How I use memory systems

Michael Nielsen


Initial release: July 20, 2023. I will add additional practices over time. Mostly a sequel or advanced version of some of my other writing about memory systems.

Did you know you can use memory systems to: understand a research paper; help you write a book; expand your visual imagination; support your search for a mathematical proof; expand your empathy; and many more things?

I've used memory systems to help with all these things, and many more. Many people most often think of memory systems as a tool to help learn basic vocabulary in a foreign language. But I've found them useful much more broadly, as a tool to deepen my conceptual understanding of unfamiliar fields, to help in my creative research work, in artistic practice, and in my personal life.

This document describes some of these practices, to help other people get more benefit from such systems. Everything here is something actually do, not merely something I speculate may be useful1. I've shown questions and answers I've used in my memory system – mostly actual Anki cards I use2, not made up examples. I'd love to hear life-tested strategies from other people who use memory systems. A good place to share those strategies is in replies to the Twitter thread companion to this document. I'd also love to see other serious users of memory systems create documents like this. For me, it's repaid the effort many times over: writing about the practices helps me improve them. Indeed, the guilty secret of this document is that a major motivation for writing it is to make me more creative with how I use memory systems. Note that many of the practices I use are described in my other writing on memory systems, and I've put a low priority on repeating ideas from that writing here.

Anki is the main memory system I use. But many of the practices described here can be adapted easily to other memory systems. Still, I'll often use "Anki" as a standin for the more general terminology. I use no Anki plugins. I have made some minor Anki customizations, but I'm afraid I don't remember what they were, and doubt they much matter, since I wasn't making principled changes so much as playing around. I often meet people who tinker endlessly with Anki, but expend relatively little effort in getting better at writing cards. Alan Kay said it well: "the music isn't in the piano". You get better at piano by playing and studying to improve how you play, not by constantly retuning your piano. And so it is with memory systems: you are best served by improving how you write cards, not tinkering with the system. Indeed, I believe most large improvements in the design of such systems will, in any case, come from serious users.

Memory systems make memory a choice. They're also a choice about who you will become, a way of sculpting your future self. The practices I report here have been developed with an eye toward the future self I want. This has many implications, but one of the most important is: I try to only add cards where I care about the content. Of course, sometimes it's sometimes difficult to tell: I'm making an informed bet that in the future I will be very glad that understanding is a part of me. Part of the art of memory systems is to gradually improve your sense of when that's true. For this reason I suggest browsing through the practices and concentrating only on those that resonate for you.

Evoking images of unusual emotional experiences

Nightwish is a remarkable Finnish band who make "symphonic metal" music. According to Wikipedia, they're the most successful Finnish band in the world. Their lead songwriter, Tuomas Holopainen, often integrates scientific ideas deeply into Nightwish's music and shows. In a show at Wembley Stadium in 2015, they concluded their set with "The Greatest Show on Earth", an extraordinary 21-minute song which includes an extended spoken monologue from the biologist Richard Dawkins, about how remarkable each individual human being is.

As Dawkins' monologue plays, the camera briefly shows a woman in the front row. She is reciting Dawkins' monologue along with him. She appears deeply moved and deeply engaged by the words she's reciting, as well as very happy and relaxed, almost in a state of rapture. It's just a few seconds long, but I find it extraordinarily beautiful: she appears to me to be having simultaneous near-transcendent intellectual and emotional experiences, the numinous through apprehension of the natural world:

When I first saw this performance, I shared it with several friends, and talked it over with them, including this remarkable few seconds. I wanted to capture something of that experience in my memory system. So I added more than a dozen cards. Here's a few of them3:

Of course, I don't actually know what she was experiencing. If I was writing for others I might write "What's an image that I believe shows…". But in Anki I prefer unqualified statements, as they express the core idea more clearly. The necessary qualifiers are usually evident later. This is, of course, common in creative work, where one wants to rapidly make webs of approximate associations, trying to find what is essential, before verifying arguments, adding caveats, and so on.

And so on, more than a dozen interlinked cards.

I find the entire Nightwish video extraordinary, and have watched the video and listened to the song many times. I therefore wonder: was there any need for me to use a memory system at all? On the other hand, I believe creating the cards helped me make more meaning out of what I was seeing than I otherwise would have. I began with a vague sense that "this is really something"; making the cards helped me more deeply apprehend what that something was4. Indeed, that meaning has, in turn, become a part of why I've watched the video many times. It's Anki as a way of seeing, or Anki as a way of meaning making. It makes the memory conceptually much richer, even though (of course) watching the video – or even being present – is richer and more emotionally compelling in other ways.

