Thank you for writing it. I have been feeling something similar. The ability to truly understand requires the ability to know few terms and what they mean. I thought it was only me who felt it like so. Most of my friends who venture on creative tasks are able to recall a lot of things because of practice and without having to be distracted!
But does it effect the field like programming? When programming, if I can remember the context, then I can easily search it (research paper, books, documentation, forums etc) Now, with Co-Pilot, isn't it effectively beneficial to understand a topic and develop a general problem solving framework for ourselves so that we can let the AI do it's thing?
Even with programming I would argue you'd be a much better programmer if you can remember more. Obviously sometimes you're going to have to look things up but the more you remember the more problems you'll be able to solve.
Specifically you won't be able to solve a programming problem if the answer requires you combining over 4 things you don't have in your long-term memory (even if you can look them up). This is the main reason why Jeff Dean is a better programmer than me even though we both have access to google - he has more knowledge & experience of programming in his memory than me that means that even though we can both look things up he is able to solve way more problems than me.
Co-pilot slightly changes the type of thing that's valuable to remember, but it doesn't change the importance of remembering things. I think, as you implied, co-pilot probably makes remembering some types of syntax or boilerplate less important.
hmmm very interesting, not sure why that could be happening! I can see loads of people are reading it now but i would have thought our server (arranged on vercel) could handle it
Learning is memory, in a sense. More broadly learning the the physical structure that process leaves on your brain. When you truly learn, you do less remembering and more simulating. The process of simulating actions (multi-step addition algorithms in elementary school or wrote memorization of math-facts) is the process that changes the structure of your mind that isn't dependent on "memorization".
I would point you to two different motivating examples:
* Childhood development. A child who has had early childhood adverse experiences mind is effected. A fMRI or chemical detection can see physical differences in the brain. But the brain is plastic (it can change) over time. Works on this subject have been published by Dr. Dan Siegel https://drdansiegel.com/books among others. The key insight here is that academic learning is not significantly different at a core aspect then behavioral learning.
* Physical sports/martial arts depend on a reaction time much smaller then what is afforded by going through the full frontal cortex. "muscle-memory" isn't real (it isn't "memory" as you think of it). What you have in these cases are "short-circuits" (this implies structural changes) that are able to act before you are consciously aware of what is going on. The same applies to math facts and other fundamentals, you move things away from memory that needs to be retrieved and into reaction. Reading C-syntax for programmers or signing your name is something that has been turned into structure that doesn't need "memory".
I think your initial premise is correct. We have limited memory. How do we overcome that? We write. Writing is important because with it we can overcome our natural memory limitations. You cannot think about complexities (well or clearly) if you cannot write.
The danger of writing is that you can produce something that is both irrational and nearly impenetrable to the casual reader. For example:
> Foucault's use of the concept is descriptive, that is, analytical and explanatory, and at the same time normative and critical: he describes the grip biopolitics have on individuals through technologies of power in a way that makes manifest the repression at work in these biopolitical processes.
The above, taken directly from an "academic" published journal, could be said to have meaning. Unfortunately, each one of these words has an alternate meaning that is not normative to English, making the entire (actual) meaning opaque. "contecpt" "analytical", "explanatory", "normative", "critical", "biopolitics", "individuals", "technologies", "power", "manifest", "repression", "processes" are all defined differently then a standard English dictionary. So even if you can get past the convoluted sentence structure, the intended meaning will still elude you.
But the answer to memory limitations is clear writing using common definitions of words. I bring this up because as you extend your memory beyond what it innately has, the more likely you are to fool yourself (and others) with sophistry.
> Writing is important because with it we can overcome our natural memory limitations.
Very interesting. Is the theory here that by writing things out on a page we are then able to manipulate the ideas in our head without the usual limits of our working memory? Working memory is still limited but because all the information is so nearby and within view we can quickly put things in and out of our working memory so its limit doesn't impede us as much?
The mind, in this sense, is composed of mental models that we use to predict how our various actions might change the state of the world. A good example is driving a car [0] - when first learning, we lack an intuitive grasp on how turning the wheel might effect the car, but as we gain more experience we can simulate just how the car will respond right before we take the action, jarring us if we're wrong. This process can be applied to robots as well [1].
assuming our brain is a CPU, the 4-working-queue-short-memory is basically our L1-cache. We need L2 and hard drive for longer term memory, otherwise the system can not really function well.
however long-term storage is just one vital factor, another one is the 'deep learning' neurons that understand the content it stores and more important to connect the dots among various neurons.
we need both: understand and store. Neurons do both for us.
yes, we need understanding - the best way to get things into memory is to make connections between neurons through understanding. And this keeps it in memory for longer than if you just memorise it — and it will enable you to deploy the information in new contexts.
My takeaway from this article is that if you only focus on understanding (and so do not commit it to memory), you cannot reason using this information in unfamiliar contexts later on, once you've forgotten it.
So the best thing to do is to:
1. Make sure you understand something thoroughly
2. Test yourself on it using spaced repetition to ensure you keep it forever
For transparency: I'm one of the cofounders of Save All (linked site) alongside Petros the author
I think what the article misses is that there is a ton of knowledge we can't 'write down' and therefore memorization is not enough. For example learning to solve Integrals. Yes, there are some rules and tricks one can memorize which helps but I would argue the only way to get good at solving integrals is to interact, resp. solve them.
Another point are second order effects of how one learns, for example curiosity and resilience. There might be a long time negative effect on motivation of a topic, when there is too much focus on 'memorization' (It certainly was that for me in my French class;)).
I really enjoyed the 'bad reputation' part and agree that it is sadly viewed as not important enough by many.
There are plenty of different way to learn, and the optimal probably depend on the material to learn.
For example, in machine learning, there is something called stochastic gradient descent, where to learn you present a single random element at a time from the dataset.
In the end it will have learned of all concepts, by becoming more and more confident in each individual concepts.
For example to learn QM, you pick a random QM wikipedia article, and try to push through the article, even though there are some things you don't understand. Then you do the same thing, for a different unrelated QM article.
For learning tennis, you don't learn specifically forehand, then learning backhand, but you alternate them at random so that you have a single unified way of playing with smooth transitions, instead of having to switch between different "modes" of thinking.
Sure more memory can allow some speed-space trade-off in learning ability, but using your memory too much may make you miss some fluency that may have emerged. For example the old-school of machine learning was using databases and K-near neighbors, which used a lot of memory and was slow. But the new-school of machine learning are using constant memory algorithm and compressing the data in it, and it can learn to generate all the pictures in the world with only 4 Gb of weights.
Learning is imagining, once you bootstrap your imagination, its bandwidth to synthesize new examples from which you can learn from, is much greater than the bandwidth of looking up new data material to learn from.
It's not bad, but you're limiting yourself "out the gate" so to speak. If you really want to explore memory and learning you must study and practice (self-)hypnosis. Most of what we think we know about how the brain and mind work is wrong. E.g. memory: "our working memory has a maximum capacity of roughly 4" is not an absolute rule. Your brain, right now as you read this, is tracking 100,000's of variables and recording a trace of everything you are seeing, hearing, smelling, tasting, touching, and doing. Almost all of this is done unconsciously, and for good reason.
Anyhow, you can learn to have better "interfaces" to your automatic unconscious abilities and leverage them to e.g. remember instantly and durably any mathematical equation, etc.
- - - -
As an aside, the strategy you sketched out at the start of your article would be workable if you have great recall, however an even more better strategy would be to recapitulate the discoveries of physics in roughly historical order. Your understanding of quantum mechanics would be much deeper and richer, and you would be following a story (an epic story made up of so many fascinating smaller stories, and one that is still going on! Albeit things have calmed down this last century or so, but no one thinks we have reached the climax yet.)
That would be the way to do it: start with the Greeks and the Alchemists and proceed to follow the trail(s) of how we as a species sussed out the mysteries of the physical Universe.
I am utterly baffled that no response in this thread so far has taken issue with the statement "As you probably know intuitively, it won't work."
For me, this _does_ work and I have proven it many many times over the years by adding entire categories of technical knowledge to my repertoire. And not superficially, either - I get paid very well to do things professionally that I taught myself by reading Wikipedia.
If my experience were commonplace, the "it won't work" statement would be highly contentious in the comments here. Since it isn't, I guess I can deduce that I must be an outlier.
IMO, such a viewpoint about "learning" is limited to learning 'knowledge'; i.e. this sort of 'learning' is limited to repeating (replicating) facts external.
