YT YouTube2Text

YouTube2Text → Transcripts → How to memorise (almost) anything you read in 20 minutes

How to memorise (almost) anything you read in 20 minutes — Transcript

by Justin Sung · 3,773 words · 548 segments · language en · Watch on YouTube

Open in transcriber Summarize with AI Download SRT/VTT

Full transcript

0:00So, today I'm going to teach you my
0:01favorite method of memorizing. I use
0:04this method to memorize over 150 facts
0:07in my third year of medical school. It
0:09took me maybe 2 hours and I had over 90%
0:12retention for over 5 years. This method
0:15is actually a combination of a few
0:17memorization techniques like the story
0:18method and the method of loci and I've
0:20put it together into a system of steps
0:23that I've personally found to be more
0:26practical and effective than anything
0:27else that I found. So, I'll show you my
0:29exact steps and my personal workflow so
0:32that you can start using this
0:33immediately. There are four steps in
0:35total and the first step is by far the
0:38most important so I'll spend a little
0:40bit longer explaining that. And this
0:42first step is
0:43using a memorization filter. This first
0:47step is crucial. So crucial in fact that
0:50I don't think there is any memorization
0:53technique that is sustainable long-term
0:56especially with high volumes of content
0:58if you do not apply this step. In short,
1:01using a memorization filter means
1:03actively deciding what you should and
1:06should not memorize. And here's why it's
1:08so important. When I was first in my
1:10first few years of university, trying to
1:12enter into medical school, going through
1:14medical school, just endlessly being
1:16bombarded with facts to learn. The
1:18pressure that I felt was there was so
1:21much I need to hold on in my memory. I
1:25have to use the method that allows me to
1:27memorize this as fast as possible. And I
1:29discovered flash cards and flash cards
1:30are very fast because you basically just
1:33read a sentence and then you make a
1:35little flash card on it. These days you
1:36can just like automatically have those
1:37created with AI. And because I was doing
1:40this for every lecture, like every day,
1:43creating hundreds of new flash cards,
1:45this is how I quickly ended up having
1:48thousands of flash cards to constantly
1:50go through. Like my spaced repetition
1:52algorithm was just like every minute
1:54something is due. And this is actually
1:56mathematically unsustainable. And so
1:58here's how I think about it now as a
2:00much more experienced learner and also
2:02as a learning coach. This is the
2:04forgetting curve. You may have seen this
2:05before, the famous Ebbinghaus forgetting
2:07curve. Basically says that when you
2:09learn a new piece of information,
2:11initially you will forget it very
2:12quickly. So this is time on the x-axis
2:15and this is your level of recall, your
2:19your percentage of memory. So you learn
2:21it the first time, you'll forget it very
2:23quickly. So therefore, if you repeat it,
2:25then you forget it a little bit more
2:27slowly. If you repeat it again, you will
2:29forget it even slower. This idea that as
2:32you repeat things multiple times, you
2:34forget it more slowly, this is called
2:35the spacing effect in cognitive science
2:38research. But the issue with relying on
2:41this to hold on to and retain all of
2:44your facts is when you're learning a
2:46very high volume of material, this is
2:49happening like a hundred times every
2:52single day. And for you to have good
2:54memory on this fact, you need to repeat
2:57it like two to maybe five times. And so
3:00every day, let's say you're learning an
3:02extra 100 new facts, that is giving the
3:04future version of you, for example, 500
3:08repetitions. And every day, you're
3:11giving yourself another 500 repetitions
3:14for the future. And so for every one
3:16fact, that is giving the future version
3:18of you five times the amount of work.
3:21It's like this endless hamster wheel.