I've gone through a similar process for many other videos and images that depict emotional experiences novel to me. Indeed, there is even a later moment in the same video. Richard Dawkins – the actual living person(!) – has come out on stage, and is reading the concluding paragraphs from Charles Darwin's "On the Origin of Species". Now, conventional wisdom suggests that you don't finish a concert by having an elderly scientist read aloud from a scientific monograph. But that conventional wisdom is utterly wrong. The looks on the faces of audience members are extraordinary, another experience of awe:

Again, I added a dozen or so Anki cards, some in reference to that image, some in reference to Dawkins' experience, some in general about the role scientific understanding plays in our lives.

What does this kind of practice achieve? The broader pattern is this: to be more alert to instances when I'm witnessing a very striking emotional experience. And to attempt to capture something of that experience. Of course, I don't capture the full richness of the emotion – far from it! It's rather to build my awareness and to expand the range of my appreciation, an increase in my emotional imagination and empathy. While I cannot recreate the richness of the emotion, I can at least do a little to evoke it.

A variation of this pattern is based on a still from the teaser trailer for the movie Interstellar:

The cards here are a little different. It's not so much that I am witnessing an unusual emotional experience. Rather, it's that I am personally having (for me) an unusual emotional experience, stimulated by the image. And so I write cards which are in part about that experience, and in part about the visual design:

We are, of course, very sensitive to both our own and others' emotional experiences. This makes me wonder if it's a mistake to use memory systems here, rather than relying on our natural interest. I already briefly discussed this above, but let me return to think about it in more depth.

In 1998, walking down the street in Los Angeles, I passed a photography shop. Up in the window was one of the photographer's photos. It had, apparently, won Third Prize in some prestigious competition or another. It showed a young bride, in her wedding dress, on her wedding day. Her back was turned to us, her face not visible, because she was being held, very tight, by her elderly father. He was an Asian man, with what appeared to ordinarily be a stern mien. But you could see: that sternness had just cracked, his face was broken, and he was weeping, giving his daughter away. It was an incredible image, and I stopped and studied it for perhaps a few minutes. I don't believe I ever saw it again, but it left a deep imprint on me.

And so, of course, for many other emotional images. This is true of strangers' experiences, of fiction and non-fiction, and, of course, of our own experiences. I shall never forget the look on my face in the mirror shortly after realizing my wife and I would divorce. I will never forget the sound in the voice of my Godfather when he called and told me his son had died in a fall from a horse. These and many other experiences are seared into my mind.

What then is the benefit of engaging in this kind of memory practice? I believe it's this: it's for the illegible but instinctively important emotional experience. When I saw that image of the father of the bride, I had no trouble parsing it. It was, for me, a highly legible experience. But when I first saw images like those from Nightwish, or from Interstellar, while I was instinctively moved, I also didn't really understand them. Making Anki cards was part of meaning making for me: helping deepen my understanding of the experience. Of course, other things helped with the meaning making – especially talking the experience over with friends. And so Anki was merely part of that overall experience. Still, it is perhaps especially valuable in making an experience available at a later date. And, it also expands the range of emotional experience I am aware of, since it gives that awareness an outlet. All these things I find immensely valuable.

Image close analysis: adding images to expand my visual imagination

I use Anki occasionally to deepen my understanding of images and to expand my visual imagination. They tend to be images that I find striking, but don't understand why. Consider, for example, this famous image of Frodo in The Lord of the Rings movie:

Like many people, I found the image striking, but didn't really understand why. I decided to interrogate it in Anki:

This is an observation I'm proud of – is Frodo doing things to the ring, or is the ring doing things to Frodo? But I'm not very satisfied with the phrasing! The thought could be improved, and that would make the card better. In general an interesting thing about Anki is that it atomizes thought, and provided a very sharp focus for improving it.

Why add cards like this to Anki? To some extent, I don't do this kind of analysis for any instrumental reason. I just enjoy doing it; I'd much rather do this than crossword puzzles! I enjoy understanding visual culture.

A related question is: why do it using Anki? After all, I could potentially enjoy doing the analysis without wanting to commit the resulting ideas to long-term memory. Here, my reasons are also instinctive. Growing up, I was often treated as "a kid who is into science and technical stuff". As an adult it was an epiphany to realize: I get a similar cognitive and emotional and meaningful experience from enjoying art. But I have also found that I get more enjoyment as my understanding of visual culture deepens. In adding these analyses to Anki, my sense is that I'm deepening my understanding of our visual culture.

