If you have heard one fact about note-taking, it is probably this one: writing by hand is better for memory than typing. It comes from a single well-titled study, it was reported everywhere, and it is repeated in study guides that have never looked at what happened next.
What happened next is that a careful replication mostly failed to find it. The truth is more useful than either headline, and it points at a variable neither camp usually names.
The study everyone quotes
Pam Mueller and Daniel Oppenheimer published The Pen Is Mightier Than the Keyboard in 2014. Students watched lectures while taking notes by hand or on laptops, then answered factual and conceptual questions.
Both groups did about the same on factual recall. On conceptual questions, the longhand group did better.
The proposed reason was not the hand or the pen. It was speed. Typists could keep up, so they transcribed. Writers could not keep up, so they were forced to select, compress and rephrase in real time, and that processing is what produced understanding. The paper's own framing was that verbatim transcription is the problem and slowness is an accidental cure for it.
Notably, telling the laptop group not to transcribe did not help. They transcribed anyway.
The replication nobody quotes
In 2019, Morehead, Dunlosky and Rawson published a direct replication and extension in Educational Psychology Review, adding conditions the original lacked, including an eWriter group and a group that took no notes at all.
They did not reproduce the result cleanly. Some trends leaned longhand, but performance did not consistently differ between groups, and a meta-analysis of the direct replications found only a small effect favouring longhand that did not reach significance.
Their conclusion was appropriately flat: the available evidence does not settle whether students should take notes by hand or by computer.
This is not a scandal. It is what usually happens to a striking single result, and the correct response is to hold the original conclusion much more loosely.
The neuroscience wave, and its critics
In 2024 van der Weel and van der Meer published a high-density EEG study in Frontiers in Psychology reporting that handwriting produced widespread theta and alpha connectivity across parietal and central regions that typing did not. It was covered enthusiastically as proof the debate was over.
It is worth reading the published commentary alongside it. Pinet and Longcamp raised objections that are hard to dismiss: participants wrote single familiar words with no learning task at all, so nothing about learning was measured; increased theta and alpha connectivity has not been established as a condition favourable to learning; and the typing condition required one-finger typing, which is not how anyone types and which would depress typing-related activity on its own.
Different brain activity is not the same claim as better learning. The study is interesting. It is not the closing argument.
What actually survives
Strip out the contested parts and a solid core remains.
Verbatim transcription is bad, and this is not in dispute. Copying words without processing them produces poor retention regardless of instrument. This is the finding that has held up across every version of this research.
Typing makes transcription easy, and people do the easy thing. This is the real mechanism. It is a behavioural tendency, not a property of keyboards, which is why it can be defeated by anyone willing to defeat it.
Handwriting imposes a speed limit that forces selection. A useful accident. You cannot copy a lecture by hand, so you must decide what matters, and deciding is the part that works.
Nobody has shown that a typist who genuinely processes does worse than a writer who genuinely processes. That comparison is the one that matters and it is largely absent from the literature.
So the operative variable is not handwriting. It is whether you are compressing or copying.
The practical answer
Type if you will impose the constraint yourself. The constraint is what does the work, and it can be manufactured:
- Write in your own words. If a phrase arrives already formed from the speaker's mouth, rewrite it.
- Cap yourself. Aim for a fraction of what was said, not a record of it. See how to take notes fast, which is really about writing less rather than typing quicker.
- Use a structure that forbids transcription. The Cornell method reserves a column for questions you have to invent, which cannot be transcribed. The outline method forces you to judge what is subordinate to what.
- Leave a gap and fill it later from memory, not from the slides.
Write by hand when the material is spatial, or when the device is the problem. Diagrams, equations, sketchnoting and anything with arrows are still faster and better on paper. And a laptop in a lecture is an open portal to everything else, which is a documented cost that has nothing to do with encoding: see context switching.
Consider the hybrid, which is what most people should do. Capture by hand where speed forces selection, then process into your system afterwards. The second pass is not duplicated effort, it is a spaced retrieval attempt, which is the single highest-return thing you can do with a set of notes. See how to review your notes and spaced repetition.
The thing paper cannot do is be found in three years. That is the trade the whole digital vs paper notes question turns on, and it is why the hybrid keeps winning: encode on paper, retrieve from a searchable archive.
The one-sentence version
The pen is not mightier than the keyboard. Thinking is mightier than copying, and the pen just makes copying harder.
More in why writing helps memory, how to take lecture notes, and the wider note-taking methods section.