Students drink coffee to study, and the assumption underneath that is that caffeine helps you remember things. The research says something more specific and considerably less flattering. Caffeine is good at one thing that looks like memory improvement from the outside, and the trials that tested memory directly disagree with each other in a way that turns out to be informative. If you want to know whether the coffee is helping you learn, the honest answer requires separating three different processes that the word memory runs together.
Three things people mean by memory
Encoding is getting information in, which happens while you are reading or listening. Consolidation is the process by which a fragile new memory becomes stable, occurring in the hours after learning and heavily dependent on sleep. Retrieval is getting it back out, which is what an exam tests. Working memory is different again: the small amount of information you can hold and manipulate right now, like keeping a phone number in mind while you find a pen.
Caffeine can affect these separately and in different directions, which is precisely why the studies disagree. A drug that helps you stay awake to encode more material is not necessarily a drug that helps the material stick.
The famous consolidation study
The study that launched this question was Borota and colleagues[1], published in Nature Neuroscience. They gave caffeine after participants had already studied the material, which is the clever part of the design: if caffeine only helped by keeping people alert during learning, dosing afterwards should do nothing.
It did something. Caffeine enhanced performance 24 hours later on a behavioural discrimination task, and the effect followed an inverted U-shaped dose-response curve, meaning a middling dose worked better than a low or a high one. Crucially, the effect was specific to consolidation and not retrieval. The authors concluded that caffeine enhanced consolidation of long-term memories in humans.
The task matters for interpreting it. The discrimination test required telling previously seen items apart from very similar new ones, a fine-grained ability that depends on the hippocampus. The finding was about detail discrimination rather than about remembering more things.
The study that found the opposite
In 2025, Leon and colleagues[2] set out to extend that result to faces, which are socially important and harder to discriminate. Using a randomised, double-blind design, 97 participants viewed ten generated faces on day one and then received 200 mg of caffeine or placebo. On day two they completed a recognition task, in one condition with the original face among five similar distractors and in another with six similar distractors and a none of the above option.
Caffeine did not improve consolidation of face recognition memory. Instead the researchers observed a general impairment in recognition performance, meaning a reduced ability to distinguish previously encountered faces from novel but similar ones. They connected this to work suggesting caffeine enhances global processing at the expense of detailed discrimination.
That is the same fine-grained discrimination ability the earlier study found improved, measured in a different domain, and moving the other way. With 97 participants this is the larger of the two trials.
What happens to working memory
If caffeine's benefits were straightforward, working memory would be where they show up, since it is the most alertness-sensitive form of memory. Kassim[3] ran a randomised, placebo-controlled, double-blind study giving 200 mg of caffeine to habitual coffee and tea drinkers and testing both spatial and verbal working memory with span tasks.
Caffeine significantly impaired spatial working memory, on both the maximum span obtained and the total score, in a delay-independent manner. Verbal working memory showed no significant effect. The conclusion stated it directly: the findings contradict the commonly held belief that caffeine improves cognitive performance, and acute caffeine may impair spatial working memory in habitual consumers.
That finding aligns with the older review by Nehlig[4], which concluded that caffeine facilitates working memory to a limited extent but hinders performance on tasks that depend heavily on it.
What caffeine reliably does instead
The most useful trial for settling this compared caffeine against prescription stimulants. Repantis and colleagues[5] gave male participants placebo and one of caffeine, methylphenidate or modafinil, testing across several cognitive domains. Methylphenidate had positive effects on self-reported fatigue and on declarative memory 24 hours after learning. Caffeine had a positive effect on sustained attention. Modafinil showed no significant effect on anything in their battery. The authors described the significant effects as domain-specific and of rather low magnitude.
Caffeine improved attention, not memory. That is consistent across this whole literature and it is the honest version of the claim: a compound that keeps you attentive will help you learn material you would otherwise have been too tired to take in, without improving memory itself.
Nehlig[4] summarised the same pattern from the older evidence. Caffeine does not usually affect performance in learning and memory tasks. It facilitates learning when information is presented passively, and has no effect when material is learned intentionally, which is what studying is. Most studies found improvements in reaction time. And memory performance improved mainly under suboptimal alertness conditions.
That last clause is the practical heart of it. Caffeine helps a tired brain get closer to its own baseline. It does not raise the baseline.
The withdrawal problem
There is a further complication that applies to every study of caffeine in habitual consumers. Rogers and colleagues[6] have shown that much of what looks like caffeine enhancing performance is caffeine reversing the withdrawal it caused, since regular drinkers tested without their usual dose are performing below their own normal, not at a neutral baseline. Rogers and colleagues[7] found that in habitual non-consumers, caffeine produced no reinforcing, mood or psychomotor performance effects at all, which is what you would expect if its apparent benefits in regular drinkers are largely restorative.