An example: several decades ago I realized that artists often think in terms of separate layers. It took a while and multiple examples to understand this. Everything from seeing a Disney animator explaining the multiple layers of plates they used in old hand-drawn cartoons, to a dissection of the way Michael Bay uses multiple layers to create "Bayhem". Understanding this then helped me better see what was happening in the image of Frodo. So I like to accumulate an understanding of visual culture; Anki has accelerated that. Of course, there are many images where I do this analysis very naturally, and the memory system isn't needed. But there's a borderline class where I find an image striking, but for reasons I don't entirely understand. In those cases, I'll often add 5 to 10 Anki questions, just exploring and (I hope) developing my understanding of visual culture.

Add 3-10 cards after listening to an audio book on the exercise bike, or while running errands

One pattern I find enjoyable is to eat breakfast while cycling lightly on the exercise bike in the morning. If I am in the mood, I simultaneously listen to an audio book. Sometimes, something said in the audio book will inspire Anki cards. Most days I don't add any questions, but sometimes I'll add a handful – typically 3-5 or so. I often follow this by doing my Anki practice for the day, while continuing light exercise. A common variation is to do this while running errands, walking to the coffee shop in the morning, and so on.

Anki makes things more meaningful; Anki as a universal (but incomplete) context for caring

Agnes Callard has made the striking observation that "twitter is a way to care about everything". It certainly resonates with my own experience. The philosopher René Girard believed that human desire is often mimetic: we desire what other people want. Twitter makes it easy to act on this, since it enables you to find people talking with great interest about an astonishingly wide range of topics. And, if you engage in that conversation, you find yourself getting more invested in that topic. It becomes more interesting: you start to care about it more.

Indeed, you can even initiate this pattern. Suppose you've been noodling on some esoteric subject on your own. To pick an example from my own life: over the past few days I've been learning about DNA nanotechnology. I'm doing this on my own, and it's a somewhat lonely experience. But: I can tweet about it. Write a hopefully interesting thread, containing questions and ideas that I'm genuinely curious about. And: others reply. Pretty soon, a mini-discussion group is going, almost like running a small conference. Of course, this is helped if you're in the sweet spot for followers: between 3k to 50k seems to be especially good, in my experience. Below about 3k or so, and you will struggle to get replies (depending, of course, on how provocative your thread is). Above 50k and the signal to noise ratio will start to decline: the problem then becomes too many low quality replies. But even with these limitations, twitter is a great way to convene a party or conference, on any subject I care to discuss. That, in turn, makes it easier for me to care, and to immerse myself more deeply.

Anki is, in a very different way, a context that will help you care about anything.

As an example, before Anki I often found it strangely pointless to read research papers in unfamiliar fields. I would read a paper, often with much effort, and then 6 months or a year later would be reminded of the paper, and realize (with a sinking feeling) that I remembered almost nothing of it, despite the effort. Indeed, anticipation of this forgetting makes it hard to read even in real time. It just seems a little meaningless. Now, of course, you can make the reading more meaningful in multiple ways. You can give a class on the subject. Or write an article. Or explain it to an interested friend. But those are high-cost things. Maybe none of your friends are especially interested. And giving a class is very high-cost indeed: that's both the benefit and the problem with it.

I thought I was a crazy for feeling this way, that it must be some unique shortcoming. But it turns out to be a commonplace experience. I first realized this emphatically when reading the mathematician Littlewood, who observed5 "I have tried to learn mathematics outside my fields of interest; after any interval I had to begin all over again." This was gratifying to read: I was not alone in my experience! Of course, conversation with others has since confirmed that this experience is common. Littlewood, correctly, identifies some sort of drive – what I would call caring, or finding meaning — as the key element to such learning, and comments: "Given the strong drive, it communicates itself in some form to the subconscious, which does all the real work, and would seem to always be on duty. Lacking the drive, one sticks."

One marvellous thing about Anki: it makes it low cost to care in this way. In part it's because of that sense of Anki making memory a choice. I switch from finding it all a little pointless because reading the paper will leave so little trace, to understanding that in a year or two or three I will still remember much from the paper. That long-term change in my self, that sense of growth, is meaningful. Anki is a meaningful context. Indeed, we might even go further: Anki is a meaning-making context. It is a near universal context for caring.