I wonder, if learning is remembering, then what is "understanding"??
from my own viewpoint, learning is about something external; for example "what's the word for such and such concept?" ..in english or in spanish?
point being that you need a corpus of consensus about what the specific linguistic-culture calls the learned concept.
but then, what does it mean to understand?
I think the way towards making sense of this (answering it) needs to consider learning of physical (do-able) actions. Because when considering such skills as learned/understood, the distinction between learn/understand seems to vanish.
So then maybe understanding has more to do with having learned something to a proficiency level that allows one to teach (show/explain) to another how to do that action?
finally, to throw a proverbial wrench into my own attempts to make sense, what does it mean to perceive something complicated, such as the meaning out of arbitrary alphabetic glyphs? how is the meaning out of a text understood? what did we have to learn to be able to do it? is it just a matter of knowing most of the contents of a dictionary??
This is cognitive science in a nutshell. Most early philosophers argued exactly what "wisdom" means both externally and internally. Look to the Socratic philosophers for simplified explanations.
I was brought up with this distinction - my family always put more pressure on understanding than remebering but now after mamy years I start to suspect that there is really no difference between those two - understanding is probably just remembering proper models that are useful to solve problems that You want know how to solve.
remembering vs understanding is uncompressed vs compressed storage:
If I tell you a sequence of numbers:
1,2,4,8,16,32,64,128
And you try to remember them, having never seen these numbers before, you have to remember each individually and it will not be so easy.
But if you, before trying to remember them, apply a little computing power to figure out that its a sequence of powers of two, starting from 2^0 going up to 2^7, then you have compressed the information I gave you. You understood (presumely) the source of the information and you will be able to remember the numbers much easier.
One strategy I see people applying to unknown data they want to remember is to try to establish links to already known information or made up stories. For example when given the sequence of numbers above but not knowing about exponentials some people would try the follwing:
* 1: the first number is one (as on the number line)
* 2: the second number is two (same)
* 4: I have four friends
* 8: I ate lunch with my friends
* 16: my sister was also there, she is 16 years old
* 32: the house number of the restaurant was 32
* 64: we ate sushi, my dad also likes sushi, he is 64 years old
* 128: one-two-eight sounds a bit like "want to eat" and yeah I also like to eat
By doing so some people seem to achieve quit good memory of an unknown topic. But from my point of view they are only re-encoding the information to sort it into already existing bins in their memory instead of compressing it. The amount of information is not reduced but increased and it seems harder to reconstruct the original encoding/information. Additional without compressing the numbers to their generating algorithm it is not possible to use the "learned" knowledge for anything but reciting.
This all leads to Solomonoff's theory of inductive inference.
It's not wrong, it just doesn't cover much breadth of learning nor how humans learn over time. Also it doesn't consider different learning styles. It's seeing learning as just chunks of conceptual information where that is reducing the problem too far.
It's not just some simple "move from short term memory -> long term memory" to make more room. If it was, we would all optimize for that outcome. People would have written books about learning that glorified this concept. Teachers would be teaching it in schools to have students score better on tests.
Learning is much more a lifelong mindset akin to the famous Socrates quote of "I know that I know nothing". Or even the idea that we change through the books we read / things we learn, but don't remember much of what we did.
So I don't agree it is all about remembering because like GI Joe said, knowing is half the battle.
You have a preference of how you learn, no? It's not a myth if there's perennial truth to it.
> There are lots of books on it actually. Make It Stick is my favourite book on this, you should try it!
I've read many titles including this one. Not everything is going to be "learned" with spaced repetition, interweaving, retrieval, and varied practice. These are great modern methods to learn effectively for the short-term, but are not by any means concepts you're going continue practicing past your formal education. (i.e. Anki flashcards for everything you want to learn)
Perhaps, but if you have a preference for one way of learning as opposed to another, you're probably more likely to stick with that preference and learn through keeping up the habit of learning via that preference. If it's a method you don't enjoy, you'll probably just drop that topic and not learn it – not necessarily because the topic itself was the issue, but how you went about learning it.
It may also be interesting to consider this in the context of learning physical skills. When you're learning something new, there are a lot of things that you are doing wrong. A not-so-good coach will see them and start telling you things that you need to change. But you can't remember all these things. A good coach will find one key thing which you can keep in mind to work on.
The Inner Game of Tennis frames this as a difference between the two selves. Maybe that could be considered through the lens of decreasing the number of things you need to remember in order to improve.
I’m glad you brought up this point. I think there are analogues to “muscle memory” when it comes to learning concepts, too.
For example, I don’t remember all the techniques to compute integrals explicitly, but I know that I’ll be able to pick it up quickly, because of my “muscle memory”.
I feel like this falls into “remembering”, but not “memorizing” when it comes to definitions.
yes! i think that's completely right, it takes so much energy & working memory to adapt to feedback that it's really easy to over-instruct people and give them too much feedback to handle.
> Maybe that could be considered through the lens of decreasing the number of things you need to remember in order to improve.
This! learning materials who master this way of information giving based on their importance is key to effective learning, and I think a lot of them don't give importance to it.
The best coaches I have worked with have cues they use instead of, or with, the explanations. Things like "pinch your shoulder blades" and "push the floor away". These give the practitioner an understanding of how it should feel, physically, without the need to go into details about which muscles and joints are involved.
The good news is that memory techniques have never been as accessible as they are now.
I am fully in agreement with the conclusions drawn in the article, but the bad news is that even with those techniques it can be a slog. Any Anki user will tell you that maintaining dedication and avoiding burn-out is your Achilles' heel, not the limits of your human ability. Understanding something for the first time (in the Feynman technique sense of being able to explain it well), and doing that multiple times per day, takes up a lot of mental energy even if you are smart and naturally talented.
In conjunction with using memory techniques, we need to add dietary practices where we become much more selective with the information we take in. Places like HN give the illusion of learning (and to a great extent help broaden your mind about certain topics), but the actually utility of all that random knowledge butts up against the opportunity cost. There are already many more worthwhile pursuits than can be fit into your lifespan.
Completely agree on Anki requiring too much effort, its the main reason why i'm making Save All (https://saveall.ai/). Save all is a simpler version of anki that use AI to try and make it less effortful
All it takes then is just move from a method of remembering to another more efficient and enjoyable one, I can illustrate what I am trying to say using your Anki example by saying that once a language learner for example reaches a certain level in the language after acquiring a sized set of vocabulary, the learner can move to "abundant reading" as a way to memorize words through frequency as a more effective method of learning new vocabulary.
So it's all about remembering, but what differs is how someone approaches it depending on their level and understanding of the topic they are trying to learn.
This is definitely true. The snowball effect and the networking of information helps a lot.
I'm trying to look at the problem with a wider lens, however. For example, if learning that language is actually something we want to do in opposition to all the things you could be doing. In the context of public education, since this is the theme the author focuses on, we don't just study certain topics, we study a certain spin on some topics that is determined by a range of government officials.
In other words, you have to deal with severely limited energy and interest compared to all that is available in life so cutting chaff is probably even more important than boosting your ability to remember. In fact, selecting what to learn is life, just like a sculptor removes the parts of the marble that aren't in the end result.
I agree with most of whats written here. I think there is a major point you're overlooking here. There is simply just too much information we're expected to know nowadays. You could argue that the amount of information we have to cram into our brains in such a short amount of time makes it very difficult to build good long term memories. High School and university curriculum's are forever expanding jamming more and more in the same period of time. Even in professional settings such as being a web developer or data scientist for example where new things are being invented by the second that then become the standard. At some point you'll need to offload long term memories to external sources. So while you may forget the actual content of what you need to know, you could for example remember what to google and the summary of the content you expect. Essentially our long term memories is transformed into pointers for external information banks. We only keep whats absolutely essential to perform our functions.
The problem is that if you don't commit information to long-term memory you can't use it reason effectively in other contexts, and you have to add it back to your short-term memory every time you look it up -- so outsourcing your memory to a knowledge bank is limiting the complexity of the tasks you can handle.
So there might be more information you're expected to know in modern jobs -- but if you spend a bit more time consolidating rather than acquiring new information, you can build the foundations on which more advanced skills can rest.
A key point here is that our brains don't work like computer hard drives. Our brains are a lot closer to how, in biology, a single cell stores the entire DNA "data" that's needed to replicate but just using a few base pairs.
