Focus Without More Caffeine: What the Fourth Cup Is Actually Doing

The fourth cup is where people start asking the question. Not because coffee has stopped working — it has not — but because the amount required to reach the same place keeps drifting upward, and the cup that used to fix the three o'clock slump now mostly makes eleven at night worse. That drift has been measured on a calendar, in controlled trials, and it is the most predictable thing about caffeine.

What follows is what caffeine does and does not do, how tolerance develops, why the afternoon cup stops earning its place, and what people reach for instead. None of it is an argument against coffee, which is one of the better-evidenced things in this category.

• 85%Of the US population consumes at least one caffeinated beverage a day, in a 37,602-person diary survey
• 226 mgMean daily caffeine intake among consumers aged 50–64 — the highest of any age band
• 18 daysTo complete tolerance to caffeine's subjective effects at 300 mg three times daily, in a 32-person trial

Worth establishing before anything else, because the rest of the page reads as an attack otherwise.

A Cochrane review of caffeine in people working against their circadian rhythms pooled 13 randomised trials. Against placebo, caffeine improved orientation and attention (standardised mean difference −0.55; 95% CI −0.83 to −0.27) and memory (−1.08; 95% CI −2.07 to −0.09), with no benefit for verbal functioning and language skills (0.18; 95% CI −0.50 to 0.87). The trials measuring errors found caffeine significantly reduced them (Ker et al., Cochrane Database of Systematic Reviews, 2010; 13 randomised trials).

The reviewers' caveats matter as much as the result: no trial measured an injury outcome, and the evidence came largely from young participants under simulated conditions. Their conclusion is quotable and rarely quoted — there is no reason for healthy people already using caffeine within recommended levels to stop doing so.

It is also nearly universal. In a 7-day diary survey of 37,602 US consumers, 85% of the population consumed at least one caffeinated beverage daily, at a mean of 165 mg a day across all ages and 226 mg among consumers aged 50 to 64 (Mitchell et al., Food and Chemical Toxicology, 2014).

Caffeine is an adenosine receptor antagonist. Adenosine accumulates through the waking day and, as it binds, produces the sensation of mounting tiredness. Caffeine occupies those receptors without activating them, so the signal stops arriving.

Three consequences follow, and all three are visible in ordinary life. The tiredness is not cancelled, only unreported, and it is still there when the caffeine clears. Nothing has been added — no sleep repaid, no store replenished. And a system that keeps finding its receptors blocked adapts, which is the next section.

• The first cupReliable, and increasingly hard to distinguish from simply feeling normal.
• The secondStill works. Arrives earlier in the day than it used to.
• Three o'clockThe cup that used to fix the afternoon now mostly maintains it.
• SaturdayA late start and a headache by mid-morning, on the one day with no schedule.
• Eleven at nightAwake, and confident the four o'clock coffee had nothing to do with it.
• The next morningShort sleep, so the first cup is bigger. The loop closes.

Thirty-two healthy adults with histories of moderate caffeine consumption were given either 300 mg of caffeine three times daily or placebo for 18 consecutive days. Afterwards, a dose of caffeine produced significant subjective effects in the group that had been on placebo — and not in the group that had been on caffeine (Evans & Griffiths, Psychopharmacology, 1992; n=32, 18 days). The authors described it as the clearest evidence to date of complete tolerance to caffeine's subjective effects.

The more uncomfortable finding came later, and it reframes what a morning coffee is doing for a habitual drinker.

In that trial, 379 adults were split by habitual intake and given 100 mg of caffeine, then 150 mg more ninety minutes later, or placebo both times, after 16 hours without caffeine. Placebo in the medium-to-high consumers decreased alertness and increased headache; caffeine did not increase alertness in the non- and low-consumers at all (Rogers et al., Neuropsychopharmacology, 2010; n=379).

That is a claim about what the morning cup is for. In a habitual consumer much of the alerting effect is the reversal of an overnight withdrawal rather than a net gain over an unmedicated baseline — one trial among several on a contested question, and the single most useful result here.

The withdrawal side is well characterised. A review of 57 experimental and 9 survey studies validated ten symptoms, headache and difficulty concentrating among them, with headache occurring in 50% of cases in experimental studies. Onset typically fell 12 to 24 hours after abstinence, peaked at 20 to 51 hours, and lasted 2 to 9 days, with symptoms after doses as low as 100 mg a day (Juliano & Griffiths, Psychopharmacology, 2004; 66 studies).

