It’s Memorization All The Way Down

At the end of the day all learning is memorization.

Want to get notified about new posts? Join the mailing list and follow on X/Twitter.

“Memorizing without understanding” means only storing surface-level details in memory… in other words, not having enough information stored in memory.

At the end of the day all learning is memorization. It’s memorization all the way down.

Even learning to generate new ideas – it amounts to searching a space of possibilities, combining pieces of memory in ways that haven’t been combined before…

and you might say “aha, the skill of searching/combining is something other than memory,” but let me ask you:

When a someone trains the skill of coming up with novel ideas (e.g. a grad student learning to come up with research ideas that contribute to the cutting edge of knowledge in the field), where is that skill stored for future use?

In memory.

If they didn’t store it in memory then they’d never improve at the skill.

Memory isn’t just declarative, it’s also procedural.

You can’t just think of it as a SQL database that stores declarative knowledge.

Does it really matter? Isn’t this just a philosophical debate?

If you want to successfully learn you have to frame the problem as writing info to memory and the way to accomplish that is to use memory-supporting training techniques.

Lots of people try to weasel their way out of using effective training techniques by saying stuff like “learning is more than memorization, you don’t need to do active retrieval practice & spaced review because those techniques are only about memorization.”

They try to shirk the hard work, use ineffective (but fun/easy/low-accountability) training techniques instead, and justify it by hiding in the idea that they’re training some kind of learning that does not amount to changes in memory.

When in reality what’s happening is their training methods are not actually producing learning (or not nearly as much learning as would be produced by more effective training techniques).

What about metacognitive skills, and skills to apply information in different contexts?

These things are themselves represented as memory.

• Metacognition strategies are learnable skills and these skills would be stored in memory. (If they're not stored in memory, where are they? How would it be possible for a student to start out with poor metacognitive abilities, acquire better metacognitive abilities via training, and then retain those metacognitive abilities into the future, if they're not stored in memory?)
• When knowledge is flexible, generalizable across different contexts, that just means that the memory representation is more robust -- the memory has a more generalized structure rather than being some brittle representation centered around superficial details.

I’m not saying these things aren’t implicated in learning, I’m just saying that they themselves amount to memory, and the way you improve your ability to do them is by extending/improving memory representations.

But don’t metacognition and flexibility arise from the interplay between memory and other cognitive systems? Yes, you have to improve memory representations, but isn’t it also the dynamic interaction between systems that enables adaptability complex tasks?

When you improve the dynamic interaction between systems, how do you retain that improved coordination into the future, if not by memory? Where is the change happening if not in memory?

I think it’s easier to think about in the context of athletics – athletes improve coordination all the time and these coordination improvements are stored in memory.

What about mental calculation? Like, multiplying large numbers in one’s head? That seems more about dynamic calculation than static memory.

The extent to which it’s not described as recall, it’s a working memory loading task – but expanding one’s working memory capacity doesn’t seem to be a learnable skill so much as acquiring long-term memory that has such effortless recall that it effectively turns into an extension of working memory.

And while there’s a lot of dynamic calculation needed – there are strategies involved to reduce that amount of calculation, and learning those strategies amounts to storing them in memory.

What about instincts, like a baby’s babbling or a fear of heights?

Instincts are really interesting… but this is still just pulling information from memory, right? Just that the memory is part of the brain’s base installation as opposed to being acquired through experience.

But when you think of a new idea that you’ve never thought before, or take a new action that you’ve never taken before, and it’s not just a hard-coded instinct, then that can’t amount to just memory. Memory is necessary but we could not create new knowledge if it were only about memory.

My claim is NOT that all actions or thoughts amount to retrieving from memory – it’s a stricter claim: all learning amounts to storing information in memory.

Sure, when you think of a new idea that you’ve never thought before, or take a new action that you’ve never taken before, and it’s not just a hard-coded instinct, then that thought or action has not been written to memory. You are experiencing it for the first time.

But experiencing is not learning. Learning is when when you increase your ability to leverage that thought or action. Which amounts to storing the thought or action in memory and improving its memory representation.

Once you know a skill well enough, don’t you forget the low level details? Like an expert performer who is unable to teach the skill that they perform?

That’s not forgetting. That’s the complete opposite of forgetting. That’s the most solidified form of memory. You know it so well it’s second nature.

Want to get notified about new posts? Join the mailing list and follow on X/Twitter.