We likely store information more in some type of loose graph structure, where we recall / "remember" something by re-creating links to that piece of information. There seems to be very very little "storage cost" for the billions of pieces of information we keep in our brains.
> There is simply just too much information we're expected to know nowadays. You could argue that the amount of information we have to cram into our brains in such a short amount of time makes it very difficult to build good long term memories.
I would also argue that the time we spend memorizing "facts", of questionable utility, takes away from the time that could be spent on learning better methodologies for thinking.
The words used make the whole thing very confusing. Understanding and remembering are two very different thing in cognitive literature. Understanding is deeper about connections, and remembering is more recollection of facts or knowledge. It is fair to say that you cannot understand something without having some knowledge, but you can certainly know a lot of facts without having the slightest understanding.
It also poses wildly different challenges, where the metrics by which one is to judge the "degree of having attained knowledge" in the broader sense of the word, depends entirely on which of those aspects one value.
Some literature refers to this dichotomy as "instrumental" or "relational" understanding. You see this very clearly in math, where students can recall the facts of equations, but they don't understand it.
It can very well be that the author of this article is aware and appreciates this distinction. But, the phrase and title of "learning is remembering" will certainly evoke suspicion. Just because it is much easier to remember something one understand, does not mean that by remembering, one understands.
I agree with the core idea of your comment, but I think the author is as far as I understood it think of learning as "fully understanding", so if you didn't fully understand it then you didn't learn it yet.
Per the author, they’re launching a service called “Save All” — which is simplified version of spaced repetition; might be worth explicitly stating this in the intro or footer of the linked blog post.
I didn't mention it in the article because the article is hosted on the same domain (with a link to it at very top of page) so thought it was quite obvious.
Also I wanted it to stand alone as an essay to see what people thought. If i linked to the product within the article it would turn the whole article into an advert rather than an essay
Okay, but you don’t even mention spaced repetition until the end and only in a single sentence.
And obviousness is relative; I did not even notice the logo and base domain, read the title and scanned the text for core concepts, spaced repetition again is only mentioned once, even though in my opinion it’s literally both the topic and your solution to the issue; which was not obvious until I read your comments. Think if you were to ask random people who are not aware of the company, what the post is about — then define spaced repetition and the reasoning behind the company’s name, then ask people again what the post is about — answers would be noticeably different; you could even hide the context and ask if concept of “save all” was mentioned on the page and what it means.
Spaced repetition is not the topic. The topic is memory and there are many solutions to trying to remember more, spaced repetition is just one of them. There's also things like elaboration, dual coding, mnemonics, memory palaces etc.
>> Well, new technologies (Save All link) leveraging techniques like spaced repetition mean it's much easier to remember what you learn so its time to rethink that. You don't have to forget what you learn anymore.
Might be wrong, but at the point topic is covered and a single solution is presented, it becomes the topic.
“[…] the purpose of memory isn’t to remember the past. The purpose of memory is to, at least in part, so that you don’t repeat the same errors that your repeated in the past [..]”
– Jordan B. Peterson
I like to think that Spaced Repetition learning pre-empts these errors, so we are apt to recall when we productively need that memory.
Whoa.
That quote is kind of amazing.
My brother committed suicide, and sometimes memories of the past really haunt me.
That quote is actually really comforting.
How do you trust your memory when (as I understand it), every time you recall something your brain changes in ways that can affect recall later?
Is there a distinction between learning and knowing? For example, in my undergrad days I learned about Simpson's rule for numerical integration. I jammed enough of it into my memory to pass the test and quickly forgot all the details. Now, 30+ years later, I needed to calculate the volume of a pretty complex space. I remembered the existence of Simpson's rule but absolutely nothing else about it. Looking it up on Wikipedia I was able to re-learn it well enough to apply it in my job and move on to the next problem.
If, over the past 30 years, I had been using flash cards to remember the details of Simpson's Rule, I would have wasted a lot of time. Re-learning it when needed also means I don't have to rely on my faulty, dynamic memory. For me, it seems like there's a sweet spot to remembering enough to know the concept and then relying on the internet to fill in the details as needed.
Firstly, in terms of the distinction between learning and knowing -- the thing that matters most is the strength of the encoding in the brain. If you just memorise something with 0 understanding, the connections in the brain aren't as strong -- so they disappear. Whereas if you know something thoroughly, the connections are much, much stronger.
These strong connections are why when you go back and look at it, you recognise it and you know how to apply it - because you still have some of the residual memories from this strong encoding. But in the meantime, you probably haven't been able to apply it in an analogy for example.
Secondly - there's a classic on the topic of Spaced Repetition written by Gwern.[0] Gwern calculated that, given the average amount of time you spend testing yourself on something, and the exponential increase in how long you remember it, if you would spend more than 5 minutes per 10 years looking something up, you should use spaced repetition to remember it.
> Firstly, in terms of the distinction between learning and knowing -- the thing that matters most is the strength of the encoding in the brain.
I'd vote for the ability to perform a skilful epistemic analysis of the retrieved information being more useful. I prefer this because it can overcome any natural immutable shortcomings in the underlying process.
> every time you recall something your brain changes in ways that can affect recall later?
Yeah that phenomenon is called "proactive & retroactive interference".
> For me, it seems like there's a sweet spot to remembering enough to know the concept and then relying on the internet to fill in the details as needed.
I think i agree with you to an extent. For example, there is no point memorising all the digits of pie so that you can use it when programming, it's much more efficient to just remember what pie is at a vaguer higher level than to put the energy into memorising all its digits.
But i would say that most people go too far the other way and only remember less than the efficient amount.
I experienced this myself when I was a kid. I wanted to do mental arithmetic like my dad but I couldn't do multiplications for large numbers even after my dad explained how he did it. Turns out I had to memorize the time tables first, otherwise I would run out of swap space.
If you want to experience this yourself but you already know the time tables, you could try to memorize log tables and then you can do those fancy arithmetic tricks that old school engineers used to do with slide rulers.
Calling learning remembering, is like calling a computer a hard drive. Sorry, but, remembering is just one part, although important part of learning. There is loads of difference between remembering the words of a foreign language and fluently speaking it. The process of learning is complicated, eg. good teaching is not giving the final answers to learner, which should seem faster, but letting the learner arrive at their answers, even sometimes arriving at a wrong answer at first.
And the point about reducing forgetting, that's also a terrible way of learning. If you want to minimize forgetting, that means you have to learn in baby steps, repeat everything using some kind of spaced repetition algorithm, and endure the pain of doing all this for a considerate amount of time. Instead, learning the content rapidly (meanwhile forgetting a lot of the content), getting a whole picture, and gradually gain more understanding repeating the process using different textbooks/courses, is a much more faster way of learning. The brain is more used to BFS (breadth-first-search) than DFS (depth-first-search), and the whole process is much more stimulating. This is because new knowledge needs to be encoded with connections to pre-existing knowledge, and most knowledge is intertwined together, for example in calculus, limits, derivatives, integrals, infinite series, etc. Learning limits in isolation of the whole picture, optimizing for less percentage of forgetting, often leaves learners confused, why am I even learning this, and what use is this for. Even though the percentage of forgetting is much higher in the latter holistic BFS process, the overall content mastered in the same month or week duration is higher. So forgetting is actually normal and you shouldn't panic when you forget things- since you will gain more understanding each time you learn and re-learn the content.
For the last ten years, I've been making products for learning, and every day now and then, I'm still learning something new about learning.
> eg. good teaching is not giving the final answers to learner, which should seem faster, but letting the learner arrive at their answers
Why is that "good" teaching? ---> Its because if they arrive at the answer themselves then both the answer & the process for figuring it out will be more ingrained in their long-term memory!
> Instead, learning the content rapidly (meanwhile forgetting a lot of the content), getting a whole picture, and gradually gain more understanding repeating the process using different textbooks/courses, is a much more faster way of learning.
I agree to an extent. It's inefficient to try and remember everything beyond a point. But spaced repetition is very efficient e.g. with Save All it might only take you 5 minutes to remember something for 10 years... so it is much more efficient than you think to try and remember more things.
>There is loads of difference between remembering the words of a foreign language and fluently speaking it.
And alternatively put, there's a difference between learning to read, and learning to speak. Yet both would boil down to primarily rote memorization, as languages tend to do.
Language is not a good example to prove your point on.