Now, of course, it's a fairly low-grade meaning. In that sense it's a quite incomplete context for caring! Certainly, it is not a burning drive, on its own, the kind that comes from a deep sense of connection and meaning! But it's a beginning, a way of bootstrapping into a deeper connection to the material. If you desire a deeper connection you'll need to find other ways of obtaining it: talking about it with friends, writing about it on Twitter or elsewhere, using it as fodder for creative projects. But having a way of getting over that initial hump at the beginning is powerful.

A common mistake: adding cards for things you feel you "should" be interested in

(This section is adapted from: )

People often tell me things along the following lines: "I don't like memory systems. I memorized all this useless information, and it was useless."

I exaggerate, slightly. But I get told essentially this surprisingly often.

A variation is: "Wow, I love your essays on memory systems. But I don't follow any of the advice! I use memory systems to memorize long lists of dull partially-digested facts that have no personal meaning!"

It's a bit like using tweezers as a hammer… to hammer the roses into your garden.

Just don't. It's a very bad mistake to add cards for things you're not genuinely interested in. People will tell me: "I used Anki to memorize the state capitals". Me: "Why did you want to know the state capitals." Them: [silence].

More broadly: someone will think: "Oh, I should be more interested in American history", and start adding lots and lots of cards about American history. But they're not actually interested in American history, they merely think they should be, and so they're adding hundreds of scattershot cards; 3 months later those cards will simply be a burden, of no interest, but a drag.

This is not at all a knock on American history. If someone is genuinely motivated (and doesn't merely feel they should be motivated) to understand more about American history, then Anki will work well. But the word "should" is, at the very least, a warning sign. Maybe you should go and do something else.

Memory systems shine in areas where I'm interested enough that I'm sure I'll find meaning in it in the future, but where I'm not so knowledgeable already that I can just learn en passant. This is always a judgement call, and I sometimes make errors of judgement; I have to cull constantly and cull aggressively. But it's a useful general principle.

Pay attention when something seems illegible: strange, but somehow interesting in a hard-to-parse way

One of the most common critiques of memory systems goes: "Oh, you should just do your work. You'll naturally memorize the stuff you really need". It's sorta fine, as far as it goes. But my spidey sense activates when I see something which I don't understand, but which somehow seems interesting. I gave the example of the Nightwish concert above. I didn't know why I found it interesting, exactly, but I did. Making the Anki cards helped me understand why. Again, as I said above: Anki is especially helpful in areas which you are very interested in, but don't yet have the expertise to enable rapid encoding of new memories.

Creating Anki cards as a powerful context in which to think

Alan Kay famously said: "A change of perspective is worth 80 IQ points".

The same is true of contexts: a powerful new context to think in is often worth 80 IQ points. Teach a class; talk to a beginning student; write a Twitter thread. All are powerful contexts. Every new person to talk with is a powerful new context. Every (good) new book or good new paper is a powerful new context.

There is something marvellous about finding reliable, on-demand, powerfully motivating contexts in which to think.

For me, Anki is such a context.

And the more powerful patterns of card creation I have, the better a context in which to think. It changes the way you interact with your experience. And it becomes easier (or possible) to have certain thoughts; you will have thoughts in this context that would be near impossible in other contexts.

Edwin Schlossberg: "The skill of writing is creating a context in which people can think". Even more important: it's creating a context in which the writer can think. Done skillfully, it makes it much easier to have certain thoughts. Creating Anki cards is an example of such writing.

Stubs for possible future development


Thanks to the many people with whom I've discussed memory systems over the years, but especial thanks to Andy Matuschak, who has influenced nearly every aspect of my thinking about memory systems.


  1. I find fascinating the productivity writers who pour out "content" – I believe I am justified in using that perjorative – about all the different actions and systems one should purportedly use. They'd need a thousand hours in a day to carry out all their own advice.↩︎

  2. With two caveats: (1) I'm quoting from memory, and no doubt the quoting will be inexact; (2) I've occasionally used synthetic cards for reasons of privacy. But I've tried to keep both effects to a minimum.↩︎

  3. I initially had some privacy qualms in adding this, since it's a personal shot of a person. However, upon reflection, I assume that anyone in such a video has given their consent for their image to be recorded and shared widely. In the event that I receive any indication otherwise, though, I'll certainly take this down, with sincere apologies.↩︎

  4. So too did talking with friends about it, and it's hard to disentangle that from the effect of the memory system. "Talk with friends" is one of our best ambient memory systems. However, I believe that making Anki cards helped considerably in making meaning. Indeed, often while making such cards I'll have a thought which I wish to share with a friend, and will message them about it.↩︎

  5. John E. Littlewood, "Littlewood's Miscellany", Cambridge University Press (1986).↩︎