Handwriting better than Typing for Memory and Learning. On Why so.

“If you want to absorb information, a pencil is best. If you want to produce it, however, a keyboard should be your best friend.”

“Taking notes by hand leads to better memory recall.”

“Increase people’s typing speed and you’ll increase the quality of their writing.”

— Clive Thompson [1] and Daphne at Publication Coach [2]

Behavioral research [3] finds that students who take notes longhand performed better on ‘conceptual’ questions than students who take notes on laptops (See the attached chart below). For ‘factual’ questions, the study does not report a significant difference even though laptop notes typically carry more (factual) details. The finding was true irrespective of whether the students were allowed to study from their typewritten/handwritten notes or not.

Brain research [4, 5] indicates that handwriting involves widespread brain connectivity. This should not be surprising however as handwriting involves more sophisticated muscle movements than typing -- So is it that handwriting leads to higher brain connectivity for the concepts being written about, or only for the handwriting process itself? :-) Like our experiences and behavioral research suggest, there indeed are indicators [5] that the taking handwritten notes leads to more connectivity for the new concepts learned (while handwriting itself may already been mastered, not requiring the connectivity to develop further).

It is less clear why all this happens this way. Here are some hints from the above-cited research articles: We note for a start that handwriting is distinctly slower than typing [6]. Laptop notetakers tend to take notes verbatim whereas writing by hand forces the notetakers to ‘compress’ the thoughts they write, simply because they cannot write fast enough. This forces the hand writers to think harder as they take notes. For the same reason, note takers are less distracted when writing by hand. On the other hand, typing is better when producing an article (from your existing knowledge and understanding) as it involves ‘transcription fluency’, i.e., you can type closer to the speed of your thought and capture more thoughts before they escape from your mind.

I further surmise:

• There are multiple processes involved: (a) Generating thoughts from existing knowledge and understanding, (b) Learning something new, i.e., assimilating and accommodating new thoughts into the current knowledge/understanding, (c) Motor neuron and muscle activity involved for handwriting and typing. These processes have different speeds. What’s slow in writing by hand is the hand, not the brain.
• When writing by hand, the brain thereby forcibly has more idle time and therefore opportunity to think and connect the dots. The brain holds the new thoughts in the working memory for longer, recalls the existing knowledge/understanding, correlates the new with the existing, generates new thoughts, refines the cumulated thoughts, and stores.
• There's a possibility that by holding the thoughts for longer, the working memory also creates a stronger stimulus for the long-term memory to store better. I have no reason to think so though.
• Connecting the dots is crucial to conceptual understanding. We go by a reasonable approximation/assumption that the brain learns and stores via ‘synaptic weights’ [7]. Understanding, for the brain, then is ultimately memorization too. Memorization however need not be understanding. The difference is how much semantic structuring happens in those neurons/weights. Rote memorization would be closer to relatively unlinked pieces of information held, whereas understanding would have more cross-linking [8].
• Such connecting the dots however is not as intense when producing information into a writeup, as recalling is easier than learning. Writing too however can involve intense thinking. Blaise Pascal’s quote [9] comes to mind, “I have made this longer than usual because I have not had time to make it shorter.” Writing correctly and crisply is significantly harder. Writing crisply is again higher ‘compression’ seen for note taking by hand. (Any thoughts or experiences on whether writing by hand is better than typing also for composing such articles? I have personally found electronic typing to be better as it makes restructuring easier.)
• Our memories seem to have a “forgetting curve” [10] in action that describes decline in memory retention with time. A learning technique called “spaced repetition” [11] is used to optimize retention. More conceptual or cross-connected our learning is [12], the less likely we may forget, as the new experiences and situations would keep reinforcing those memories [15]. Hence note taking by writing would lead to better retention.
• When taking notes by hand, we are more likely to miss out on what the teacher is continuing to explain. Overall, as the experimental studies are indicating, writing notes is still better.
• A personal observation on myself. I've noticed that when memorizing by rote (e.g., practicing a specific song on a piano), I tend to make mistakes when starting out, gradually become better, but then I again start making mistakes, and finally improve and settle. This may be related to [19].

While we are at this topic, here are some additional remarks that may relate the above characteristics of the human cognitive abilities to the observations in artificial intelligence developments and research.

• There may be a deep relation between the human/brain characteristics of forgetting curves and the machine learning technique stochastic gradient descent [16] and its batch size. This hasn’t been formally explored as yet though something called ‘hyperparameter optimization’ [17] essentially covers this for Artificial Intelligence.
• The Dunning-Kruger effect seen in humans “is a cognitive bias in which people with limited competence in particular domain overestimate their abilities” [18]. The effect is controversial; however, it may relate to the “double-descent” phenomenon [19] seen in modern AI [20].