I think the point being made is that remembering is a huge if not key aspect of learning anything, and people underestimate the importance of memorization. Maybe that title is a bit misleading, could've been "No serious learning can happen without memorization" but that doesn't flow as well.
If anything the problem used to be that we overestimated the importance of memorization. Maybe it's changed now, but trying to free cognitive load is important. It's the same reason why literary societies beat oral ones. The emphasis on memorization was reduced.
I 100% agree. This is true in my experience as well. Is your content available in English? If not, can you recommend some books for derivatives/ integrals or statistics please
A less eloquent but probably more helpful translation/phasing might be:
You have not truely mastered something until you've forgotten the step-by-step process of it (and presumably had to reconstruct it by doing it and paying attention to how you do so).
(This is also why it can be hard to teach something you're very good at - often, you literally don't know (well, remember) "how to do it" in a form that's actually useful to them.)
Learning is a lot of things, fact collection is one small part.
Learning is mostly model building. Building a model to predict future sensory inputs from past sensory inputs. Building a model to predict future sensory inputs from past sensory inputs and control inputs.
Learning is not just recording sensory inputs for future recall, it is taking them and building useful abstractions with them.
This is why abstraction is so important (came here to say the same thing).
Wikipedia is factually correct but often lacks insight. It puts the learner at the wrong level of abstraction, limiting how much more can be learned.
For example, a sine wave looks complex, and has a great deal of inherent complexity around stuff like transcendental functions. But it's just a spiral, the side view of a radius arm turning through time along the x axis with a period of 2 pi radians and a radius of 1.
But if readers don't know that, they get stuck at the abstraction of trigonometry instead of the far deeper relations between things like complex numbers and higher dimensions.
That's why I think it's difficult to learn quantum mechanics without a teacher. It just ends of being a bunch of matrices and handwaving that makes little sense intuitively.
This is why the debate around higher education is silly IMHO. Sure, someone can avoid college and get hands-on experience in application. But they'll miss out on the theory and abstraction that allows them to transcend their area of expertise. That's good enough for most people, but most likely won't result in true mastery. No schooling is not better than schooling if one wants to do important work.
Your brain is neither BFS or DFS. Your brain is a biological neural network. It is associative, that is, your brain follows paths of relationships. The brain actually is "like calling a computer a hard drive" in that memory and processing is wrapped up in the same mechanism. "Thinking", "knowing" and "remembering" are not separate things for a brain, they are all flavors of the same thing.
You wrote "learning limits in isolation of the whole picture, optimizing for less percentage of forgetting, often leaves learners confused". The is true. The reason is because, if you do it in isolation, you're not forming connections to related knowledge.
The main split in brains is short-term memory vs long-term memory. And I suspect (speculating here) that forming more connections to items in long-term memory helps in moving a fact from short-term memory to long-term memory.
When I study something, I go for awhile, but eventually it becomes difficult, confusing, hard to see the forest for the trees. Particularly with technical information and skills like programming (or natural) languages.
When I come back a little bit later, I find that I only remember the things that made sense; my confusions are forgotten, and there is fresh mental space and energy to master a bit more of the terrain before I need another break.
Our subconcious mind often works on our problems while we aren't thinking about them, i that's more what's going on here than forgetting. your brain figured it out for you while you were doing something else
That's sometimes true but I think my point still stands.
When I'm studying a foreign language, I learn some words and they stick, but I'm exposed to a bunch more that I don't remember next time. I forget those meanings, but the ones that stuck are now vivid and with me, brighter.
When I'm studying Kubernetes, I end up reading a ton of information that's irrelevant to the task at hand, and lots of it doesn't make that much sense because I'm new to it. The next day, when I come back, the things that I actually understood remain, ready to be the foundation for new learning, which they couldn't have been when they were mere data points in an overwhelmed brain. I don't remember the parts I was confused about yesterday, just this stuff that now makes sense.
To unite our points, I might be thinking of something like: the immensity of sensory and cognitive data that pass through (sub)consciousness during the learning task are sifted and sorted in the unconscious while not learning; one might call the sifting "forgetting" and the sorting "figuring out".
I'm still staggered how potent a post effort pause can get. I can spend hours and days trying to improve something and not being able to see the easiest spots. I come back 4 days later and everything just jumps out as obvious as day. No confusion, no fatigue, lots of ideas, enthusiasm, creativity..
Um, there’s like 300(0) years of research on this, and, really, it isn’t anything like this simple. For “recent” science you can start on volume 1 of the jep journal of learning and memory, which is like 50+ years of continuous monthly publication.
“Some simplification”, even if it is technically correct can be misleadingly incorrect in the sense that it misleads you into thinking that the object of analysis is simple, which is incorrect. Analogous example: A computer is a machine that you stick numbers into, and it does stuff to them, and other numbers come out. But you can only put in a limited number of numbers because it can only do stuff to a handful of them at a time. The rest need to get saved someplace else, called RAM. So machine learning is RAM lookup. Is that misleadingly incorrect in the sense I described?
Learning basic QM is 90% complex differential equations and 10% physical intuition.
There is a reason why you generally don't get to introductory quantum mechanics until second year after (or while) you're doing calculus, differential equations and linear algebra.
To quote a lecture I once saw: the deepest point of any state is a mine shaft somewhere. You don't find that shaft by gradient descent.
As someone who has been using Anki for the past 4 years to learn Japanese and now Chinese, I've recently found that the initial "learning" step is not that hard. (i.e., given 辞書, I remember "dictionary") What is hard is keeping the content I've learned, _learned_, for more than a day.
So now I'm wondering, is my initial "learning" process wrong? Since definitions are fairly simple and I find them easy to remember at first but hard to hold onto. And my process in Anki for New cards is to rep them like reviewed cards until I pass the card; so really, I'm just staring at the kanji + definition until I remember it.
This brings me to my question: is there more Anki can do when it comes to learning new cards? Is there something we can do when learning new information that will help make it "sticky"? Especially in regards to "simple" facts, like basic kanji -> definition mappings, where there are no mechanics to understand, just simple mappings.
Edit (an addition):
Also, I love SRS, but I don't understand how it can be advertised as completely different than rote memorization. When you learn a new physical flashcard, you're learning it the same way as you would be when learning a new Anki card. The only difference is that Anki will show it to you again at a more efficient time in the future, rather than at some regular interval.
I learn a lot. The way I've learned best for me is to mimic children. Especially small children.
They focus intently at seemingly incomprehensible complexity. Then they play with the small fragments they can get their hold on. "ba-ba-ba-ba" when they're learning to speak.
They play. They combine two (or more) different things they're interested in into a single experiment which we think of more as play. Mushy food? I wonder what happens when I throw it on the floor? Oh look how it landed all funny! Look how mum and papa reacted!
Grinding through cards learning languages never really did anything for me. Immersion and regular (but not exhaustive!) play stretched over time makes the neurons of my mind much more reliable.
Learning is often a byproduct of play, but the actual point of play is highly subjective to whoever is doing the playing. Although, in my experience, if you need to learn something, making it into a game is a highly effective and fun way to go about something that might otherwise be pretty dull.
Anki is great, but it doesn't hold a candle to full sensory integration. If you moved to Japan or China and rid yourself of English, you'd be forming all sorts of associations a white screen with text can't replicate.
SRS is great for the characters -> definition pathway, but we need so much more. Listening, speaking, interacting, using the language to describe the world around us in conversation and navigate through it.
Yeah it needs to help you do better "encoding". This is what determines how sticky information is straight after learning it.
One way that Save All (a company i run that's like anki, https://saveall.ai/) helps you with encoding is that it suggests alternative ways that we can quiz you on a card you just made. This helps you engage with the card a bit more as you're making it which makes it more sticky.
Other than that a good way to also improve encoding is to link new knowledge to existing knowledge. Things like Roam help you do this with their backlinks
I thought active and free recall were both superior to mixed encoding and interleaved practice. Is there a study linking more encoding to enhanced active recall? From what I understand, the suggestion is to add desirable difficulty (which I guess leads to better encoding, but different encodings dont nefessarily add desirable difficulty).
Not sure about the relative importance, but save all also has active recall & interleaving. The encoding happens when you first learn something & create a card, and the active recall, spaced repetition & interleaving happens later when you review cards.
It seems like you've gone pretty far. I assume you started with something like radicals (constructive strokes) and built up to more complex Kanji/pictograms. I know in speaking to native Mandarin speakers that they see stroke order in the characters and were intrigued that I don't see that in english. Often we just learn the sound/typed elements (e.g. pinyin), which I guess might be like modern asian kids.