This cuts both ways in the studies above. A trial that gives habitual drinkers caffeine after overnight abstinence and finds improvement may be measuring restoration. A trial like Kassim[3], which found impairment in habitual consumers, is harder to explain that way, which makes it more interesting rather than less.
Why two good trials can disagree
It is worth sitting with the contradiction between Borota and colleagues[1] and Leon and colleagues[2], because how you read it determines what you do with the whole field. Three explanations are on the table, and they are not mutually exclusive.
The first is the material. Objects and faces are processed differently, and face recognition is a specialised ability with its own neural machinery, so a compound could plausibly help one and not the other. The second is the proposed mechanism in the 2025 paper: if caffeine shifts processing toward the global and away from the detailed, then any task that rewards fine discrimination could be hurt, and the earlier positive result becomes the outlier needing explanation. The third is ordinary statistical life. The first study was small and striking, published in a high-profile journal, and striking early findings in psychology have a poor record of surviving replication.
The honest position is that the consolidation claim is unresolved. That is a less satisfying answer than either headline, and it is the one the evidence supports.
It is also worth noting how the popular version of the 2014 finding drifted. The study dosed caffeine after learning and tested 24 hours later on a fine discrimination task, and it was reported widely as evidence that coffee boosts memory. Neither the timing nor the narrowness of the task survived the retelling, which is how a careful finding about consolidation of detail became a general claim about studying.
What about the dose curve?
The inverted U-shape in Borota and colleagues[1] deserves attention because it complicates the intuition that more caffeine means more effect. An inverted U means performance rises with dose to a point and then falls, so a large coffee can be worse than a medium one for the same task.
That pattern recurs elsewhere in caffeine research, particularly where anxiety is involved, since higher doses increase jitteriness and that interferes with performance, as covered in our article on caffeine and anxiety. For anyone using coffee to study, it means escalating the dose through an evening is unlikely to keep buying benefit and may cost some.
Before or after studying?
This is the practical question people arrive with, and the evidence supports a narrow answer rather than a confident one. If you are too tired to concentrate, caffeine before studying will help you attend to the material, which is a real if indirect benefit. The consolidation case for caffeine after studying rests primarily on one influential trial that a larger and more recent one failed to replicate in a different domain.
What is not in dispute is the thing that competes with both: sleep is where consolidation happens, and caffeine taken late enough to help you study can measurably reduce sleep, as covered in our article on caffeine and sleep. A late coffee that buys two more hours of revision and costs an hour of sleep may be a poor trade for memory specifically.
Reading this from the Philippines
Café study culture is well established here, and Marikina has a long list of cafés people work in, covered in our guide to study-friendly cafés. Nothing in this research argues against that, and the environment, a quiet table, few interruptions and a few hours set aside, plausibly does more for learning than anything in the cup.
The finding worth carrying into an exam period is the sleep one. A study pattern built on late coffee and short nights is trading the process that consolidates memory for the process that keeps you awake to make more of it, and only one of those is doing the remembering.
What the long-term brain research says, and why it is separate
People often mix this question up with a different one: whether coffee protects against dementia. That literature is observational, spans decades rather than days, and is covered in our article on coffee, tea, stroke and dementia. It asks whether coffee drinkers are less likely to develop a disease, not whether a cup this afternoon helps you remember a list.
Keeping them apart matters because a finding in one is routinely used to sell the other. Evidence that habitual coffee drinkers have a slightly different dementia risk says nothing about whether caffeine will help tonight's revision, and the trials above, which are the relevant evidence for that, are decidedly unenthusiastic.
The limits of this evidence
These are small laboratory trials, mostly in young adults and students, using tasks such as picture discrimination and span tests that resemble exams only loosely. The two consolidation studies used different materials, objects and faces, so their disagreement may reflect the task rather than a true contradiction, and neither has been replicated widely. Doses, timing and participants' habitual intake vary between trials, and the withdrawal-reversal problem contaminates almost all of it. Nobody has run a study that randomises students to coffee across a semester and measures their grades.
The practical answer
Caffeine is an attention drug, not a memory drug. The strongest consistent finding across these trials is improved sustained attention and reaction time, which helps you learn by keeping you engaged rather than by improving retention. The evidence that caffeine after studying strengthens consolidation rests on one notable trial, and a larger 2025 trial found impaired discrimination in a related task. Caffeine may impair spatial working memory in habitual consumers. If you are studying, the defensible uses of coffee are to stay alert enough to work and to avoid withdrawal, and the intervention with the better evidence for remembering things is sleeping after you learn them. See also our article on caffeine and focus and our article on coffee and long-term brain health. Nothing here is medical advice.