I’ve made my position on edtech clear plenty of times: tools matter less than what teachers do with them. But the medium children use to do their thinking still leaves traces, and those traces show up in how much they understand a week after the lesson, not in what they remember an hour later.
That’s what makes Horbury and Edmonds’ (2021) study on note-taking by hand versus typing in a primary classroom worth a second look. The paper is small. Its sample is awkward, and the data are now seven years old. The result still does real argumentative work in 2026.
Handwriting vs Typing Notes: What the Study Actually Did
Horbury and Edmonds ran two short studies with 26 boys aged 10 to 11 in a UK primary school. In the first study, half the boys handwrote notes on a video lesson about the Black Death, and half typed their notes on Chromebooks. The roles flipped a month later for a second video lesson, this time on cells in biology. Each child took a multiple-choice test immediately after the lesson and another a week later. The questions were split into factual recall items (dates, names, basic facts) and conceptual understanding items (why something happened, how a process works).
That last split is the part of the design I think is doing the most work. Most note-taking research treats “recall” as a single thing. Horbury and Edmonds insist on separating shallow retrieval from deeper understanding, and the entire story turns on that split.
The Finding That Holds Up
On factual recall, note-taking mode did not make a difference. Children who handwrote and children who typed scored about the same, in both subjects, at both time points. Horbury and Edmonds found no statistically meaningful gap on the surface-level questions.
The conceptual results look different. A week after the lesson, children who handwrote their notes scored significantly higher on understanding questions in both the history and the biology lesson. The effect sizes were 1.6 in history and 1.3 in biology, both classed as very large in the standard tables. The authors observe that “at one week after history and science lessons, children who had made notes by handwriting had better conceptual understanding than children who made notes using a laptop computer” (p. 61).
That gap did not show up on the immediate test. It opened up over the week between the lesson and the second test. Horbury and Edmonds connect this to revision time: handwritten notes, they suggest, support deeper review during the gap. They were careful to offer revision time to both groups equally, but they did not track who actually used it. I’ll come back to that.
The Detail That Surprised Me
Here is the part of the paper I find most useful. In adult studies, typists usually generate a far higher word count than handwriters, and that extra volume is part of why researchers think laptop notes encourage shallow, verbatim copying. With these children, both groups produced about the same number of words. Typists did not write more. They wrote about the same amount as the handwriters. And they still came out behind on conceptual understanding a week later.
Horbury and Edmonds note that “it is striking that, in our study, the two modes of note-taking had such different effects on understanding when the same amount of notes were recorded under both conditions” (p. 62). The volume explanation that gets tossed around in the adult literature does not account for what they observed in children.
So what does account for it? The authors suggest something about the cognitive cost of typing for children who are not yet fluent touch typists. If you are still hunting for the keys, you are using working memory to type, not to think about what you’re hearing. That leaves less capacity for the deeper encoding that produces conceptual understanding later. They flag this as one possibility among several, including differences in how handwriting itself supports active processing.
The Connection to the AI Conversation
This paper was not written about AI. It was written about laptops in 2018-2019 and published in 2021. But the cognitive logic it describes is exactly the logic at the center of the current debate over AI tools in classrooms. When students offload mental work to a machine, whether that machine is a laptop or a chatbot, the surface output can look fine while the underlying learning thins out.
That argument runs through Gerlich’s (2025) work on cognitive offloading and critical thinking, where heavier AI reliance correlated with weaker thinking skills. Fan et al. (2025) found a similar pattern with metacognitive laziness: students using AI feedback did not gain new knowledge or transfer skills, even though their immediate work improved. The EEG evidence in Kosmyna et al. (2025) lines up with the same picture. Brains of students writing essays with ChatGPT showed less engagement than those writing without it. The Horbury and Edmonds paper belongs in that same family of findings, just predating it by a few years.
The pattern is consistent: the medium through which students do their thinking shapes how much thinking actually happens.
Where I’d Be Cautious
This is a small study, with twenty-six boys from a single school, all male. The authors are clear about this. They explicitly note that note-taking habits and digital distraction patterns differ by gender, so the results may not generalize. They also did not assess typing speed or accuracy, did not analyze the quality of notes, and did not record who actually revised before the second test. Each of those gaps matters for how confidently we can read the result.
I’d add another caution. The data are now seven years old, collected before tablets with handwriting recognition became standard, before AI tutors arrived in classrooms, and before the wider conversation about cognitive offloading had matured. The basic finding still feels relevant, but the digital environment children are working in today is not the digital environment of 2018. Replication with current devices, current AI tools, and a more representative sample would settle a lot.
What This Means for Teachers and Schools
I’m not arguing schools should abandon laptops. I have argued the opposite many times. But the choice of writing medium is a pedagogical choice, not a logistical one. Horbury and Edmonds recommend that “part of preparing children for the workplace, and/or the demands of further study, is to offer students touch typing training and clear advice on the best way to produce notes” (p. 65). I’d go further. Schools need to think about the cognitive demands of note-taking the same way they think about the cognitive demands of any other learning task.
That means teaching children how to take notes well, in either mode. It means choosing the mode based on what the lesson is asking children to do, not on what the school happens to have purchased. Conceptual lessons may earn handwriting time. Heavy retrieval lessons may not. The authors close with a similar nuance: “the optimal note-taking strategy may depend on skill with keyboards, mode of presentation, and time between learning and test, among other factors” (p. 65).
This is the shape of every good edtech question, including every good AI question. The medium is not neutral. The medium is not destiny either. The medium is part of the pedagogy, and the pedagogy is what teachers control.
Pick the mode that fits the lesson. Teach children to use both well. Then ask them what they remember a week later, not an hour later.
References
- Fan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., … & Gašević, D. (2025). Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance. British Journal of Educational Technology, 56(2), 489-530.. https://doi.org/10.1111/bjet.13544
- Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15(1), 6. https://doi.org/10.3390/soc15010006
- Horbury, S. R., & Edmonds, C. J. (2021). Taking class notes by hand compared to typing: Effects on children’s recall and understanding. Journal of Research in Childhood Education, 35(1), 55-67. https://doi.org/10.1080/02568543.2020.1781307
- Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., … & Maes, P. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. arXiv preprint arXiv:2506.08872.