The Engineer as an Athlete

Here is a finding from the neuroscience literature that I have not been able to stop thinking about since I first encountered it: people who speak more than one language develop the clinical symptoms of Alzheimer's disease roughly four to five years later than monolinguals — even when the underlying brain pathology is identical. The plaques are there. The neural damage has accumulated. The disease is progressing on schedule inside the skull. And yet something built up over a lifetime of cognitive demand holds the symptoms at bay for half a decade.

In a disease for which there is currently no cure, and no reliable means of reversal, this is not a marginal finding. It is an enormous one. What the research points to is something called cognitive reserve — the brain's developed capacity to compensate for damage through alternative neural pathways, routes constructed not by medication but by the accumulated demands of a mentally active life. The brain can be trained to absorb punishment that would otherwise be disabling. The question is whether you are training it, or spending it down.

I am a software engineer. I have been one for more than twenty years, and I have led engineering teams for most of that time. The industry I grew up in has a specific and well-developed answer to that question, and the answer is: spend it down, as fast as necessary, in the service of shipping. Run on caffeine. Compress sleep. Treat the weekend as unallocated capacity. Wear the grind as a credential — the visible suffering that signals, to the people who evaluate you, that you are serious.

There is an irony embedded in this moment that the industry has not yet found the language to acknowledge. The same technology sector that built and still sustains this culture of systematic self-depletion is also betting its entire future on the proposition that human judgment — the capacity to ask the right question, catch the wrong assumption, make the architectural call under uncertainty — is the one thing artificial intelligence cannot replace. The machines, the argument goes, will handle production. What remains irreplaceably human is the quality of the thinking that directs them. And yet the conditions under which that thinking is expected to occur — the sleep-deprived, caffeine-dependent, physically sedentary conditions of the professional norm — are precisely the conditions under which the research says thinking degrades most reliably.

This is not a productivity argument. It is a structural one. And it begins, as most structural arguments do, with what you are willing to stop.

Subtractions

Over the years, I had normalized three substances — smoking, alcohol, and eventually caffeine — as minor habits with manageable trade-offs. The research corrected my perspective on each of them.

Smoking I gave up decades ago, and I will not spend much space on it. The neurological case is unambiguous: chronic smoking accelerates cognitive decline and is independently associated with increased risk of dementia, including Alzheimer's disease, a conclusion drawn not from isolated studies but from the accumulated weight of longitudinal research across large populations.¹ It was the clearest subtraction I ever made.

Alcohol required more honesty. For years I held the comfortable belief, shared by much of the professional world, that moderate drinking was at worst a benign pleasure — perhaps even mildly cardioprotective, as older studies had suggested. The neuroimaging research that emerged in the early 2020s made that position very difficult to sustain. A 2022 study published in Nature Communications, analyzing MRI brain scans from more than thirty-six thousand participants in the UK Biobank, found that even light-to-moderate consumption — as little as one to two drinks per day — was associated with measurable reductions in both gray and white matter brain volume.² A companion study of twenty-five thousand participants, published in NeuroImage Clinical, reached a harder conclusion still: no level of alcohol consumption was identified as safe when examining its relationship with brain structure and function.³

The glass of wine at the end of a hard day — the ritual that signals the work is over, that the pressure has lifted, that you have earned some relief — is, by the best current neuroimaging evidence, measurably shrinking the organ you spent all day relying on. The team dinner toast, the client drinks, the industry conference bar tab: these are not neutral social lubricants. They are, in aggregate, a tax on the very cognitive capacity that the industry simultaneously claims to value above all else. The irony would be almost elegant if the stakes were not so high.

Caffeine was the hardest, and the most counterintuitive to let go of. At my worst, I was consuming nine to twelve espresso shots daily to sustain my output. The short-term productivity felt real — and in a narrow sense, it was. What I had not honestly accounted for was the full cost. A double-blind, randomized crossover study published in the Journal of Clinical Sleep Medicine found that caffeine consumed even six hours before bedtime significantly reduced total sleep time compared to a placebo.⁴ This matters because sleep is not, as the engineering culture tends to treat it, a variable to compress when deadlines press. It is when the brain consolidates memory, clears the metabolic waste generated by a day of thinking, and restores the cognitive resources that the following day will require. Anything that reliably compromises that process does not disappear from the ledger — it rolls forward, with interest.

After pushing through a withdrawal period I will describe, charitably, as difficult, what I found on the other side was immediate and substantial: deeper sleep, more consistent energy, clearer thinking in precisely the hours I had previously lost to the post-caffeine drop. The science supports what I experienced. The subtraction was worth it.

Fertilizer

The relationship between physical exercise and brain function is among the most robustly established findings in modern neuroscience — and also among the most consistently undervalued by exactly the people whose work depends most on it. Two mechanisms are directly relevant to the demands of sustained cognitive work.

The first involves a protein called BDNF — brain-derived neurotrophic factor — which aerobic exercise elevates consistently in key brain regions, particularly the hippocampus and prefrontal cortex. BDNF acts as a biological growth factor for neurons: it supports the generation of new neurons, strengthens synaptic connections, and is sometimes called "fertilizer for the brain" for exactly that reason — it is, quite literally, the molecular substrate of learning. A meta-analysis of twenty-nine studies involving more than eleven hundred participants found a moderate, statistically significant increase in BDNF following a single exercise session, with sustained elevations following regular aerobic training.⁵

The second mechanism is structural. A 2024 meta-analysis of randomized controlled trials found that aerobic exercise training produced measurable increases in hippocampal volume in healthy older adults⁶ — increases in the physical size of the region most critical for forming new memories and adapting to novel problems. Physical growth, in the organ responsible for learning, driven by the act of moving your body. For anyone in a field where continuous learning is not optional but definitional, this is not an abstraction. It is, quite literally, part of the job description.