3:22Some of my students have called it Anki
3:24hell because Anki is you know one of the
3:25most popular flashcard apps. And so if
3:28at all possible, you want to not need to
3:31memorize things. And I think this is one
3:33of the things that I was really confused
3:35about initially is that I didn't
3:37understand that memory and memorization
3:41are not the same thing. Memory is just
3:44having stuff stored in your memory so
3:46that you can recall it. Memorization
3:49is the process of how you achieve
3:51memory. And typically, when we're
3:53talking about memorization, we're
3:54talking about memorizing things through
3:56rote, through repetition. It's very
3:58tedious, it's very boring, it's very
4:00passive. And despite it being called
4:02memorization,
4:04what we know from modern research is
4:06that memorization
4:08is a incredibly inefficient way of
4:11developing memory. Right, so instead of
4:14calling it something like memorization,
4:15which is actually misleading, we should
4:17really just call it rote repetition. And
4:19we actually know this from just
4:21day-to-day experience that most things
4:24that we can preserve in our memory don't
4:27need to be memorized. And so bringing
4:28this back to the memorization filter,
4:31this is the way that we decide what we
4:33should be memorizing and what we should
4:35try to learn through some other
4:37mechanism. We ask ourselves three
4:39questions in this order. First of all,
4:41do we understand why this fact is
4:45important to learn about? If the answer
4:47to this is no, I don't understand, then
4:51try to look it up. If the answer is,
4:54after you've tried, you still do not
4:56understand why it's important, we move
4:58on to question number two.
5:00Is this
5:01similar or related to anything else that
5:04you've learned? Perhaps you can find an
5:06analogy to help you understand this. If
5:09the answer is no, it doesn't seem
5:12similar to anything else, again, try to
5:15look for similarities. And if after
5:17you've looked, you still cannot find
5:20anything that it's similar or related
5:22to, then we move on to question number
5:24three. Do I need high-level
5:28mastery over this concept or fact? So
5:31high-level mastery refers to not only
5:35being able to recall and list or
5:36describe that particular fact, but being
5:39able to use that for complex
5:41problem-solving, especially if that
5:43concept or fact is going to have have
5:45fit with other concepts and facts. If
5:48you know that the piece of information
5:49you're trying to learn works together
5:51with lots of other pieces of
5:52information, then you probably don't
5:54want to be relying too much on rote
5:56repetition because rote repetition
5:58doesn't build up your brain's ability to
6:00apply the knowledge in diverse ways. But
6:03if the answer to this is no, I basically
6:06just need it for
6:08low-level recall, this is something that
6:11yes, you can go ahead and memorize. So,
6:15the idea with using this memorization
6:16filter is that not every single piece of
6:18information will decay this quickly. And
6:22this is also one of the key
6:24breakthroughs that really allowed me to
6:26save time in learning so many facts
6:29every single day. Is the idea that not
6:32all information decays rapidly. How
6:34quickly something decays really depends
6:36on how well integrated this fact is with
6:41everything else you already know. So, if
6:43this is your existing body of knowledge,
6:45we call this a schema. When a new fact
6:47can be related to many other things in
6:50your existing knowledge, this
6:52information naturally is retained
6:56for a very long time. Like I remember
6:57when I was like a 13 to 16 years old, I
7:01played this game called Dynasty Warriors
7:03a lot. And this game franchise is set in
7:07the like old wartime China period. And
7:11there's lots of famous names that appear
7:14like historical figures and generals.
7:16And I played this game so much and I
7:18learned so much about the battles and
7:21who was in charge of which dynasty and
7:23the names of all the generals that
7:25later, like years later, when I actually
7:27learned about Chinese history a little
7:29bit, I was like, "Wow, that's
7:31so interesting. Oh yeah, that's the
7:32battle that I I played in that
7:35particular level and on the left was
7:37that particular base with that annoying
7:38archer that kept shooting at me." So,
7:40that information, which normally might
7:43be very dry and hard to remember, for me
7:45it was very easy to remember because I
7:46had so much of this prior knowledge,
7:50prior experience from just playing the
7:51game. And so, it's the same thing with
7:53learning any new piece of information or
7:55any new fact is that some facts that you
7:57learn will be very close to what you
8:00already know, and then some facts that
8:02you learn are going to be way out here.
8:04They're so disconnected from everything
8:06else that you've learned about, all of
8:07your existing prior knowledge, that
8:09there's no real way to connect it. And
8:11so, this stuff we feel this is random.
