This is page is showing published content. Click here to enter preview mode!


Thoughts on conceptual learning, anki, books and physicality

Micheál Reilly-Sendal

Conceptual learning issues

I want to spell out what I see are some core issues with the learning of abstract ladder like concepts, think with mathematics (think algebra to calculus) and building on earlier ideas, or physics building up a more complex picture from simpler parts.

When learning a new topic, whether from a textbook, article or podcast, there are always aspects of the learning experience that have for me have been a struggle. When wading into a new field, you face an onslaught of terminology and nested abstraction. This concept relies on this other concept which sits on this core idea, all labelled and named with unique vocabulary.

Now additionally, I'm an avid user of Anki, the open source spaced repetition application. With approximately 20,000 cards, 95% created by myself.

So naturally to ease the pain of learning a new conceptual field, or at least mitigating some of the hardest aspects. I have attempted to use tools like space repetition to catalogue my journey and remind me of the core concepts. But outside of vocabulary and grammar for languages, I've found it incredibly difficult to maintain effective usage of spaced repetition for conceptual or contingent knowledge. Ideas that depend on other ideas, or formulas that build on some earlier intuition.

Why should this be?

It has puzzled me why this should be. For I receive great value out of space repetition in one area, languages, but outside of that it fails completely. I know it's not a memory issue, because with an enormous collection of 20,000 cards I have remarkably little retention issues.

So what do I find so difficult with Abstract ladder (AL) like knowledge when placed in the spaced repetition format of Anki.

Well, for one, the spatial relationship between knowledge is lost once placed in the context of a single lone card floating in the ether. True it's possible to organize the cards, but depending on your response, your ability to recall and the way you select the decks, the cards themselves will never have consisted spatial relations.

There are card specific adjustments like cloze deletions, whereby you hide some aspect of an image or sentence to then reveal it on selection. But these are card specific spatial relations. Not spatial relationships between pieces of knowledge.

What would a solution look like?

If we look outside the world of spaced repetition, there are hints at possible solutions. In the humble form of a physical book, with you have a natural spatial organization around the knowledge. And to a surprising degree when discussing with others on how they learn technical topics, a consistent point is the use of physical books vs digital books. The digital book, itself, is perhaps useful once the concept is understood or the general field is known. But the initial hard ingestion of information and ideas seems to be easier with a physical form. The book itself additionally if effectively written exhibits some degree of spaced repetition with writing in the later stages referring to knowledge and concepts from the earlier stages.

There are also form of knowledge progression and prior abstractions within the world of gaming. Even in the simplest of games, you have a game loop whereby you learn a new technique or concept and are then asked to use those ideas to confront a problem.
This then through the life of the game builds, with new ideas, mechanics and concepts introduced, leading to a heady build up of abstractions (so to speak) from the initial ideas.

Whether playing Minecraft or some city builder, or combo moves in platformer, there is a surprising degree of complexity and abstraction convened in this spatially arranged loop of progression, challenge and reminder.

Another area of effect learning for novel concepts is in switch contexts. Typically, when reading, one wishes to continue reading. You enjoy it in some fashion, there is a nice sense of progression, and you feel like you have achieved something after finishing. But for learning of concepts, we know you need to be questioned on the knowledge and challenged to use it. But unfortunately, the emotional pull to continue reading typically means that even though I know the exercises are useful, I just don't want to do them. It's boring, it's annoying, leave me alone and let me read my mind tells me.

But this experience can be short-circuited and this emotional reaction sort of tricked by switching of the context.

Take the example of reading a book about electronics, it describes the functioning of some circuit and some programming associated with it. You then proceed to switch contexts completely away from reading to physically building the mini circuit and programming it. Why does this feel less of a burden and less likely for one to reject or rebel against?
It's functionally doing the exercise to implement the concept or knowledge you have learnt.

I think the reason why it works so well is that the context has shifted enough away from reading vs. reading, an exercise to do. That your mind accepts that these are too independent steams of progression, with just some casual cross linkage. You are not doing the exercises and feeling judged by your poor reading comprehension of the previous material. No, instead, you are now creating something wholly different that can't be judged by the same metrics.

Conclusion for now

I think there is a possible way out, and a possible landscape of solutions, whereby you can retain the benefits of spaced repetition tooling and the benefits of spatial related knowledge found in the likes of physical books, while vastly improving over the experience of both. To hint at, I think the solution lays in using the physical intuition of your own mind, the flexibility of the computing platform and the interlinkage powers of large language models in AI.

But do email me at "micheal (not the irish spelling of michael) at" if you have thoughts too.

More Stories

Thoughts on Context and Conversation for Ideas

Micheál Reilly-Sendal