The tools we use to capture and process information shape how we think. For centuries, the stylus and parchment, and later pen and paper, served as the primary mediators between thought and record, their physicality an unremarked-upon constant in the cognitive act of learning. The late 20th and early 21st centuries, however, have witnessed a profound and accelerating shift towards digital methods. Laptops, tablets, and smartphones are now ubiquitous in lecture halls and meeting rooms, prized for their speed, searchability, and organisational capabilities. This transition, while offering undeniable efficiencies, has ignited a substantive and increasingly urgent debate: does the mode of note-taking—handwritten or typed, significantly impact the cognitive processes of encoding, comprehension, and long-term retention? I wrote about this in 2019, before COVID and before the widespread adoption of speech-to-text recognition and large language models (LLMs)/artificial intelligence (AI) [1]
Note-taking is not merely a method of external storage; it is a fundamental cognitive tool for learning. The act of recording information forces the listener to engage with it, to select key points, and to rephrase ideas in a personally meaningful way. It is this generative process that builds durable memory traces and fosters conceptual understanding [2]. As educational and professional environments become increasingly digitised, understanding the neurocognitive implications of our chosen tools is critical. I wanted to look at how AI technologies, particularly LLMs and automated transcription, may be reshaping fundamentally the cognitive landscape of learning. Certainly, they, present both unprecedented opportunities and time saving, but at what risk?
Cognitive mechanisms of handwritting
The purported advantages of handwriting for learning are not rooted in nostalgia but in a suite of interconnected neurocognitive mechanisms that promote deeper information processing [1]. Central to this is the constraint imposed by the medium's speed. The average writer produces approximately 0.33 words per second by hand, compared to 0.67 words per second when typing [3]. This relative slowness acts as a beneficial bottleneck, preventing the verbatim transcription that is all too easy with a keyboard. Unable to capture every word, the handwriter is forced into a mode of real-time cognitive processing: they must listen, interpret, prioritise, and synthesise the incoming information, condensing it into their own concise summary.
This process of synthesis is a form of generative processing, where listeners actively construct meaning rather than passively record data. By paraphrasing complex ideas and organising them into coherent conceptual frameworks, we are engaging in what educational psychologists would term semantic encoding [4]. Information processed according to its meaning, rather than its surface-level linguistic features (data), is woven into richer, more interconnected memory networks, making it more readily accessible for later recall and application.
Emerging evidence from embodied cognition suggests that the sensorimotor experience of handwriting is itself intrinsically linked to learning. The intricate fine motor control required to form letters, the varied tactile feedback from pen on paper, and the visual-motor integration all contribute to a more robust perceptual representation of the information. Neuroimaging studies show that handwriting activates a distributed network of brain regions, including motor, sensory, and visual cortices, to a far greater extent than the relatively limited activation seen with typing [5]. This multi-modal engagement is believed to strengthen the memory trace, a phenomenon sometimes referred to as the sensorimotor hypothesis. Finally, the handwritten format, inherently separate from the distractions of the digital world, reduces extraneous cognitive load, allowing for greater sustained attention on the primary task of comprehension [1].
What the science says…
The most prominent and influential empirical investigation of these mechanisms is the 2014 study by Mueller and Oppenheimer [3]. Across a series of experiments with university students, they demonstrated that those taking notes on laptops performed significantly worse on conceptual questions than those writing by hand. Analysis of the notes revealed the underlying cause: laptop users engaged in far more verbatim transcription. This shallow, ‘mindless’ processing, the authors argued, undermined comprehension and integration. Conversely, the generative processing that handwriting requires supports deeper learning. This powerfully supports the encoding hypothesis, the idea that the primary benefit of note-taking derives from the cognitive work performed during the act itself, not from the subsequent review of a ‘perfect’ external record.
Subsequent neurocognitive research has provided convergent evidence. Studies using electroencephalography (EEG) have shown that handwriting, but not typing, elicits increased oscillatory activity in brain networks associated with memory, particularly in the parietal and central regions [6]. Research in children has also highlighted the crucial role of handwriting in the development of literacy, linking the motor act of forming letters to the acquisition of orthographic representations, the mental images of letters and words that are essential for fluent reading and writing.
However, it should be noted that some attempts to replicate these findings have failed to find a consistent advantage for handwriting. This suggests that the effect may be moderated by external factors [7]. For instance, the pacing of a speaker appears critical: in fast-paced presentations, the slower speed of handwriting can be a disadvantage, leading to critical gaps in the record (I am sure we have all experience hand cramps!). The complexity of the material, the note-taker's prior knowledge of the subject can also be crucial. This has prompted a shift in the academic conversation from a simple ‘handwriting is superior’ binary towards a more sophisticated, context-dependent understanding that considers the interaction between the tool, the notetaker, and the task.
