MAJOR DRAWBACK in applying memory palace to learning

For the past 12 months, I have been experimenting with learning using memory palaces where I literally completely rely on memory palaces in learning everything. I have made over 400 memory palaces with a total of about 20000 loci. No doubt I was able to significantly increase the amount of things I can recall. However, I also come across a MAJOR problem in applying memory palaces to learning.

I find retaining more information via memory palaces DOES NOT TRANSLATE to better applying those knowledge acquired in problem solving. Let me try to explain what I mean.

Memory palaces allow a person to place things he/she want to remember in loci in memory palaces so that he/she can easily recall. However, those information acquired are essentially “locked” inside the memory palace. What I mean is that to access those information, he/she will need to revisit the memory palace. By recalling those information over and over again, one might get quicker in retrieving the information, but the process still require revisiting the memory palace (at least subconsciously).

Now the problem is: Say there is a question that can only be solved by retrieving information stored in 5 different memory palaces. Since information stored in each memory palace are essentially self-contained, the brain cannot simultaneously access 5 different memory palaces to solve the problem.

In contrast, learning via the conventional method is like storing information in one BIG void. It is harder to find the information since one need to search through a much larger part of the brain, but at the same time the brain is able to simultaneously access those information required to solve the question.

I don’t know if I have explained myself well enough. But that’s the DRAWBACK I find. I would be interested in knowing what you guys think.



I face the same problem. I have like 10000 Locus. Loci is not a problem, I can create thousands. I use them all, but to retain that information I have to revisit them. That’s tome consuming and also not fluent enough. Rote memory take much more time but give long lasting recall without hesitation and most of the time, NO REVISION. but by no means I am saying that rote memory is good.

That’s why I use Loci for pump and dump type memorization… I memorise for exams and then just forget it. No permanent storage for a lifetime. I start before the exam(like 3 months before, I am lazy, and also I know I can do it) and then forget it after exams.

BUT, I think there is a solution, though it takes a little time. I have noticed that when I am recalling the information from Locus again and again in my mind or saying out loud,each time it’s becoming faster and ultimately I need a little or no mental imagery at all. Like after 10 to 20 recall.

I am not sure if this is a solution at all. Because I don’t have that much time to revise it 20 times and also I have no intention of doing that.

Pump and dump.

Anybody have any solution then please post here. I want to know about it too.

Thank you for posting the truth that nobody talks about. Dark side of memorization. It stored INSIDE LOCI. not fluent. Takes time. Take revision again and again.


Yes, I find to use information stored in multiple memory palaces to problem solve, I need to revisit each of these memory palaces at least 20 times (as you have said). Isn’t that rote learning in a way? Sigh~~

So I doubt the fluency obtained via this way is from revisiting the memory palaces, BUT rather re-exposed to the information 20 times (ROTE LEARNING) as one use the memory palaces.

That is, I find using memory palaces is only good for exams. But even one has stored massive amount of information using memory palaces in a field, he/she will not be able to become EXPERTS of the field (at least using memory palaces does NOT make one become experts in their field faster) as he/she will not be able to use the information acquired to problem solve.

Nah, problem solving is a completely different skill. You know what the irony is? All my life my teachers told me that you cannot memorise something properly without understanding them. And now, when I know these techniques, it seemed my teacher was wrong all that time. Now if I don’t know a subject, even if I don’t know a single word or single terminology, I still can memorise it completely.

So, my point is, memorization is memorization, which requires not much understanding. A Mnemonist can memorise almost anything, if he doesn’t have much brain power. But problem solving requires memorization + understanding + brain. By doing only one thing (that’s memorization), I don’t think anybody could solve a problem properly.

Yeah I kind of think it’s a type of rote memorization. But as you know, our teachers told us to write things down after memorising. What we did that time? We repeat it, repeat it, repeat it, recall it, again again and again, then we write it, revise it, write it again. But with Loci, we are memorising once then revisiting it again again again. But not making much mistake. So I think it’s really good.

First, we are giving only a third of our effort and getting 100% result.
Second, we need about a third of our time in memorising and revisiting.

It still like Rot memorization but you can’t deny that it takes one third of your usual time. Yeah, I don’t think we can solve this problem.

But, the question is, is it truly a real problem?

Sorry for my bad English. I am from India.



@m1313s - I’m curious, what subjects are you trying to learn with memory palaces that also require problem solving skills?

Hi Everett_C, i am doing medicine. But it seems like memory palaces isnt the best method to study for it. Although i understand my materials, i find using memory palaces are significantly slower in problem solve.

One other thing is, i find using memory palaces slows down the rate at which information is converted to natural memory. Some information even seems to stay in the palaces for so long that it does not seem to get assimilated.

That is, by using memory palaces i can memorize lists using shorter periods of time, but this comes at a significant price. Students study via the conventional method may do worse initially, but once they have assimilate the knowledge, they perform significantly better. This is because information in memory palaces are kind of like stored in “locked houses” and some of this information never becomes natural memory as they are being locked away too well.

So in summary, i find studying via memory palaces do BETTER IN THE SHORT RUN, but DO WORSE THAN OTHERS IN THE LOMG RUN.

1 Like

One more thing i would like to point out is that the statement “revisiting your memory palaces over and over will allow one to recall without going through the palace” is ONLY true to a certain extent.

I find it definitely speeds up recall as one does not need to go through each loci one by one to access the information. But no matter how much I practice, i find subconsciously a brief mental image of the palace will have to appear in my brain first before i can recall. This inhibits problem solving that requires multiple pieces of information in different palaces.

This is only from my personal experience but i have devoted 4+ hrs per day (on average) for the past 12 months to study using memory palace and i dont think im particularly dumb (well at least im in Mensa).