Have you looked at something like wanikani.com with the mnemonics or some of the more historical derivations of the different characters? Those might help build an internal story for why those characters mean what they do. As an adult, that helped me.
Anki is great for higher level learning as well. Instead of having just individual words, have full sentences or even paragraphs. If you can read it aloud and understand it, then you can mark it as learned, but if you need to use the definition or explanation for any of it, mark it as unlearned. Once you have a decent initial vocabulary and grammar, this is far more effective at pushing things into long-term memory. Most of language is contextual, so anything to increase the amount of context on an individual card will help with getting better holistic knowledge of the language.
Fluent Forever book has helped me a lot with Language learning. It shows you how to learn a language mainly using flashcards, apart from other things.
I'm recommending you some things that have completely changed my experience using Anki for language learning:
- Don't use translations. Use images instead. For learning how to say dictionary put a picture of a dictionary. It will be easier to remember. Also, you usually won't find a perfect translation for a word.
- Use cloze cards for grammar. Instead of learning the rules, understand them and put four sentences with placeholders and repeat them. This process will make this way of constructing sentences stuck in your head.
To learn a word or character, you need to be able to understand it immediately and be able to form a sentence with it without delay (at least not more than the occasional "uh..."). Remembering the meaning after some fiddling is not good enough.
That has to be achieved by actually practicing the language, either via listening, speaking, or writing.
Anki is there to help after that step by keeping you refreshed with the exact tones, strokes, multiple pronunciations or more ambiguous CJK characters etc. but it can't be the motor behind the process.
For spaced repetition, although it is a form of rote memorization, the key insight is that memory is formed through recall, not review. The system then (purportedly) hits you at the very moment you are about to forget the card. It's counter-intuitive but it's easier to encode things in long term memory by recalling them at that point as opposed to when it's still fresh in your mind.
This is analogous to studying in university via explaining things to yourself versus just highlighting and rereading the textbook. Every student knows the former is more effective.
A friend who is amazingly good at languages made the analogy, you can't read about how to play a musical instrument and get better at playing it. You actually have to play it.
Anki might help some, maybe reading, but speaking and listening require actually speaking and actually listening.
The analogy can be taken too far; if you can recognise a note by sight on sheet music, you’ve learnt everything there is to directly memorise in paying an instrument. All the rest is practice. This is not the case in language learning. A language which is not closely related to one you already know has tens of thousands of words, the meanings of which can’t be deduced from other things. There is no use to doing listening practice if you do not understand almost all the words being used; you will be stuck looking them up one at a time in a dictionary. It is also very difficult to correctly pick out words you don’t know in a language you are new to.
I think you are understating the difference this "more efficient time" makes. SRS outclasses every other approach we have for long-term memorization by several zeroes.
Bringing your car to the shop when the "check engine" light is on and bringing it into the shop after every single drive is superficially the same, but only one of them makes actual sense...
The answer is yes. I’m working on it with https://reader.manabi.io which aims to enrich memorization with more colorful contextual anchoring. Working on a SwiftUI rewrite with Anki integration currently.
> Is there something we can do when learning new information that will help make it "sticky"?
For me, a big part of learning a new word involves creating chunks. Those chunks then form bigger chunks. Let's use 辞書 as an example: 辞 is talking/words (and also used as kanji for quitting) and 書 is books. So word book = dictionary.
The kanjis are also broken down by radicals. 辞 is 舌 (tongue) + 辛 (spicy). There isn't a logical connection between the combination of radicals and the kanji, so I'd make a memorable story instead [1]. Thus the only things I have to purely memorize are the radicals.
I go through above process when I'm creating a card and I believe the process itself helps me learn the word much faster than just rote memorization. Of course, you need to start using the word in your daily life too!
Edit:
[1] When I create a story, I close my eyes and try to imagine the story happening as vividly as possible. I also have a rule where radicals in the story must appear in the same order as in the kanji
I dont understand why some people still go by brute force memorization when there are better methods. Those that our brain naturally uses. You won't see memory champions using Anki for competitions as their main memorization tool that's for sure.
I've studied a few languages, and by far the easiest, fastest and most long-term proof way to memorize words is by free association.
It takes some initial effort to be able to create these associations, but when done, it's done. There are some words I now basically can't forget, because the association is just too strong.
Example:
Waarschuwing = dutch word for warning. With free association i did it like this.
the "waar" part of the word immediately made me think of english word "war"
schu, shoe
wing, wing
So i vividly imagined a flying shoe with a trompet warning about an upcoming war. Done. Unforgettable for me. The more ridiculous, emotional (funny, stupid, ...) and obvious the better.
Some words are easier than others, and the ones that don't have a quick association jump out quickly are at a higher risk of being forgotten, but even then, they will stay way longer and without much less effort than with brute force, which is basically just creating memories out of nowhere, with no associations to anything on your brain. Our brain loves to relate things to others. Unrelated things = useless things. Everything that is useful has a relationship to something else.
So when you try to memorize by a way that is not based on association, indeed, only brute force works, which is a way to tell your brain "welp i can't really tell you what's the usefulness of knowing this, but somehow it keeps coming back into consciousness, so i guess i need to keep it in here somewhere. A self sustaining island, a dot of knowledge with no connection to anything else i know or care about."
> I dont understand why some people still go by brute force memorization when there are better methods. Those that our brain naturally uses. You won't see memory champions using Anki for competitions as their main memorization tool that's for sure.
Thats because Anki isn't for memorization, it's for time optimization and scheduling. Nothing about Anki makes you remember a fact better, that's still on you, and the memory champions use ALL those tricks and techniques.
Anki just schedules the non-memory-champions needing to mostly remember most of the things, and minimize the time taken to try and do that.
Said as a multi-year daily Anki user getting really close to a 1000 day streak.
Some other things to consider. A working memory of 7 is a median number for the population. Some folks will have a greater working memory (some outliers are in the 80+ range). Some folks will have a smaller working memory.
Specifically, someone with ADHD (a topic close to my heart), the size of working memory is typically around 3 items. Related, long term memory is also a bit more chaotic for those with ADHD; when you can only have three linked concepts at a time in working memory, it's going to be encoded with fewer overall links data.
That all is to say, if you're building a tool to help people remember things, don't just build it with the median in mind.
If storing information and creating associations between bits of it were 'learning', then databases would be very smart indeed.
No, learning is not 'storing pieces of information in long-term memory and recalling them'. It's not the ability to recall information. At the very least, learning is information+behavior change+understanding+values and attitudes associated with the information. It's much more complex than memorization and recall.
Thanks for sharing this and I've been certainly thinking along the same lines.
To add to this, I've known many folks that can accomplish certain tasks almost automatically and creatively. If you asked them to recall exactly what they did to achieve it they couldn't. And this usually isn't action on concrete information either but on intuition alone.
If humans worked primarily on memory we'd have been toast a long time ago. There's too much variation in the natural world to confront it solely on the basis of memory. I'd say we're more experientially oriented as opposed to memory oriented
240 comments
[ 4.6 ms ] story [ 269 ms ] threadBut does it effect the field like programming? When programming, if I can remember the context, then I can easily search it (research paper, books, documentation, forums etc) Now, with Co-Pilot, isn't it effectively beneficial to understand a topic and develop a general problem solving framework for ourselves so that we can let the AI do it's thing?
What are your opinions regarding it?
Specifically you won't be able to solve a programming problem if the answer requires you combining over 4 things you don't have in your long-term memory (even if you can look them up). This is the main reason why Jeff Dean is a better programmer than me even though we both have access to google - he has more knowledge & experience of programming in his memory than me that means that even though we can both look things up he is able to solve way more problems than me.
Co-pilot slightly changes the type of thing that's valuable to remember, but it doesn't change the importance of remembering things. I think, as you implied, co-pilot probably makes remembering some types of syntax or boilerplate less important.
https://saveall.ai/blog/learning-is-remembering
"The process of simulating actions is the process that changes the structure of your mind"
* Childhood development. A child who has had early childhood adverse experiences mind is effected. A fMRI or chemical detection can see physical differences in the brain. But the brain is plastic (it can change) over time. Works on this subject have been published by Dr. Dan Siegel https://drdansiegel.com/books among others. The key insight here is that academic learning is not significantly different at a core aspect then behavioral learning.
