor Hayes has covered the full Lancet checklist elsewhere on this site, and she has also written the practical guide for telling normal-aging forgetfulness from the early signals of dementia — the clinical sorting question this piece deliberately does not try to answer. I want to go orthogonal. Not what the risk factors are. Why the underlying biology makes those risk factors real.
BDNF — The Growth Factor Your Workout Produces
Brain-derived neurotrophic factor is the cleanest mechanism in the whole literature. BDNF is a signaling protein that promotes the growth and survival of neurons, strengthens synaptic connections, and supports neurogenesis in the hippocampus — the region that takes the first hit in Alzheimer's. As we age, hippocampal volume shrinks. BDNF is the molecular signal that pushes back.
The most reliable way to produce BDNF is aerobic exercise. Carl Cotman's lab at UC Irvine has been working on this since the late 1990s. Cotman and Berchtold's foundational 2002 paper in Trends in Neurosciences, and their follow-up 2007 review with Christie, established the exercise-BDNF axis as one of the best-supported mechanisms in cognitive aging. The dose-response data points to roughly 150 minutes of moderate aerobic exercise per week — brisk walking, cycling, swimming — as the floor for measurable effect. Resistance training contributes, but aerobic is the heavier lever for BDNF specifically.
Here is the engineer's read. BDNF is the system's reinvestment signal. It tells neurons: this infrastructure is being used, allocate maintenance budget. When you do not move, the signal goes quiet, and the system reads that as permission to downscale. The trade-off is that the signal does not bank. You cannot do five hours of cardio on Saturday and coast for two weeks. The dosing is daily-ish, and consistency beats intensity every time.
I restarted running a couple of years ago. First run was 0.8 miles before my lungs filed a formal complaint. I am not the master of this yet. The goal is not to look impressive. The goal is to keep the signal alive.
The same lever matters past the point of prevention. Aerobic and resistance exercise is one of the few interventions with real evidence for slowing the progression of Parkinson's disease, not just lowering the risk of cognitive decline. Movement is medicine in both directions.
The Glymphatic System — Why Your Brain Cleans Itself at Night
This is the part of the literature that surprised me most, and I want to spend time on it because most people have never heard of it.
In 2013, Maiken Nedergaard's lab at the University of Rochester published a paper in Science showing that during sleep, the space between brain cells expands by roughly 60% and cerebrospinal fluid flushes through the tissue, clearing out metabolic waste — including amyloid-beta and tau, the two proteins implicated in Alzheimer's. They named it the glymphatic system, because it functions like the lymphatic system but uses glial cells to do the plumbing. The clearance runs primarily during slow-wave (deep) sleep. It is dramatically less active when you are awake.
Which is to say: sleep is not optional rest. It is a scheduled maintenance cycle. In a chemical plant you do not run a reactor twenty-four hours a day without filter changes and cleaning windows. The plant accumulates fouling and the throughput drops. The brain is the same. Short-sleep nights are the equivalent of skipping the maintenance shutdown. Do it occasionally and the system recovers. Do it for years and the residue builds.
The target is 7–9 hours, and the architecture matters as much as the duration. Deep sleep — the slow-wave portion that dominates the first half of the night — is when the clearance runs hardest. Alcohol close to bedtime, late screens, and irregular sleep timing all degrade slow-wave sleep specifically. A 2015 paper from the same lab (Lee et al., Journal of Neuroscience) reported that lateral sleeping position — on your side — appears to optimize glymphatic flow more than supine or prone. Animal data, with a sample size that says "plausible mechanism," not "settled clinical recommendation."
If I had to pick the single highest-leverage input on this list, sleep architecture would be it. Granted, it is also the hardest to fix once it has drifted. Which gets me to the next mechanism, because the things that wreck sleep also wreck the pipes.
Cerebrovascular Health — The Pipes Matter
The brain is roughly 2% of body weight and consumes about 20% of the oxygen you breathe. It stores almost nothing. It runs entirely on continuous blood delivery, which makes it uniquely sensitive to vascular health. Damage the pipes and the reactor starves.
Hypertension is the biggest single offender. Chronic high blood pressure damages small vessels in the brain — the microvasculature — and small vessel disease is a leading cause of vascular dementia and probably contributes to Alzheimer's pathology too. The SPRINT MIND trial, published in JAMA in 2019, randomized older adults to either standard blood pressure control (systolic under 140) or intensive control (under 120). The intensive group had a 19% reduction in mild cognitive impairment over the follow-up period. That is a large effect from a fairly modest intervention.
If you take one practical thing from this article, it is probably this: track your blood pressure at home. Not at the doctor's office where white-coat effects distort the reading — in the morning, sitting, after a few minutes of rest. I wrote a comparison of the blood pressure monitors worth using at home a while back. The point is the trend, not a single reading. My father's history is why I started doing this in my forties.
And frankly, the cardiologist's playbook for blood pressure — Mediterranean-pattern eating, aerobic exercise, sleep, weight management, limit alcohol — is the same playbook for the brain. The pipes and the reactor share a maintenance schedule.
Glucose, Insulin, and the Type-3-Diabetes Hypothesis
One of the more striking shifts in Alzheimer's research over the past fifteen years is how central glucose metabolism has become. The brain is a glucose-hungry organ, and insulin resistance — the same problem that drives type 2 diabetes — appears to play out inside the brain too, independent of the rest of the body. Suzanne de la Monte at Brown University and others have informally called late-stage Alzheimer's "type 3 diabetes" because of the overlap: insulin resistance in brain tissue, reduced glucose uptake by neurons, mitochondrial dysfunction. The framing is not fully settled as a clinical category, but the metabolic link is no longer controversial.
