Caffeine: Part One: What It Actually Does

Abstract

Caffeine is the most widely used psychoactive substance in the world and the best-evidenced compound in this library. That makes it a useful test of how much a well-studied stimulant actually delivers.

This is the first of three parts. It follows a 2021 systematic review and meta-analysis in Nutrients, and the sentence worth carrying away is the authors' own: after pooling the data, significant impacts of caffeine only emerged on attention, accuracy and speed.

Only. In a literature this large, that word is doing important work.

1.What Was Reviewed

Jorge Lorenzo Calvo, Xueyin Fei and colleagues searched Medline and Web of Science for studies published between August 1999 and March 2020, following PRISMA guidelines and defining inclusion by the PICOS model.

The inclusion criteria were strict in one respect that matters. Only randomised crossover studies were eligible, in which caffeine — as drinks, capsules, energy bars or gum — was compared against an identical placebo condition. Every participant served as their own control.

There were no filters on training level, sex or age. Thirteen studies examining objective or self-reported cognitive performance were selected for the systematic review; five of them supplied data usable for meta-analysis.

Thirteen studies, five pooled. That is the size of the quantitative evidence base for caffeine and cognition in this context — smaller than the compound's reputation implies.

2.What Survived Pooling

Attention. Accuracy. Speed.

Those three, and the authors are explicit that the significant impacts only emerged there. The narrative synthesis of all thirteen studies is broader — suggesting that a low or moderate dose before or during exercise can improve self-reported energy, mood and cognitive functions such as attention, and may improve simple reaction time, choice reaction time, memory or fatigue.

But that broader list carries its own qualifier: this may depend on the research protocols.

The gap between the two lists is the interesting part. When individual studies are described, caffeine appears to improve almost everything measured. When the numbers are pooled, three things survive.

That is the normal relationship between narrative and quantitative synthesis, and it is why we read meta-analyses rather than counting positive studies.

3.Attention Is Not Cognition

The three surviving outcomes have something in common. Attention, accuracy and speed are all measures of how reliably and quickly a person executes a task they already know how to do.

None of them is learning, reasoning, problem-solving or memory formation. A drug that makes you faster and less error-prone on a vigilance task has not made you smarter; it has made you more consistent.

This matters for how the compound is sold. "Improves cognitive function" and "improves sustained attention on a task you are already performing" describe the same evidence and produce different expectations.

It also fits what Part One of our L-theanine series found: every analysis separating theanine from caffeine concluded that caffeine was doing most of the visible work, and the work in question was attentional.

4.The Population Caveat

This review is about sport. The studies examined caffeine before or during exercise, in participants of varying training levels.

That is a specific context — physically aroused, often fatigued, frequently competitive — and it is not the context in which most people take caffeine, which is sitting at a desk in the morning.

Whether the attentional effects transfer to sedentary knowledge work is a reasonable assumption and not a demonstrated one. Exercise itself changes arousal, catecholamine levels and perceived exertion, all of which interact with what caffeine does.

We flag this because our own product literature has to be honest about the same boundary. Evidence gathered in athletes is evidence about athletes until someone shows otherwise.

5.What Is Not in This Review

The most important question about caffeine does not appear here at all, and it is the subject of Part Two.

Every participant in these crossover studies was a person who normally consumes caffeine, tested after a period of abstinence, against a placebo. If abstinence produces withdrawal — headache, reduced alertness, low mood — then the placebo condition is not neutral. It is a mildly impaired state.

On that account, a portion of what looks like enhancement is restoration: caffeine returning a habitual user to their own baseline rather than lifting them above it.

The question has been open for decades and is more tractable than it sounds. Part Two covers a study designed to answer it.

Editorial Comment

Boost contains 100 mg of caffeine, from anhydrous caffeine and guarana. We make no claim about it in this article.

Caffeine occupies an unusual regulatory position in the European Union, with claim conditions attached to dose and to the carrier product. We are not quoting any authorised wording here because we have not re-verified the current register entry, and quoting a claim from memory is precisely the error this library exists to avoid.

What we will say is what the evidence above supports and no more: pooled across randomised crossover trials, caffeine produced significant effects on attention, accuracy and speed, in studies conducted around exercise, in participants who habitually consume it.

Part Two asks whether those participants were being improved or merely restored.