8:16We feel that this is irrelevant. And our
8:19brain can tell that it's irrelevant. And
8:21so, it makes a decision, you are a waste
8:24of energy, and it will just throw it
8:25away. And that is what happens when we
8:28forget. So, this type of information
8:30that's really disconnected, this is the
8:32stuff that you forget rapidly. But, if
8:35we find a way to connect it, if we
8:38realize, oh, actually I thought it was
8:41really random and irrelevant, but
8:43actually there is something that I
8:45learned in a different class, or there
8:48is a different experience that I've had
8:50that reminds me of something that's very
8:51similar, we can create a connection that
8:54maybe didn't come to us naturally at
8:57first. And in doing so, instead of it
9:00being forgotten like this, it gets
9:01forgotten like this. And so, even
9:04without doing any of these repetitions,
9:07we're still able to forget the
9:09information much more slowly. And this
9:12is what we want to happen most of the
9:14time. If you've got a hundred new facts
9:16you're learning every day, and the slope
9:17of every new fact looks like this,
9:20you're in for a rough time. And that's
9:22exactly what using this memorization
9:24filter does, is it says, hey, do we see
9:27why it's important? In other words, can
9:29we see if it connects to anything in our
9:32existing knowledge straight away? If
9:34not, try. Try to find a way to connect
9:37it. If you still can't, can you at least
9:40create an analogy from something
9:43distant in a different topic? Some kind
9:46of connection so it's not just stranded
9:48there in the distance, isolated, waiting
9:50for your brain to just prune it away. If
9:52the answer is still no, but we know
9:54we're going to need to use this at a
9:56higher level eventually, then maybe one
9:59day we can build it. We can aim for
10:00that. Right now is very far away from
10:03everything we know, but let's try to
10:06scaffold our way to be able to connect
10:08it instead of just accepting that it's
10:10going to be completely random and
10:11disconnected. So, I started doing this
10:14starting from my fourth year of
10:16university onwards. The amount of flash
10:17cards I needed to make on a daily basis
10:19reduced by 80% by keeping rote
10:24repetition as my last resort, my memory
10:27actually started improving dramatically.
10:29However, there's always going to be a
10:31bunch of stuff that falls into that last
10:33resort, especially in something like
10:34medicine. There are a bunch of these
10:36things that eventually pass this filter
10:38and yes, we just need to memorize it
10:40through rote repetition. And so, when
10:42that happens, we now move on to step
10:45number two. And so, step number two is
10:47to prioritize. What we're going to do in
10:49this step is decide what specific facts
10:54we're going to try to memorize. So, I'm
10:55going to walk you through an actual
10:57example so you can see the process that
10:59I used step by step. So, this is a table
11:02of different antibiotics, side effects,
11:05the class of antibiotics, how they work,
11:08when you should use them, the type of
11:09resistance. So, there's a lot of dense
11:11information here. And so, what we do
11:13with prioritizing is instead of just
11:15taking the facts and just learning
11:17everything. So, for example, if we look
11:19at the top line, which is called
11:20sulfonamides, which is a group of
11:22antibiotics, if we wanted to learn about
11:24the side effects, the way you do this is
11:26look at each fact and try to apply the
11:30memorization filter to it. What you
11:32should find is that if you look through
11:34uh this list, nausea, vomiting,
11:37hypersensitivity, these are actually
11:39repeated in like most of these
11:42antibiotics. And so normally I would
11:44have to commit that to memory, but now I
11:46don't because it's much easier to
11:47remember pretty much all the antibiotics
11:50will make you feel a little nauseous.
11:52And actually the more outstanding fact
11:54on this is this bone marrow depression.
11:56This is not very common. I don't really
11:58see it anywhere else on this list. And
12:00so I would prioritize
12:02that particular fact. And so during this
12:04prioritized step, what you do is you
12:06write down what are the actual
12:08individual things that you want to make
12:10sure that you remember. So for this one,
12:11I wanted to remember that sulfonamides
12:14are static, that they cause bone marrow
12:18depression, that allergic reactions are
12:21common. Everything else is either simple
12:23enough for me to remember or I can just
12:25figure it out through logic. And then
12:26you move on to step three. Step three is
12:29to obs- certify. And what I like to do
12:31is I create a character out of the list
12:36of facts. This is very similar to the
12:38memory palace or method of loci. And for
12:40me personally,
12:42I like to use AI these days to do this
12:45image generation. I used to do this by
12:49like literally drawing
12:51every single thing, but it You can still
12:53do that, but it just takes much, much
12:55longer. And so for me, my personal
12:56workflow is I actually use Hicksfield,
12:58who is also the sponsor of this video.