I go to the gym five times a week. I have for years. The sessions are non-negotiable, and the framing I use to defend them to myself has nothing to do with how I look — it is about the only piece of equipment I cannot replace, which happens to be biological.

The Cheapest Thing on the Shelf

The exercise science is, by the standards of this field, unusually clean: large samples, randomized trials, consistent replications across populations. The supplementation literature is the opposite, and it is worth understanding that difference before spending any money on it.

The global brain health supplement market currently generates roughly nine to ten billion dollars in annual revenue and is projected to approach eighteen billion dollars by 2030.⁷ It is also, by the frank assessment of the peer-reviewed literature, largely selling products whose cognitive benefits range from unproven to actively misleading. Nootropic stacks, lion's mane mushroom capsules, phosphatidylserine formulations, racetams — the shelves are crowded with compounds marketed for mental performance whose evidentiary basis, when examined by someone without something to sell, tends to dissolve.

The one compound in my regimen I can discuss with any real confidence is, by some distance, the cheapest and least glamorous thing on offer: creatine monohydrate, typically sold in a plain tub for a few dollars per month. Most people know it as a muscle supplement. Its mechanism in the brain is identical — it increases the availability of phosphocreatine, which supports ATP regeneration in high-demand cells. The brain accounts for roughly two percent of the body's mass and consumes approximately twenty percent of its energy at rest; neurons are not passive processors.⁸

A 2023 systematic review and meta-analysis of ten randomized controlled trials, published in Nutrition Reviews, found that creatine supplementation significantly improved memory performance in healthy individuals, with the most pronounced effects in older adults.⁹ A 2024 meta-analysis of sixteen trials confirmed positive effects on memory and information processing speed, while finding no significant improvement in overall cognitive function or executive function, and noting that the certainty of evidence ranges from moderate to low across domains.¹⁰

The honest picture: real, if modest, benefits for memory and processing speed, particularly under conditions of cognitive load. It is not a performance drug. It is nutritional support for a brain doing demanding work. That the one compound with meaningful research support is also the cheapest, most generic thing on the shelf — while the aggressively marketed, expensively branded stacks are the ones the evidence doesn't hold up — tells you everything you need to know about the relationship between the supplement industry and the science it claims to represent.

Reserve

I did not understand, until I read the bilingualism research in any depth, what I had been accidentally doing. I grew up in two languages — English and Korean — and spent portions of my adult life building working fluency in Japanese and Thai. I thought of none of this as maintenance. It was circumstance, then curiosity, then habit. The research reframed it as something more deliberate-sounding than it was: evidence that decades of navigating multiple linguistic systems had been building cognitive reserve without my conscious participation.

The mechanism is well-documented. Active use of more than one language requires constant switching between frameworks, suppressing one system while activating another, maintaining parallel grammars in a state of ongoing competition. That executive demand directly strengthens the prefrontal circuits responsible for inhibitory control, working memory, and task-switching¹¹ — which are, not incidentally, the precise capacities that distinguish a senior engineer from a fast typist.

The long-term protection this builds is striking — and, it should be said, still actively contested in the scientific literature, where retrospective studies show stronger effects than prospective ones and methodological questions remain open.¹² What the evidence does consistently support, across multiple populations and research designs, is a meaningful association: bilingual individuals appear to develop the clinical symptoms of Alzheimer's disease later than monolinguals with equivalent pathology, with the protective effect on age of symptom onset confirmed by meta-analysis. A 2024 neuroimaging study from Concordia University pressed the finding further: bilingual individuals with Alzheimer's had measurably larger hippocampi than matched monolinguals at equivalent stages of the disease — more brain matter in the region the disease most reliably destroys, even as the disease progressed.¹³

I am now working on French, more deliberately, with that understanding in place. The goal is not fluency as a credential. It is to keep the executive control systems under sustained, effortful demand — to keep building redundancy into an architecture I cannot afford to lose. The disease does not care whether you had time to prepare. The only question is whether, when it arrives, there is something waiting for it.

What the Individual Decides

The technology industry has decided that human judgment cannot be automated. What it has not decided — and may not decide, at least not quickly — is whether to protect the conditions under which good judgment is actually possible. The engineering culture that rewards visible suffering does not dismantle itself because the underlying economics have shifted. It persists by inertia, and by the fact that grinding, performed publicly, still functions as a credential.

This means the calculation falls to the individual. Which is, in the end, clarifying.

No serious professional athlete is asked to perform at the highest level while subsisting on disrupted sleep, chemical stimulants, and deliberate neglect of the body that performs. The instrument is the job. Its maintenance is not optional. What the neuroscience is saying, across all of these domains, is that the same logic applies to the cognitive worker — and has always applied, even when the economics didn't make it visible. The machines have simply made it impossible to ignore.

The junior engineer who came in that third consecutive Saturday morning — running on caffeine, proud of the grind, performing commitment for an audience that had taught him this was what commitment looked like — eventually asked me what I actually did differently. I told him. He listened with the careful skepticism of someone who was not yet ready to believe that the grind might be the problem rather than the solution.

He will make his own choice, and make it without waiting for anyone's permission. That is the only way it gets made.

I recognized the look because I wore it for years. What changed was the moment I understood that the brain is not something you use to do the work — it is the work. Everything else follows from whether you treat it that way.