* Physical sports/martial arts depend on a reaction time much smaller then what is afforded by going through the full frontal cortex. "muscle-memory" isn't real (it isn't "memory" as you think of it). What you have in these cases are "short-circuits" (this implies structural changes) that are able to act before you are consciously aware of what is going on. The same applies to math facts and other fundamentals, you move things away from memory that needs to be retrieved and into reaction. Reading C-syntax for programmers or signing your name is something that has been turned into structure that doesn't need "memory".
I think your initial premise is correct. We have limited memory. How do we overcome that? We write. Writing is important because with it we can overcome our natural memory limitations. You cannot think about complexities (well or clearly) if you cannot write.
The danger of writing is that you can produce something that is both irrational and nearly impenetrable to the casual reader. For example:
> Foucault's use of the concept is descriptive, that is, analytical and explanatory, and at the same time normative and critical: he describes the grip biopolitics have on individuals through technologies of power in a way that makes manifest the repression at work in these biopolitical processes.
The above, taken directly from an "academic" published journal, could be said to have meaning. Unfortunately, each one of these words has an alternate meaning that is not normative to English, making the entire (actual) meaning opaque. "contecpt" "analytical", "explanatory", "normative", "critical", "biopolitics", "individuals", "technologies", "power", "manifest", "repression", "processes" are all defined differently then a standard English dictionary. So even if you can get past the convoluted sentence structure, the intended meaning will still elude you.
But the answer to memory limitations is clear writing using common definitions of words. I bring this up because as you extend your memory beyond what it innately has, the more likely you are to fool yourself (and others) with sophistry.
Very interesting. Is the theory here that by writing things out on a page we are then able to manipulate the ideas in our head without the usual limits of our working memory? Working memory is still limited but because all the information is so nearby and within view we can quickly put things in and out of our working memory so its limit doesn't impede us as much?
0. https://www.sciencedirect.com/science/article/pii/S187705092...!
1. https://www.creativemachineslab.com/uploads/6/9/3/4/69340277...
however long-term storage is just one vital factor, another one is the 'deep learning' neurons that understand the content it stores and more important to connect the dots among various neurons.
we need both: understand and store. Neurons do both for us.
My takeaway from this article is that if you only focus on understanding (and so do not commit it to memory), you cannot reason using this information in unfamiliar contexts later on, once you've forgotten it.
So the best thing to do is to: 1. Make sure you understand something thoroughly 2. Test yourself on it using spaced repetition to ensure you keep it forever
For transparency: I'm one of the cofounders of Save All (linked site) alongside Petros the author
I think what the article misses is that there is a ton of knowledge we can't 'write down' and therefore memorization is not enough. For example learning to solve Integrals. Yes, there are some rules and tricks one can memorize which helps but I would argue the only way to get good at solving integrals is to interact, resp. solve them.
Another point are second order effects of how one learns, for example curiosity and resilience. There might be a long time negative effect on motivation of a topic, when there is too much focus on 'memorization' (It certainly was that for me in my French class;)).
I really enjoyed the 'bad reputation' part and agree that it is sadly viewed as not important enough by many.
Speaking completely personally, forgetting a lot of information is the easiest way to ruin my motivation to learn a topic.
[0] https://link.springer.com/article/10.1007/s10648-022-09677-2
For example, in machine learning, there is something called stochastic gradient descent, where to learn you present a single random element at a time from the dataset. In the end it will have learned of all concepts, by becoming more and more confident in each individual concepts.
For example to learn QM, you pick a random QM wikipedia article, and try to push through the article, even though there are some things you don't understand. Then you do the same thing, for a different unrelated QM article.
For learning tennis, you don't learn specifically forehand, then learning backhand, but you alternate them at random so that you have a single unified way of playing with smooth transitions, instead of having to switch between different "modes" of thinking.
Sure more memory can allow some speed-space trade-off in learning ability, but using your memory too much may make you miss some fluency that may have emerged. For example the old-school of machine learning was using databases and K-near neighbors, which used a lot of memory and was slow. But the new-school of machine learning are using constant memory algorithm and compressing the data in it, and it can learn to generate all the pictures in the world with only 4 Gb of weights.
Learning is imagining, once you bootstrap your imagination, its bandwidth to synthesize new examples from which you can learn from, is much greater than the bandwidth of looking up new data material to learn from.
Anyhow, you can learn to have better "interfaces" to your automatic unconscious abilities and leverage them to e.g. remember instantly and durably any mathematical equation, etc.
- - - -
As an aside, the strategy you sketched out at the start of your article would be workable if you have great recall, however an even more better strategy would be to recapitulate the discoveries of physics in roughly historical order. Your understanding of quantum mechanics would be much deeper and richer, and you would be following a story (an epic story made up of so many fascinating smaller stories, and one that is still going on! Albeit things have calmed down this last century or so, but no one thinks we have reached the climax yet.)
That would be the way to do it: start with the Greeks and the Alchemists and proceed to follow the trail(s) of how we as a species sussed out the mysteries of the physical Universe.
I am utterly baffled that no response in this thread so far has taken issue with the statement "As you probably know intuitively, it won't work." For me, this _does_ work and I have proven it many many times over the years by adding entire categories of technical knowledge to my repertoire. And not superficially, either - I get paid very well to do things professionally that I taught myself by reading Wikipedia.
If my experience were commonplace, the "it won't work" statement would be highly contentious in the comments here. Since it isn't, I guess I can deduce that I must be an outlier.
I wonder, if learning is remembering, then what is "understanding"??
from my own viewpoint, learning is about something external; for example "what's the word for such and such concept?" ..in english or in spanish?
point being that you need a corpus of consensus about what the specific linguistic-culture calls the learned concept.
but then, what does it mean to understand?
I think the way towards making sense of this (answering it) needs to consider learning of physical (do-able) actions. Because when considering such skills as learned/understood, the distinction between learn/understand seems to vanish.
So then maybe understanding has more to do with having learned something to a proficiency level that allows one to teach (show/explain) to another how to do that action?
finally, to throw a proverbial wrench into my own attempts to make sense, what does it mean to perceive something complicated, such as the meaning out of arbitrary alphabetic glyphs? how is the meaning out of a text understood? what did we have to learn to be able to do it? is it just a matter of knowing most of the contents of a dictionary??
If I tell you a sequence of numbers: 1,2,4,8,16,32,64,128
And you try to remember them, having never seen these numbers before, you have to remember each individually and it will not be so easy.
But if you, before trying to remember them, apply a little computing power to figure out that its a sequence of powers of two, starting from 2^0 going up to 2^7, then you have compressed the information I gave you. You understood (presumely) the source of the information and you will be able to remember the numbers much easier.
One strategy I see people applying to unknown data they want to remember is to try to establish links to already known information or made up stories. For example when given the sequence of numbers above but not knowing about exponentials some people would try the follwing:
By doing so some people seem to achieve quit good memory of an unknown topic. But from my point of view they are only re-encoding the information to sort it into already existing bins in their memory instead of compressing it. The amount of information is not reduced but increased and it seems harder to reconstruct the original encoding/information. Additional without compressing the numbers to their generating algorithm it is not possible to use the "learned" knowledge for anything but reciting.This all leads to Solomonoff's theory of inductive inference.
It's not just some simple "move from short term memory -> long term memory" to make more room. If it was, we would all optimize for that outcome. People would have written books about learning that glorified this concept. Teachers would be teaching it in schools to have students score better on tests.
Learning is much more a lifelong mindset akin to the famous Socrates quote of "I know that I know nothing". Or even the idea that we change through the books we read / things we learn, but don't remember much of what we did.
So I don't agree it is all about remembering because like GI Joe said, knowing is half the battle.
https://www.techlearning.com/news/busting-the-myth-of-learni...
> People would have written books about learning that glorified this concept.
There are lots of books on it actually. Make It Stick is my favourite book on this, you should try it!
You have a preference of how you learn, no? It's not a myth if there's perennial truth to it.
> There are lots of books on it actually. Make It Stick is my favourite book on this, you should try it!
I've read many titles including this one. Not everything is going to be "learned" with spaced repetition, interweaving, retrieval, and varied practice. These are great modern methods to learn effectively for the short-term, but are not by any means concepts you're going continue practicing past your formal education. (i.e. Anki flashcards for everything you want to learn)
I might have an emotional preference but there's no good evidence that different people learn better in very different ways
The Inner Game of Tennis frames this as a difference between the two selves. Maybe that could be considered through the lens of decreasing the number of things you need to remember in order to improve.