Would people who type very slowly have the same benefits of writing by hand? :-) I guess, no. Those who type very slowly would be doing so because they haven't learnt typing enough yet. They would be having a higher cognitive load to type, which would rather distract from the new concepts being learnt. The situation may be different if someone could deliberately slow their typing down to spend more time thinking as they type.

What would we guess for people taking notes by speaking to their recording devices? Note that speaking is typically faster than both typing and writing. For me at least, speaking disrupts the internal train of thoughts. My guess therefore would be that taking notes by speaking would not be good for memory and learning.

References and footnotes:

1. Clive Thompson "How The Way You Write Changes the Way You Think", https://www.youtube.com/watch?v=89vzfTFu1Vw
2. Which is faster — typing or writing with a pen? https://www.publicationcoach.com/best-tool-for-writing/
3. The Pen Is Mightier Than the Keyboard: Advantages of Longhand Over Laptop Note Taking, https://linguistics.ucla.edu/people/hayes/Teaching/papers/MuellerAndOppenheimer2014OnTakingNotesByHand.pdf. Excerpt: “We show that whereas taking more notes can be beneficial, laptop note takers’ tendency to transcribe lectures verbatim rather than processing information and reframing it in their own words is detrimental to learning.”
4. Why Writing by Hand Is Better for Memory and Learning, https://www.scientificamerican.com/article/why-writing-by-hand-is-better-for-memory-and-learning/. Hacker News discussion on the article: https://news.ycombinator.com/item?id=39482641
5. Handwriting but not typewriting leads to widespread brain connectivity: a high-density EEG study with implications for the classroom, https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2023.1219945/full
6. https://en.wikipedia.org/wiki/Words_per_minute. Excerpts: “… the average rate for transcription was 32.5 words per minute, and 19.0 words per minute for composition.” “For an adult population (age range 18–60) the average speed of copying is … 8 wpm, with the range from … approximately 5 to 20 wpm.” “… the speed of handwriting of 3–7 graders” is somewhat slower.
7. https://en.wikipedia.org/wiki/Synaptic_weight. Under both neuroscience and artificial intelligence, a synaptic weight represents how strongly a neuron excites or inhibits another that it feeds to.
8. The same thing applies to machine learning too. Certain machine learning models basically just store the training data. When a good model generalizes, it is able to interpolate/extrapolate better to unseen situations, which is going beyond rote memorization.
9. “If I Had More Time, I Would Have Written a Shorter Letter”, https://quoteinvestigator.com/2012/04/28/shorter-letter/
10. https://en.wikipedia.org/wiki/Forgetting_curve. Forgetting serves multiple purposes including optimization of brain’s resources, generalization [12] via regularization [13], and helping avoid sunk cost fallacy [14].
11. https://en.wikipedia.org/wiki/Spaced_repetition
12. For Machine Learning, cross-connectivity would enable better generalization (https://en.wikipedia.org/wiki/Generalization_(learning))
13. https://en.wikipedia.org/wiki/Regularization_(mathematics)
14. https://en.wikipedia.org/wiki/Sunk_cost#Fallacy_effect
15. I personally tend to be very good at conceptual understanding, lesser so at memorizing raw facts though they say I am above average for that. :-)
16. https://en.wikipedia.org/wiki/Stochastic_gradient_descent
17. https://en.wikipedia.org/wiki/Hyperparameter_optimization
18. https://en.wikipedia.org/wiki/Dunning-Kruger_effect
19. https://en.wikipedia.org/wiki/Double_descent. Considering the number of synapses/weights in the brain to be constant, double descent may still happen in the brain via involvement of more synapses for learning a given thing.
20. Upon searching the Internet, it is seen that others have also wondered about a potential link: (i) https://news.ycombinator.com/item?id=28606752, (ii) https://www.dhirubhai.net/posts/srinivas-anumasa-phd-70b45324_i-found-it-intriguing-to-see-a-similarity-activity-7156256301491073024-Um55
21. Cover image generated using Dezgo AI via prompt "A woman's hand typing on a typewriter. Paper coming out of the typewriter. The paper has cursive handwriting with blue fountain ink. The writing on the paper is prominently visible. Bright lighting. Office setting. Photorealistic. The camera's focus on the paper." using model "JuggernautXL 5.0 (realistic)", random seed 3439325244 and landscape orientation. Upscaled to 1920x1080 resolution using Img2Go.