Learning is Remembering
by Petros Christodoulou • 4 minute read • #1 on Hacker News
You and I are going to try learning quantum mechanics together. I don’t know anything about physics and it doesn’t matter if you don’t either. Our strategy is going to be:
- Start reading the quantum mechanics Wikipedia page
- When we encounter a concept we don’t understand (let’s call it Concept 1):
- Go to Concept 1's Wikipedia page
- Read about Concept 1 until we understand it. Then go back to the quantum mechanics page and pick up where we left off
- When reading about Concept 1, if we encounter a Concept 2 that we don’t understand then head to Concept 2's Wikipedia page and read until we understand it
- And so on…
For example, on the quantum mechanics Wikipedia page we might at first not understand “subatomic particles”. We click on it and go to its page. Then on that page we don’t understand “composite particles” so we click on that, and so on, remembering where we were each time so we can come back to the right place.
Do you think this strategy will let us learn quantum mechanics?
As you probably know intuitively, it won't work. It will let most of us learn quantum mechanics for about 5 minutes before we feel lost and give up. Try it if you’re not sure what will happen.
All the information is on Wikipedia though so why doesn't learning like this work?
It doesn't work because our working memory has a maximum capacity of roughly 4. When reading about quantum mechanics we encounter new Concept 1 and store it in our working memory. Then when learning about Concept 1 we encounter Concepts 2, 3, and 4 and our working memory becomes full. We then cannot understand Concept 5. Our working memory is at capacity so our brain cannot combine Concepts 1 to 5 all together to understand Concept 5. We get the feeling of being lost.
This means we are severely limited in the new things we can understand and gives us the following rule:
You can only understand something if it combines less than 4 new pieces of information.
Worse than this, using our full capacity of 4 is difficult and requires a level of effort we often cannot provide. Most of the time to understand something it must be within 1 (or sometimes 2) steps of what we already know.
How does anyone learn quantum mechanics then? If our working memory's capacity is 4 and quantum mechanics involves thousands of new concepts how can we ever understand it?
Our working memory's capacity is limited but our long-term memory’s capacity is effectively unlimited (seriously, your long-term memory could store the entire internet as it was in 2016).
To learn more than 4 new concepts we must move some of them into our long-term memory before learning the rest. Incorporate Concept 1 into our long-term memory and we are able to understand Concept 5.
Learning is Remembering
Unfortunately, it is effectively impossible to increase your working memory's capacity. Given this, to get faster at learning you must get more efficient at moving things into your long-term memory, i.e. stop forgetting things you learn. The less you forget the more you'll understand and the faster you'll learn. You won't feel lost as often because fewer things will be multiple steps away from what you already remember.
If memory was a solved problem and we could instantaneously move new information into our long-term memory then we'd all be able to learn quantum mechanics by reading Wikipedia on a Sunday afternoon.
Memory therefore heavily influences our learning speed. In 2019 the UK school inspection body Ofsted went further than this and changed their definition of ”learning” itself to “an alteration in long-term memory”. I agree with them.
Learning is an alteration in long-term memory
Learning is intertwined with memory to the extent that they are almost the same thing. Learning cannot happen without a change to your memory. Everything we know about learning efficiently is directly related to memory - "good" teachers, "good" explanations, images, diagrams, maths problems, essays, practical assignments all are good for learning because they help move things into your long-term memory.
Memory's bad reputation
Some estimate that we forget 90% of what we learn within a month so there's much room to improve on. Why then do many people think improving your memory is inefficient and only helpful for “lesser” forms of learning? Why does memory have a bad reputation?
This story goes back to the 1970s. Unfortunately, schools in many countries around then prioritised extremely inefficient rote memorization of low-level information. This bad experience left many with the incorrect but persistent view that memory is unimportant. What was actually unimportant was the information they were forced to memorize, not memory itself.
The emerging efficiency of Google then tempted some to think you can "Just google it". Besides being slower and relying on you being able to remember what to google, me and you know another problem with this argument. If you have to google it then it's going to use 1 of your 4 working memory slots. A slot down and your thinking and creativity becomes more limited. Whole swathes of arguments, insights and creative leaps become unreachable to you. Googling something is better than nothing but you're literally smarter when you don't have to.
Some also discount forgetting as a problem because it doesn't feel like a problem. This feeling is not surprising because by definition you've forgotten the things you've forgotten. It's the unknown unknowns that are holding you back, not the known unknowns.
Lastly some know forgetting is a problem but don't want to acknowledge it for fear that there's nothing they can do about it. Well, new technologies leveraging techniques like spaced repetition mean it's much easier to remember what you learn so it's time to rethink that. You don't have to forget what you learn anymore.
Let's join the UK education sector in correcting memory's bad reputation. Learning is remembering and the easiest way to start learning faster is to stop forgetting the things you learn.