13:00Now,
13:01I will say that when it comes to using
13:04AI for learning, generally I found that
13:08it's pretty overrated. But this is a
13:11genuinely time-efficient and worthwhile
13:13way of using AI for learning. But as I
13:15said, if you don't want to use AI, you
13:17can still do this just drawing it out.
13:19It's just going to take you a little bit
13:20longer. So what I did is I took my list
13:22of prioritized facts and elements, and I
13:26turned that into a character. So, when I
13:28thought sulfonamide, sulfa, that reminds
13:32me of hot pools. When I thought static,
13:35in this case on this table when it says
13:36the word static, it's referring to the
13:38fact that it doesn't kill the bacteria,
13:40it just like stops the bacteria from
13:42growing so that your own body can kill
13:44it by itself. Static, I started thinking
13:46like like radio static, like glitching
13:49in and out. And so now in my mind I've
13:50got this this geothermal rock monster
13:54glitching in and out. For the bone
13:56marrow depression, I thought of like
13:57this big broken femur bone. And because
14:00it's not very useful with infections
14:02with pus, I wanted this
14:05rock monster to be sitting in a big pool
14:08of pus. One of the other things that we
14:10wanted to remember is that you don't use
14:12it very often because there's lots of
14:14resistance. So in my mind I kind of
14:16pictured that this monster is shackled
14:18up. And to emphasize the fact that
14:20there's lots of allergy to this, I
14:22wanted this rock monster to have lots of
14:25red uh rashes along its body. And so I
14:28put that into the prompt, uh and I got
14:30this image, which as you can see uh
14:32seems pretty memorable. You can look at
14:34this and those elements that I described
14:36to you, all of these things
14:38are cues for you to remember the facts
14:40that you need to remember. And so I did
14:42this for a few of the other items on
14:44this uh table, trimethoprim,
14:47uh penicillin, cephalosporins,
14:49carbapenems. You can see these are the
14:52uh prompts that I gave for each one and
14:54the image that I was able to generate
14:56for all of them. You'll notice that uh
14:58some of these are cows or like cow-human
15:00hybrids. I had a lot of fun with this
15:02one because
15:03uh penicillin, cephalosporins, and
15:05carbapenems are all a class of
15:07antibiotic, which are called
15:09beta-lactams. And lactam kind of reminds
15:11you of lactose, you know, like milk. And
15:14I thought, well, what if all the lactams
15:17were cows? And penicillin is like the
15:21most basic of the beta-lactams uh in the
15:24antibiotics. So, I thought, "Well, what
15:26if penicillin is just a straight-up
15:27cow?" It's not even like a cow-human
15:28hybrid, it's just a cow. You'll also
15:30note that a lot of them are wearing
15:32hats. This represents broad-spectrum
15:34coverage. It means that the antibiotic
15:36is useful for a lot of different types
15:38of bacteria. So, I would give these
15:39characters big straw hats,
15:42broad-spectrum.
15:43I don't know if that makes sense to you.
15:45This is This is step three where you are
15:47absurdifying these clusters of facts and
15:50to these discrete characters. And the
15:51reason I like using Hexfield for this is
15:53because I can type in the prompt and it
15:55will generate it. It takes a little bit
15:56of time to generate, but then while it's
15:58generating that, I go back and do step
16:00two, prioritizing for the next one. So,
16:02I'll do sulfonamides, figure out the
16:04elements, generate the thing. As it's
16:06generating, I move on to the next
16:07antibiotic. Now, you can actually stop
16:09here, but what happens if you keep doing
16:11this and you end up with lots of
16:12different characters, so if you've got a
16:14lot of stuff to memorize, is that you
16:16forget what character represents what
16:19group of facts or concepts. It's kind of
16:21like when you remember that there is a
16:23mnemonic for something, but you can't
16:24remember what the mnemonic is meant for.
16:26And especially when groups of facts
16:29relate to each other. So, for example,
16:32sulfonamides are often given in
16:34combination with trimethoprim, which is
16:36that second antibiotic. So, that's
16:38really hard to represent in a single
16:40character for each. And this is where
16:42step four comes in, which is to storyfy,
16:45which I'm pretty sure it's not a word,
16:47but you basically put them together into
16:48a story. This is taking from the link
16:50method or the story method technique.
