What caffeine actually does

Caffeine is the most widely used stimulant in the world and the one lever almost everyone reading this is already pulling. What it does is narrower and more interesting than the folklore.

The mechanism, in one paragraph

While you are awake, a molecule called adenosine accumulates in the brain. As it builds up and binds to its receptors, you feel progressively sleepier. Sleep clears it.

Caffeine is shaped enough like adenosine to occupy those receptors without activating them. It is an antagonist: it sits in the lock without turning it, so the sleepiness signal cannot be delivered.

That single fact explains almost everything worth knowing.

What follows from it

It masks tiredness rather than removing it. The adenosine keeps accumulating while its receptors are blocked. When the caffeine clears, it all binds at once — which is why the drop can feel abrupt rather than gradual.

It does more for a tired person than a rested one. If there is little adenosine to block, there is little to gain. This is why studies on sleep-deprived participants show large effects and studies on well-rested ones show modest ones, and why your own results here will show a bigger caffeine effect on a bad week than a good one.

It cannot substitute for sleep. Everything sleep does besides removing adenosine — memory consolidation, clearance of metabolic waste, the rest of it — carries on not happening.

The half-life, and why the afternoon coffee costs you

Caffeine’s half-life in a healthy adult is roughly five to six hours. That is the single most useful number on this page.

A 4pm coffee leaves about half its caffeine in you at bedtime. It does not necessarily stop you falling asleep — but it measurably reduces deep sleep, which means you wake with more adenosine debt, which means tomorrow needs more caffeine. That loop is how most heavy consumption starts.

The half-life varies a lot between people, largely down to a liver enzyme whose activity is partly genetic. Some people clear caffeine in three hours and some in nine, which is why “coffee doesn’t affect my sleep” and “one cup at lunch and I’m awake at 2am” are both true statements about different people.

Tolerance changes the transaction

With regular use, the brain responds to blocked receptors by making more of them. The result is that the same dose does less, and — importantly — that your baseline shifts.

For a habitual heavy user, caffeine largely restores normal function rather than lifting you above it. Skipping it produces withdrawal: headache, irritability and genuinely impaired performance for a day or two. That is not evidence the caffeine was helping; it is evidence the absence is hurting.

The practical implication is not that you should stop. It is that the benefit is largest when use is occasional and targeted — spent deliberately on your worst hours rather than poured continuously from nine in the morning.

What it does to each thing this site measures

Reaction time improves, and this is the best-supported effect. Expect it to be clearest when you are tired.

Sustained attention improves, which is most of why it feels effective. Vigilance tasks — staying alert with little happening — are where it does the most good.

The Stroop interference cost is a more complicated case. Caffeine helps the speed of both conditions, so raw times fall, but the difference between them — which is the actual effect being measured — moves less. A faster overall run with a similar interference cost is the expected pattern.

Memory shows little consistent benefit. Caffeine does not appear to improve encoding or recall in a rested person to any useful degree.

Fine motor control can get slightly worse at higher doses. If your aim trainer accuracy drops while your speed rises, that is a plausible reason.

Dose, and where the returns stop

The benefits plateau early. Most of the effect on alertness arrives somewhere around 100mg — roughly one filter coffee or two teas — and higher doses buy jitteriness, a raised heart rate and worse fine motor control rather than more alertness.

For rough orientation: a filter coffee is around 95mg, an espresso around 65mg, a mug of tea around 45mg, and a can of cola around 35mg. Energy drinks vary enormously and are worth reading the label on.

Health authorities generally put the safe upper limit for a healthy adult around 400mg a day, and considerably lower in pregnancy. This page is about performance rather than health, and that limit is not a performance recommendation — it is well above the point where the mental benefits stop increasing.

Why the first coffee of the day is the least useful one

Adenosine has been cleared overnight, so at 7am there is comparatively little for caffeine to block. The alertness you feel is largely the circadian signal rising on its own.

Cortisol also peaks in the half hour or so after waking, and while the popular advice to delay coffee for ninety minutes is stated with more confidence than the evidence supports, the underlying point holds: the morning is when you least need it and the early afternoon is when you most do.

If you drink two coffees a day and want them to work, the timing that matters is putting one into the trough rather than both into a period when you were going to be alert anyway.

The nap that beats the coffee

One technique has better evidence than either component alone, and it exploits the half-life rather than fighting it.

Caffeine takes roughly twenty to thirty minutes to reach useful blood levels. A short nap takes roughly the same time to be useful and starts becoming counterproductive after about twenty-five, when you risk waking out of deep sleep into grogginess.

So: drink the coffee, then immediately nap for twenty minutes. The nap clears some adenosine while the caffeine is still absorbing, and you wake as it takes effect. Studies of this combination generally find it beats either the nap or the caffeine on its own for subsequent alertness and driving performance.

It sounds like a trick and it is simply good timing. The one requirement is the alarm: past thirty minutes the sleep inertia costs more than the nap gained.

Withdrawal is a real effect, and it distorts your judgement

If you drink caffeine daily and stop, expect headache, irritability, low mood and genuinely measurable performance decline, typically peaking around a day or two in and resolving within about a week.

This matters for interpreting your own results here. Someone testing themselves without their usual coffee is not measuring their caffeine-free baseline — they are measuring themselves in withdrawal, which is a worse state than either. It takes a week or more of abstinence before a comparison means anything, which is why casual “I tried a day without coffee and felt awful” experiments always reach the same conclusion.

If you want to know what caffeine is actually doing for you, the honest test is a fortnight either side, not a day.

Seeing it in your own numbers

The comparison that works is the same one everything on this site uses: you against yourself, on the same equipment, at the same time of day.

Take the daily round at a fixed hour for a fortnight and note which days had caffeine and how much sleep preceded them. Most people find two things. The caffeine effect is real but smaller than expected on a rested day, and the sleep effect is larger than expected on every day.

That second finding is the useful one, and it is the reason this page keeps returning to sleep. Caffeine is a way of borrowing alertness against tomorrow. Sleep is the only way of actually having it.