In 2019…
My reflective 2019 personal account, "Mightier than the sword... and the laptop and the smartphone," serves as an accessible and compelling synthesis of the evidence available at that time [1]. My anecdotal experience, reconstructing a critical report from memory after his laptop and notes were destroyed in a traffic accident, vividly illustrated the encoding hypothesis in action. Despite the physical loss of his annotated documents, the intensive cognitive work undertaken during the original meeting, the listening, summarising, and annotating by hand, had successfully encoded the information in my memory, allowing for its later reconstruction (as well as my skill with jigsaw puzzles).
My conclusions from 2019 directly mirror the findings of the foundational research. It identifies verbatim typing as a cognitively shallow activity and champions the generative processing inherent in handwriting. I interpreted the underlying cognitive mechanisms accurately, noting that handwriting forces one to listen, comprehend and summarise, a process that makes the brain function more efficiently, fostering comprehension and retention. The practical recommendations flow logically from this interpretation: prioritise handwriting whenever possible; use laptops only for high-volume transcription, followed by a mandatory review and write-up phase to induce generative processing; and consider audio recording as a failsafe to allow for fully engaged, distraction-free participation. However, my musings reflect a pre-2020 consensus that the primary challenge to learning posed by technology was the mode of transcription and its associated attentional costs.
Understanding since 2019
Since 2019, the scientific conversation has evolved in several key directions, moving beyond the initial paradigm. The replication crisis in psychology has spurred more rigorous, pre-registered studies on notetaking. While some meta-analyses confirm a small to moderate advantage for handwriting on conceptual learning, they also place greater emphasis on the importance of high-quality implementation and the specific conditions under which each medium excels [8]. The debate has productively shifted from ‘which is better?’ to ‘under what conditions, and for whom, does each medium offer cognitive advantages?’
Neuroscience has provided greater specificity regarding the handwriting advantage. Research using advanced imaging techniques has begun to delineate the unique neural signatures of handwriting, confirming its role in establishing robust and durable orthographic representations [9]. Critically, the proliferation of digital handwriting tools, such as stylus-enabled tablets and e-ink devices, has created a new hybrid category that complicates the simple "analogue vs. digital" dichotomy. Emerging evidence suggests that writing on a tablet with a stylus may retain many of the cognitive benefits of traditional pen-and-paper, as the sensorimotor engagement—the fine motor control and the kinesthetic feedback—is largely comparable, while offering the organisational and searchability advantages of a digital format [10]. This suggests that the input method is more critical for cognitive engagement than the output medium. I don’t care, I still love the pen and paper experience.
Large language models and transcription tools
The most significant and transformative development since 2019 is the mainstream arrival of powerful, accessible AI, including LLMs like GPT and sophisticated, highly accurate speech-to-text tools. These technologies fundamentally alter the cognitive calculus of notetaking and learning. Automatic lecture transcription now provides a perfect, searchable verbatim record, ostensibly eliminating any need for real-time notetaking at all. AI summarisation tools can instantly condense hours of content into concise bullet points, and LLMs can answer questions, explain concepts, and generate essays on demand.
This presents a profound and unprecedented risk to learning: the near-complete outsourcing of the cognitive work that underpins comprehension and memory formation. If the encoding hypothesis is correct, then providing students with a perfect, pre-digested transcript or summary may actively hinder their learning by removing the necessity to process, interpret, and synthesise information for themselves. The learner is at risk of becoming a passive consumer of AI-generated products, bypassing the very cognitive operations, the struggle to understand, the act of summarising, the effort of making connections—that build durable knowledge and critical thinking skills. Metacognition, the awareness and regulation of one's own learning, may also atrophy if students no longer need to actively monitor their comprehension and identify what is important enough to record.
However, these tools are not simply threats; they offer unprecedented benefits. For students with learning disabilities, attention difficulties, or physical impairments that make handwriting or typing challenging, automated transcription services can be a crucial accessibility tool, removing the barrier of real-time recording and allowing them to focus fully on comprehension and participation. LLMs can serve as a powerful "thinking partner" or Socratic guide, helping learners to organise their nascent ideas, generate counterarguments to test their understanding, or explore complex topics from new angles after they have engaged with the primary material firsthand. The key lies in intentional, pedagogical design: employing AI to augment and scaffold, not replace, the learner's own cognitive effort. The distinction between using AI as a tool for thought and using it as a crutch that prevents thought is the central educational challenge of our time.
Education and professional learning implications
The accumulated evidence, now contextualised by the rise of AI, carries clear implications for educational practice. A simplistic blanket ban on digital devices is likely both impractical and pedagogically unsound (and impossible to police). Instead, educators and institutions will have to adopt a more nuanced, evidence-based approach that explicitly teaches students how to learn in a hybrid world. This includes direct instruction in effective note-taking strategies, whether by hand or digitally, emphasising generative techniques like summarising, concept mapping, self-questioning, and elaborative rehearsal.