This! learning materials who master this way of information giving based on their importance is key to effective learning, and I think a lot of them don't give importance to it.
I am fully in agreement with the conclusions drawn in the article, but the bad news is that even with those techniques it can be a slog. Any Anki user will tell you that maintaining dedication and avoiding burn-out is your Achilles' heel, not the limits of your human ability. Understanding something for the first time (in the Feynman technique sense of being able to explain it well), and doing that multiple times per day, takes up a lot of mental energy even if you are smart and naturally talented.
In conjunction with using memory techniques, we need to add dietary practices where we become much more selective with the information we take in. Places like HN give the illusion of learning (and to a great extent help broaden your mind about certain topics), but the actually utility of all that random knowledge butts up against the opportunity cost. There are already many more worthwhile pursuits than can be fit into your lifespan.
So it's all about remembering, but what differs is how someone approaches it depending on their level and understanding of the topic they are trying to learn.
I'm trying to look at the problem with a wider lens, however. For example, if learning that language is actually something we want to do in opposition to all the things you could be doing. In the context of public education, since this is the theme the author focuses on, we don't just study certain topics, we study a certain spin on some topics that is determined by a range of government officials.
In other words, you have to deal with severely limited energy and interest compared to all that is available in life so cutting chaff is probably even more important than boosting your ability to remember. In fact, selecting what to learn is life, just like a sculptor removes the parts of the marble that aren't in the end result.
https://web.archive.org/web/20220926125538/https://saveall.a...
So there might be more information you're expected to know in modern jobs -- but if you spend a bit more time consolidating rather than acquiring new information, you can build the foundations on which more advanced skills can rest.
For transparency: I work with the OP
We likely store information more in some type of loose graph structure, where we recall / "remember" something by re-creating links to that piece of information. There seems to be very very little "storage cost" for the billions of pieces of information we keep in our brains.
I would also argue that the time we spend memorizing "facts", of questionable utility, takes away from the time that could be spent on learning better methodologies for thinking.
It also poses wildly different challenges, where the metrics by which one is to judge the "degree of having attained knowledge" in the broader sense of the word, depends entirely on which of those aspects one value.
Some literature refers to this dichotomy as "instrumental" or "relational" understanding. You see this very clearly in math, where students can recall the facts of equations, but they don't understand it.
It can very well be that the author of this article is aware and appreciates this distinction. But, the phrase and title of "learning is remembering" will certainly evoke suspicion. Just because it is much easier to remember something one understand, does not mean that by remembering, one understands.
For some reason we then give them degrees and call them educated if they remember all those things they don't understand in an exam.
Also I wanted it to stand alone as an essay to see what people thought. If i linked to the product within the article it would turn the whole article into an advert rather than an essay
And obviousness is relative; I did not even notice the logo and base domain, read the title and scanned the text for core concepts, spaced repetition again is only mentioned once, even though in my opinion it’s literally both the topic and your solution to the issue; which was not obvious until I read your comments. Think if you were to ask random people who are not aware of the company, what the post is about — then define spaced repetition and the reasoning behind the company’s name, then ask people again what the post is about — answers would be noticeably different; you could even hide the context and ask if concept of “save all” was mentioned on the page and what it means.
>> Well, new technologies (Save All link) leveraging techniques like spaced repetition mean it's much easier to remember what you learn so its time to rethink that. You don't have to forget what you learn anymore.
Might be wrong, but at the point topic is covered and a single solution is presented, it becomes the topic.
– Jordan B. Peterson
I like to think that Spaced Repetition learning pre-empts these errors, so we are apt to recall when we productively need that memory.
[1] https://m.youtube.com/watch?v=yJ6_cV_RtQU
How do you trust your memory when (as I understand it), every time you recall something your brain changes in ways that can affect recall later?
Is there a distinction between learning and knowing? For example, in my undergrad days I learned about Simpson's rule for numerical integration. I jammed enough of it into my memory to pass the test and quickly forgot all the details. Now, 30+ years later, I needed to calculate the volume of a pretty complex space. I remembered the existence of Simpson's rule but absolutely nothing else about it. Looking it up on Wikipedia I was able to re-learn it well enough to apply it in my job and move on to the next problem.
If, over the past 30 years, I had been using flash cards to remember the details of Simpson's Rule, I would have wasted a lot of time. Re-learning it when needed also means I don't have to rely on my faulty, dynamic memory. For me, it seems like there's a sweet spot to remembering enough to know the concept and then relying on the internet to fill in the details as needed.
These strong connections are why when you go back and look at it, you recognise it and you know how to apply it - because you still have some of the residual memories from this strong encoding. But in the meantime, you probably haven't been able to apply it in an analogy for example.
Secondly - there's a classic on the topic of Spaced Repetition written by Gwern.[0] Gwern calculated that, given the average amount of time you spend testing yourself on something, and the exponential increase in how long you remember it, if you would spend more than 5 minutes per 10 years looking something up, you should use spaced repetition to remember it.
For transparency I work with OP on Save All.
0: https://www.gwern.net/Spaced-repetition
I'd vote for the ability to perform a skilful epistemic analysis of the retrieved information being more useful. I prefer this because it can overcome any natural immutable shortcomings in the underlying process.
Yeah that phenomenon is called "proactive & retroactive interference".
> For me, it seems like there's a sweet spot to remembering enough to know the concept and then relying on the internet to fill in the details as needed.
I think i agree with you to an extent. For example, there is no point memorising all the digits of pie so that you can use it when programming, it's much more efficient to just remember what pie is at a vaguer higher level than to put the energy into memorising all its digits.
But i would say that most people go too far the other way and only remember less than the efficient amount.
If you want to experience this yourself but you already know the time tables, you could try to memorize log tables and then you can do those fancy arithmetic tricks that old school engineers used to do with slide rulers.
https://en.wikipedia.org/wiki/Forgetting_curve
See plot with spaced repetition halfway down.
- https://andymatuschak.org/
- https://www.patreon.com/posts/71081197
- Literally a related quantum computing / physics example (with Michael Nielsen): https://quantum.country/
Why is that "good" teaching? ---> Its because if they arrive at the answer themselves then both the answer & the process for figuring it out will be more ingrained in their long-term memory!
> Instead, learning the content rapidly (meanwhile forgetting a lot of the content), getting a whole picture, and gradually gain more understanding repeating the process using different textbooks/courses, is a much more faster way of learning.
I agree to an extent. It's inefficient to try and remember everything beyond a point. But spaced repetition is very efficient e.g. with Save All it might only take you 5 minutes to remember something for 10 years... so it is much more efficient than you think to try and remember more things.
And alternatively put, there's a difference between learning to read, and learning to speak. Yet both would boil down to primarily rote memorization, as languages tend to do.
Language is not a good example to prove your point on.
- You do not truly understand until you "forget" what you've learned.
I know, it's weird and opposite of what Feynman Said. But it makes sense.
I think that generally is a characteristic of something who's mastered something.
It does make sense, but I don't think it's strictly true. You can understand something without making it second nature.
You have not truely mastered something until you've forgotten the step-by-step process of it (and presumably had to reconstruct it by doing it and paying attention to how you do so).
(This is also why it can be hard to teach something you're very good at - often, you literally don't know (well, remember) "how to do it" in a form that's actually useful to them.)
Learning is mostly model building. Building a model to predict future sensory inputs from past sensory inputs. Building a model to predict future sensory inputs from past sensory inputs and control inputs.
Learning is not just recording sensory inputs for future recall, it is taking them and building useful abstractions with them.
Wikipedia is factually correct but often lacks insight. It puts the learner at the wrong level of abstraction, limiting how much more can be learned.
For example, a sine wave looks complex, and has a great deal of inherent complexity around stuff like transcendental functions. But it's just a spiral, the side view of a radius arm turning through time along the x axis with a period of 2 pi radians and a radius of 1.
But if readers don't know that, they get stuck at the abstraction of trigonometry instead of the far deeper relations between things like complex numbers and higher dimensions.
That's why I think it's difficult to learn quantum mechanics without a teacher. It just ends of being a bunch of matrices and handwaving that makes little sense intuitively.
This is why the debate around higher education is silly IMHO. Sure, someone can avoid college and get hands-on experience in application. But they'll miss out on the theory and abstraction that allows them to transcend their area of expertise. That's good enough for most people, but most likely won't result in true mastery. No schooling is not better than schooling if one wants to do important work.