16:52And so, what you do is you download
16:53these images that you've generated, you
16:55go into video, and you make a cinematic
16:57video where you upload all of these
16:59images as characters. And then think
17:02about a story where these characters are
17:05relating together. What I recommend is
17:07if you're going to use Hexfield is
17:08changing some of these settings. For
17:10this one, I picked epic as the genre.
17:12You don't have to pick epic, but I just
17:15thought it worked with the theme of what
17:16I was trying to visualize as a story,
17:19but I definitely recommend changing the
17:20camera mode to single shot. This allows
17:24the characters to kind of all feel like
17:26they're in one cohesive story as opposed
17:29to there being like random cuts in the
17:31video that's generated and getting a
17:33little bit disjointed in terms of how
17:34the characters are interacting with each
17:35other. And so you create the prompt and
17:37you generate your story and this is the
17:39video that ended up creating for me. So
17:42we start with my sulfonamide monster and
17:45then next to him you can see a sitting
17:46down trimethoprim. I made trimethoprim a
17:49triceratops and he pulls out his letters
17:53that represents another thing. And so
17:54I've got that they are sitting together.
17:56They are often given together. You have
17:58this guy who's carbapenem. Carbapenem
18:00was like a serious antibiotic. Like you
18:02don't often give it for little
18:03infections, so this is why he's really
18:05kitted out. You have this a family of
18:07cows which stands for the three
18:09generations of cephalosporins and then
18:12suddenly here appears this cow this this
18:16the penicillin cow charging through. And
18:18the reason I wanted this to happen is
18:19because penicillin is a very broad
18:21spectrum antibiotic, often one of the
18:23first line treatments. So before you
18:25even give any other antibiotic, when
18:27you're not even sure what to give,
18:29sometimes you just give penicillin
18:30first. So with the story you're trying
18:32to create interactions between the
18:33characters and by doing that each
18:35character becomes a memory cue for the
18:38other characters and that helps with
18:40your retention. And if possible, you
18:42want those interactions to be
18:43meaningful. Stuff that you can't
18:45normally represent just within the
18:47character itself, you try to represent
18:48that into the story. And so this video,
18:51although it doesn't look like it,
18:52actually includes 27 facts across five
18:56classes of antibiotics. It is really
18:58easy to remember. I could definitely
19:00double or triple the the amount of facts
19:01that I fit into this story. You probably
19:03remember the story as well. Even if you
19:05don't know what each piece means, you
19:06can still visualize it. I did this a few
19:08days ago and I then tested my retention
19:12on all 27 facts
19:14at 6 hours and at 48 hours and I got
19:17100% for both tests. So, if you use this
19:19method, you can expect that your
19:20retention is going to be really high.
19:22You will still need to review it, just
19:25making sure that you remember all the
19:26elements and what they stand for, but
19:28instead of repeating it like every week,
19:30you can get away with just refreshing it
19:31maybe every 3 to 5 weeks. But remember,
19:34please, that even though this is a very
19:36powerful technique, which is now much
19:38faster than it used to be because you
19:39can just use AI to generate the images
19:41and videos, this is still a last resort
19:44technique. You don't want to be
19:45memorizing your entire curriculum using
19:47this method because you will not be able
19:48to use and apply your knowledge at a
19:50higher level. Now, if you want to check
19:51out Higgsfield for yourself and play
19:53around with this method, then I'll leave
19:54a link for you in the description below.
19:56And if you want to upskill on the parts
19:57outside of memorization, all the other
20:00things that have to come together to
20:01create an entire learning system, then
20:03I'd recommend checking out this video
20:05here where I break down how I, as a
20:07learning coach, design learning systems.
20:09But otherwise, thank you so much for
20:10watching and I'll see you in the next
20:12one.

About this transcript

This page contains the full transcript of How to memorise (almost) anything you read in 20 minutes by Justin Sung, generated from the public captions YouTube serves with the video. The transcript has 3,773 words across 548 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to translate the transcript into 100+ languages.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, translation to 100+ languages, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.