The engineer's read: glucose is the process input, and insulin is the receptor that lets the input land. If the receptor is degraded, it does not matter how much substrate you pump in — the cell is starving in the middle of a sugar bath. Midlife metabolic health (the decades when nothing feels wrong) shows up with a long lag in brain function.
The practical implications track what you would expect: tight glycemic control, Mediterranean-pattern eating, exercise (which improves insulin sensitivity independently of weight), and avoiding the blood-sugar roller coaster of ultra-processed food. The GLP-1 receptor agonists — semaglutide, tirzepatide — are the newer wrinkle. Early observational data and small trials suggest possible cognitive benefits beyond glucose lowering, but the EVOKE and EVOKE+ trials in early Alzheimer's are still ongoing. The hypothesis is plausible. The proof is not in yet. Eleanor's article on GLP-1 medications and what they actually do to your body is the practical companion read.
I know how it sounds, but I will say it anyway: glucose discipline in your fifties is brain insurance in your seventies.
Neuroinflammation — The Slow Fire
If BDNF is the growth signal and the glymphatic system is the cleanup crew, neuroinflammation is the slow fire that grinds the system down. Chronic low-grade inflammation is increasingly understood as a driver of Alzheimer's pathology, not a byproduct of it. The mechanism involves microglial activation, cytokine signaling, and an inflammasome called NLRP3 — but the practical inputs are the familiar ones: obesity, insulin resistance, poor sleep, sedentary lifestyle, ultra-processed diet.
This is where the Mediterranean diet keeps showing up in the literature, and the proposed mechanism is not really about antioxidants in the loose pop-nutrition sense. It is about systemic inflammation. A 2024 meta-analysis on Mediterranean-pattern eating and cognitive aging confirmed a consistent association with reduced cognitive impairment risk — a finding that has now shown up across enough cohorts and meta-analyses that it is one of the steadiest signals in the diet-and-brain literature. Olive oil, fish, legumes, nuts, vegetables, modest wine. Not glamorous. Just lower inflammatory load.
The engineer's frame: inflammation is what happens when the reactor runs on contaminated feedstock. You can tune the reactor perfectly, but if the input is dirty, the output degrades. I have written before about building a senior-friendly Mediterranean rhythm of eating, so I will not rehash it here.
Cognitive Reserve — Why Novelty Beats Repetition
The last mechanism is the one I find most interesting because it inverts the others. Everything above is about protecting the substrate. Cognitive reserve is about building redundancy on top of it.
Neurons and synaptic connections you build through learning create alternate pathways to the same cognitive function. When one pathway degrades, reserve allows the brain to route through another. In engineering terms, this is standard practice — you do not run a critical process on a single pipe. The autopsy literature shows that people with high cognitive reserve sometimes have substantial Alzheimer's pathology in their brains and yet showed no clinical symptoms in life. The redundancy carried the load.
The load-bearing variable is novelty. Repeating mastered activities maintains the connections you already have. Learning something genuinely new — a language, an instrument, a craft you have not done before — builds new ones. The ACTIVE study, the largest randomized cognitive training trial in older adults, found that speed-of-processing training produced durable benefits years later. Eleanor has written more about what the science actually says about brain training, and Victoria's piece on why your hands might be your brain's best workout covers the Mayo Clinic data on knitting and pottery.
My own read: do not get attached to the specific activity. What counts is whether the brain is being asked to build something it does not already know how to do. Sudoku does not stop being useful, but if you have done sudoku for fifteen years, it is maintenance, not construction. Pick something that frustrates you a little.
What I Actually Do (and What I Don't)
In any case, here is the ground-level honest version. Not prescriptions. My own operating parameters.
I prioritize sleep architecture above everything else — 7 hours as a non-negotiable floor, closer to 8 when I can. I shut down screens earlier than I used to. I track my blood pressure at home — partly because of my father's history, partly because SPRINT MIND moved me. I run consistently rather than intensely, because the BDNF data says consistency is the variable, and frankly my back will not let me push intensity anyway. Mediterranean-pattern eating about 80% of the week. The other 20% includes Chickenjoy with my parents, and I am not negotiating on that.
I am skeptical of most brain-health supplements. Omega-3 trial data is mixed at best. Vitamin D evidence is mostly correlational. Phosphatidylserine claims are thin. Ginkgo did not pan out in the large trials. I take a multivitamin and call it good. The money I do not spend on supplements goes into produce and shoes.
The ADHD aside, because someone always asks: I take stimulant medication for ADHD, and the literature on long-term cognitive effects of stimulants in midlife is genuinely uncertain. I do not assume my personal experience generalizes.
The honest closer: I do not know if I am doing enough. The process-engineer in me hates not having clear output metrics. The brain is not a spectrometer — you cannot run a purity check on your hippocampal reserve. But you can track the inputs. Sleep. Pressure. Movement. Glucose. Novelty. Inflammation. That is what I do.
If I had to recommend one thing to start this week, it would not be a supplement or an app. It would be scheduling the preventive checkups worth scheduling you have been putting off, and buying a decent home blood pressure cuff. Twenty minutes of work. Decades of leverage.