You wrote "learning limits in isolation of the whole picture, optimizing for less percentage of forgetting, often leaves learners confused". The is true. The reason is because, if you do it in isolation, you're not forming connections to related knowledge.
The main split in brains is short-term memory vs long-term memory. And I suspect (speculating here) that forming more connections to items in long-term memory helps in moving a fact from short-term memory to long-term memory.
When I study something, I go for awhile, but eventually it becomes difficult, confusing, hard to see the forest for the trees. Particularly with technical information and skills like programming (or natural) languages.
When I come back a little bit later, I find that I only remember the things that made sense; my confusions are forgotten, and there is fresh mental space and energy to master a bit more of the terrain before I need another break.
When I'm studying a foreign language, I learn some words and they stick, but I'm exposed to a bunch more that I don't remember next time. I forget those meanings, but the ones that stuck are now vivid and with me, brighter.
When I'm studying Kubernetes, I end up reading a ton of information that's irrelevant to the task at hand, and lots of it doesn't make that much sense because I'm new to it. The next day, when I come back, the things that I actually understood remain, ready to be the foundation for new learning, which they couldn't have been when they were mere data points in an overwhelmed brain. I don't remember the parts I was confused about yesterday, just this stuff that now makes sense.
Learning basic QM is 90% complex differential equations and 10% physical intuition.
There is a reason why you generally don't get to introductory quantum mechanics until second year after (or while) you're doing calculus, differential equations and linear algebra.
To quote a lecture I once saw: the deepest point of any state is a mine shaft somewhere. You don't find that shaft by gradient descent.
It doesn't matter how much of the wiki page you remember, without understanding a few key pieces you're wasting your time.
So now I'm wondering, is my initial "learning" process wrong? Since definitions are fairly simple and I find them easy to remember at first but hard to hold onto. And my process in Anki for New cards is to rep them like reviewed cards until I pass the card; so really, I'm just staring at the kanji + definition until I remember it.
This brings me to my question: is there more Anki can do when it comes to learning new cards? Is there something we can do when learning new information that will help make it "sticky"? Especially in regards to "simple" facts, like basic kanji -> definition mappings, where there are no mechanics to understand, just simple mappings.
Edit (an addition):
Also, I love SRS, but I don't understand how it can be advertised as completely different than rote memorization. When you learn a new physical flashcard, you're learning it the same way as you would be when learning a new Anki card. The only difference is that Anki will show it to you again at a more efficient time in the future, rather than at some regular interval.
They focus intently at seemingly incomprehensible complexity. Then they play with the small fragments they can get their hold on. "ba-ba-ba-ba" when they're learning to speak.
They play. They combine two (or more) different things they're interested in into a single experiment which we think of more as play. Mushy food? I wonder what happens when I throw it on the floor? Oh look how it landed all funny! Look how mum and papa reacted!
Grinding through cards learning languages never really did anything for me. Immersion and regular (but not exhaustive!) play stretched over time makes the neurons of my mind much more reliable.
For others, no. It's joy. And their secondary benefit is learning.
SRS is great for the characters -> definition pathway, but we need so much more. Listening, speaking, interacting, using the language to describe the world around us in conversation and navigate through it.
Don't stop using Anki though.
One way that Save All (a company i run that's like anki, https://saveall.ai/) helps you with encoding is that it suggests alternative ways that we can quiz you on a card you just made. This helps you engage with the card a bit more as you're making it which makes it more sticky.
Other than that a good way to also improve encoding is to link new knowledge to existing knowledge. Things like Roam help you do this with their backlinks
Have you looked at something like wanikani.com with the mnemonics or some of the more historical derivations of the different characters? Those might help build an internal story for why those characters mean what they do. As an adult, that helped me.
I'm recommending you some things that have completely changed my experience using Anki for language learning:
- Don't use translations. Use images instead. For learning how to say dictionary put a picture of a dictionary. It will be easier to remember. Also, you usually won't find a perfect translation for a word.
- Use cloze cards for grammar. Instead of learning the rules, understand them and put four sentences with placeholders and repeat them. This process will make this way of constructing sentences stuck in your head.
That has to be achieved by actually practicing the language, either via listening, speaking, or writing.
Anki is there to help after that step by keeping you refreshed with the exact tones, strokes, multiple pronunciations or more ambiguous CJK characters etc. but it can't be the motor behind the process.
For spaced repetition, although it is a form of rote memorization, the key insight is that memory is formed through recall, not review. The system then (purportedly) hits you at the very moment you are about to forget the card. It's counter-intuitive but it's easier to encode things in long term memory by recalling them at that point as opposed to when it's still fresh in your mind.
This is analogous to studying in university via explaining things to yourself versus just highlighting and rereading the textbook. Every student knows the former is more effective.
Anki might help some, maybe reading, but speaking and listening require actually speaking and actually listening.
I think you are understating the difference this "more efficient time" makes. SRS outclasses every other approach we have for long-term memorization by several zeroes.
Bringing your car to the shop when the "check engine" light is on and bringing it into the shop after every single drive is superficially the same, but only one of them makes actual sense...
For me, a big part of learning a new word involves creating chunks. Those chunks then form bigger chunks. Let's use 辞書 as an example: 辞 is talking/words (and also used as kanji for quitting) and 書 is books. So word book = dictionary.
The kanjis are also broken down by radicals. 辞 is 舌 (tongue) + 辛 (spicy). There isn't a logical connection between the combination of radicals and the kanji, so I'd make a memorable story instead [1]. Thus the only things I have to purely memorize are the radicals.
I go through above process when I'm creating a card and I believe the process itself helps me learn the word much faster than just rote memorization. Of course, you need to start using the word in your daily life too!
Edit:
[1] When I create a story, I close my eyes and try to imagine the story happening as vividly as possible. I also have a rule where radicals in the story must appear in the same order as in the kanji
I've studied a few languages, and by far the easiest, fastest and most long-term proof way to memorize words is by free association.
It takes some initial effort to be able to create these associations, but when done, it's done. There are some words I now basically can't forget, because the association is just too strong.
Example:
Waarschuwing = dutch word for warning. With free association i did it like this.
the "waar" part of the word immediately made me think of english word "war"
schu, shoe
wing, wing
So i vividly imagined a flying shoe with a trompet warning about an upcoming war. Done. Unforgettable for me. The more ridiculous, emotional (funny, stupid, ...) and obvious the better.
Some words are easier than others, and the ones that don't have a quick association jump out quickly are at a higher risk of being forgotten, but even then, they will stay way longer and without much less effort than with brute force, which is basically just creating memories out of nowhere, with no associations to anything on your brain. Our brain loves to relate things to others. Unrelated things = useless things. Everything that is useful has a relationship to something else.
So when you try to memorize by a way that is not based on association, indeed, only brute force works, which is a way to tell your brain "welp i can't really tell you what's the usefulness of knowing this, but somehow it keeps coming back into consciousness, so i guess i need to keep it in here somewhere. A self sustaining island, a dot of knowledge with no connection to anything else i know or care about."
Thats because Anki isn't for memorization, it's for time optimization and scheduling. Nothing about Anki makes you remember a fact better, that's still on you, and the memory champions use ALL those tricks and techniques.
Anki just schedules the non-memory-champions needing to mostly remember most of the things, and minimize the time taken to try and do that.
Said as a multi-year daily Anki user getting really close to a 1000 day streak.
Specifically, someone with ADHD (a topic close to my heart), the size of working memory is typically around 3 items. Related, long term memory is also a bit more chaotic for those with ADHD; when you can only have three linked concepts at a time in working memory, it's going to be encoded with fewer overall links data.
That all is to say, if you're building a tool to help people remember things, don't just build it with the median in mind.
No, learning is not 'storing pieces of information in long-term memory and recalling them'. It's not the ability to recall information. At the very least, learning is information+behavior change+understanding+values and attitudes associated with the information. It's much more complex than memorization and recall.
To add to this, I've known many folks that can accomplish certain tasks almost automatically and creatively. If you asked them to recall exactly what they did to achieve it they couldn't. And this usually isn't action on concrete information either but on intuition alone.
If humans worked primarily on memory we'd have been toast a long time ago. There's too much variation in the natural world to confront it solely on the basis of memory. I'd say we're more experientially oriented as opposed to memory oriented
[1] https://en.m.wikipedia.org/wiki/Meno