Two separate things are happening to the three o'clock coffee, and conflating them is why the usual response is to make it larger.

The first is tolerance, which operates across weeks and applies to every cup equally. The second is timing, measured directly: a placebo-controlled study gave 400 mg of caffeine at bedtime, three hours before bedtime, and six hours before bedtime, with sleep monitored objectively as well as by self-report at home. All three timings produced significant sleep disturbance against placebo, and the reduction in total sleep time from the six-hour dose was large enough that the authors treated it as empirical support for the standard six-hour cut-off (Drake et al., Journal of Clinical Sleep Medicine, 2013).

So a four o'clock coffee is not neutral with respect to a ten o'clock bedtime, and the sleep it costs is what the following afternoon runs on. This loop is what keeps the dose climbing.

A third factor gets misattributed most often of all: the afternoon dip is a clock rather than a caffeine deficit. Two independent physiological processes produce a predictable early-afternoon trough regardless of what anybody drank — set out, with the datasets, in what happens to mental stamina between 1pm and 5pm.

The substitutions follow a predictable order, and two of them have better evidence than their reputations suggest.

• A bigger coffeeThe default, and the one the tolerance data speaks to directly. Escalation restores the dose, not the baseline.
• Energy drinks and nootropic drinksMostly caffeine with additions, often at doses not printed per serving in a form anyone can compare. The format question is worked through in drinks versus capsules.
• Switching to teaA genuine change rather than a smaller dose of the same thing, because tea carries L-theanine alongside its caffeine — the substitution here with the most directly relevant pooled randomised evidence behind it.
• A walkBetter than expected and smaller than hoped. A meta-analysis of 79 studies found the effect of a single exercise session on cognitive performance positive and small (Hedges' g = 0.097; n=1,034) — real, replicated, modest (Chang et al., Brain Research, 2012).
• Quitting entirelyDefensible, and the first week is the withdrawal curve above rather than a preview of life without it. What returns afterwards is a baseline, not a gain.

The tea answer deserves its own paragraph. A 2025 systematic review screened 50 randomised trials in healthy participants and meta-analysed 15, finding small-to-moderate differences favouring theanine plus caffeine over placebo within the first two hours: digit vigilance accuracy at hour two, standardised mean difference 0.20 (95% CI 0.02–0.38), and attention-switching accuracy 0.33 (95% CI 0.13–0.54) (Payne et al., Nutrition Reviews, 2025).

Small effects with their confidence intervals printed, in the hours after a dose. The individual products, and what each states on its panel, are compared on the L-theanine shelf.

The head-to-head version of this comparison — what another coffee does that a capsule does not, and the reverse — is in what another coffee actually does.

The awkward part first: a theanine capsule contains no caffeine, and nothing in it substitutes for sleep. A product in this category that implies otherwise is selling against the evidence it cites.

The theanine capsules built on that literature carry 200 mg of L-theanine — the dose used across most of the trials pooled in the 2025 review, and the compound tea supplies alongside caffeine while coffee does not. Alpha-GPC is the other studied ingredient, routinely misreported: trials in healthy adults have generally used single doses of roughly 250 to 630 mg, while the higher-dose trials quoted in marketing were run in people with diagnosed cognitive impairment — a different population answering a different question. What each set measured is in what the choline research measured.

What this page settles

Five things. Caffeine genuinely improves attention and reduces errors, with pooled effects and printed confidence intervals. It removes a signal rather than adding a resource, so nothing it does is repayment. Complete tolerance to its subjective effects developed within 18 days at 300 mg three times daily. Much of the morning lift in a habitual consumer is withdrawal reversal. And a dose six hours before bed still measurably disturbs sleep, which is how the afternoon cup is paid for twice.

What none of that settles is whether anybody should drink less coffee. The Cochrane reviewers declined to say so, and this desk is in no position to.

Of every substitution above, switching to tea is the substitution with the most directly relevant pooled randomised evidence behind it — and the compound that makes tea a different answer rather than a smaller dose of the same one is L-theanine, which the better-specified capsule formulas print at 200 mg, beside every other active with its amount and its chemical form.