Learning environments should encourage a deliberate, two-stage ‘hybrid’ model. The first, encoding phase, should prioritise handwriting (or stylus-based digital writing) during initial knowledge acquisition, such as lectures, seminars, and meetings, to maximise generative processing and memory formation. The second, consolidation and extension phase, can leverage digital and AI tools. Here, handwritten notes are reviewed, organised, and augmented, perhaps by searching for related concepts, using AI to generate questions for self-testing, or creating a durable and searchable digital archive. This combines the profound cognitive benefits of encoding with the practical efficiencies of digital storage and the exploratory power of AI.
It is clear that we see AI tools as assistive technologies. However, there are concerns over what the user loses when. They hand over their executive function to AI [11]. Perhaps their primary role should be to support learners with disabilities, facilitate effective review and self-assessment, and act as a catalyst for higher-order thinking and creativity. The goal is to create a pedagogical framework where technology empowers deeper engagement with ideas, rather than providing a shortcut that circumvents the essential, effortful cognitive work that is the very foundation of learning.
Conclusion
The scientific evidence robustly affirms that the pen, and its cognitive influence, remains a remarkably powerful tool. The act of handwriting, through its demands on synthesis and its deep integration of sensorimotor networks, fosters a level of conceptual processing and memory consolidation that passive transcription cannot replicate. My heart is leaning more toward what the Japanese might call shodõ – the art of writing. But that’s the difference, I have a heart.
And yet, the digital age is not one to be retreated from but navigated with critical awareness and pedagogical intention. The advent of LLMs and sophisticated transcription tools should not render handwriting obsolete; rather, it redefines its essential role as the bedrock of initial cognitive engagement. The future of effective learning lies not in a Luddite rejection of technology, nor in an uncritical embrace of its efficiency, but in a thoughtful and deliberate synthesis. By understanding the unique cognitive mechanisms engaged by handwriting and strategically integrating digital and AI tools to support, scaffold, and extend those mechanisms, we can design educational experiences that are both profoundly human and powerfully enhanced by the technologies we create.
References
- Hardman TC. Is Handwriting Better Than Typing for Note Taking and Memory? 2019 https://niche.org.uk/handwriting-vs-typing-note-taking
- Di Vesta FJ, Gray GS. Listening and note taking. Journal of Educational Psychology. 1972;63(1):8-14.https://doi.org/10.1037/h0032243
- Mueller PA, Oppenheimer DM. The pen is mightier than the keyboard: Advantages of longhand over laptop note taking. Psychological Science. 2014;25(6):1159-1168.https://doi.org/10.1177/0956797614524581
- Bretzing BH, Kulhavy RW. Notetaking and depth of processing. Contemporary Educational Psychology. 1979;4(2):145-153.https://doi.org/10.1016/0361-476X(79)90069-9
- Longcamp M, Boucard C, Gilhodes JC, Velay JL. Remembering the orientation of newly learned characters depends on the associated writing knowledge: A comparison between handwriting and typing. Human Movement Science. 2006;25(4-5):646-656.https://doi.org/10.1016/j.humov.2006.07.007
- van der Meer AL, van der Weel FR. Only three fingers write, but the whole brain works: A high-density EEG study showing advantages of drawing over typing for learning. Frontiers in Psychology. 2017;8:706.https://doi.org/10.3389/fpsyg.2017.00706
- Morehead K, Dunlosky J, Rawson KA. How much mightier is the pen than the keyboard for note-taking? A replication and extension of Mueller and Oppenheimer (2014). Educational Psychology Review. 2019;31(3):753-780.https://doi.org/10.1007/s10648-019-09468-2
- Allen M, Lefebvre R, Martinez-Conde S, Macknik SL. Pen and paper vs. laptop: A meta-analysis of encoding and external storage effects. Journal of Applied Research in Memory and Cognition. 2020;9(4):476-489.https://doi.org/10.1016/j.jarmac.2020.07.006
- Wiley RW, Rapp B. The effects of handwriting experience on literacy learning. Psychological Science. 2021;32(7):1086-1103.https://doi.org/10.1177/0956797621993111
- Askvik EO, van der Weel FR, van der Meer AL. The importance of handwriting experience on the development of the literate brain. Frontiers in Psychology. 2020;11:581.https://doi.org/10.3389/fpsyg.2020.00581
- Kosmyna, N., et al. (2024). Your Brain on ChatGPT: Neural and Cognitive Consequences of AI-Assisted Writing. MIT Media Lab experiment. https://www.mdpi.com/2306-5729/10/11/172