I am DESPARATE to hear from ppl who have similar situations (not necessarily doing medicine) who have had success overcoming this problem.

Everett, I think there is a difference between learning and memorising. I think memorizing is about training neurons whereas learning is about gaining worldly insights or intelligence. When you “learn” about memorizing, you gain insights not about the world but about your own inner world. So learning about memorizing shouldn’t be considered learning unless you think of your own inner world and memory castles as worldly things, of course.
This is why I personally think that education and brain training are not necessarily compatible activities. I think you have to decide what you want to really focus on.
Personally, when I heard that a high education (other than in languages) does not strengthen the brain as much, my decision to focus on brain training became a no brainer.

Not sure if my approach will help, but I often avoid permanently storing information (that I wish to retain long term) in one memory palace. When it comes to refreshing my memory, I try not to rely on simple retrieval. I find repetitive retrieval of the same image tends to create a rather static and, for me, an unsatisfying result.

The three principles that drive my thinking are:

  1. to try to form dynamic connections where the information is associated with changing contexts, not a repetitive, static contexts
  2. to find strategies to frequently use the information, instead of always relying on straight retrieval practice
  3. to start, soon after memorizing the info, to create ongoing connections with the ‘real’ interactive world, as opposed to only focusing on establishing associations in the fantasy universe of mnemonics.

In concrete terms, here are some of the things I (often) do:

  1. After I have placed a set of items in one memory palace (or set of pegs), I often move them to another (or link them to a set of pegs, chain them together, etc)

  2. When doing retrieval practice, I try to constantly transform the memory, rather than keep it static. For example, at some interval after memorizing a scene, rather than just retrieving it as I first encoded it, I might try to imagine the prequel to the scene–i.e.: I would try to dream up an earlier scene that led to the crazy mnemonic image that I want to reinforce. Later, I would imagine the sequel, if you understand my meaning. (There is more theory behind this approach, but that’s another discussion.)

  3. I turn the information to be memorized into a set of loci or pegs, which I then use to recall other information. This is especially useful for memory items that I want to store but only have infrequent natural uses for, such as the periodic table of elements

  4. I spend time thinking about and using my memory objects to create more real world associations and interactions, rather than sticking with in mnemonically encoded contexts. In my view, it is important to begin disrupting the exclusive linking to the mnemonic universe as soon as possible.

Not sure if my approach is clearly explained, or if it is what you are looking for, but a metaphor that seems apt is that one approach to using memory techniques is like pinning a butterfly to a corkboard and leaving it there for future inspection, while the other is like trapping it momentarily, but then encouraging it to fly again.

Best regards,



I think that you are right with the assumption that memory palace can be some kind of rote learning due to the fact that it is based on out-of-context repetition of specific pieces of information. Moreover the research papers that I read suggest that rote memorization weakens remembering the context of a given information and improves recalling the information itself. So even the results of rote learning look similar to the one obtained by using memory palaces.

You also need to take into account the types of retrieval. Memory palace allows for organised recalling of some pieces of information that does not require searching all your memory. During normal retrieval of information (without a memory palace), which involves the testing effect (Roediger & Karpicke), an additional mechanism appears that organises the information in your memory, improving the speed of retrieving them from memory. As a result of this process the access to the pieces of information improves, facilitating the problem-solving ability. In case of memory palace such organising mechanism can be absent or weakened due to lack of appropriate factors triggering it. It can be the cause of problems connected with using memory palaces.

Very interesting discussion; very hard-to-answer questions. I have never thought about or felt m1313s “symptoms”. Really thought-provoking.

I wonder if the nature of most images we conjure and also the kind of memory palaces we use might have some take on all that. I risk saying that the most common approach to come up with images for abstract or weird-named concepts (such as the ones in Medicine) is to use sound resemblance. Our memory palaces are well-known places to us or places we make ourselves acquainted with just for memorisation purposes; these usually have nothing to do with the information being memorised. Therefore, the moment we “see” our images in the context of their memory palaces, they do not immediately prompt us to think of the concepts we have memorised. At first, they are just nonsensical imagery; only after an interpretation phase they become the concepts we are looking for. The memory palaces themselves do nothing to assist the associations. Maybe that hampers the necessary “jumps” among the five memory palaces m1313s mentioned in his/her example?

Maybe we are looking at this “superhero doing crazy things to a neighbour of ours by the entrance of downtown’s museum” and, although such image is promptly decoded into the stuff we need to memorise for our next exam, the associations it elicits from our natural memory are not the connections we need for problem-solving: they are the feelings of pure awe we had when we visited that museum for the first time or the noise that neighbour used to make every night or how we wanted to fly like that superhero of our infancy. This can all just be going on in the background of our minds, subtle half-thoughts that we don’t even take notice, but they might be enough to lead our subconscious thoughts astray – those thoughts that would, eventually, as if through magic, allow us to make that singular connection to solve our problem.

I’ll think more about that, thanks for the stimulating thoughts.

BTW, m1313s, have you ever used the site Picmonic? I’d be curious to hear your comments about it.


1 Like

Hi m1313s,
Congratulations on your achievements with memory techniques thus far.

Are you not interested or curious about how the material you have uploaded into your memory palace is relevant to other material already in there?

Why wait for someone else to set the question?
Do you agree with everything you heard or read, yet already uploaded to a memory palace?

What conclusions are yours vs. the one’s you memorized?
I haven’t gotten anywhere near the recall ability you appear to have taught yourself, but I have noticed that it is easier to think about material if I have memorized it.

Complex problem solving requires us to evaluate, examine, find supporting reasons, and discard opinions that are unsupported. A trained memory can help in that effort, but you probably need to add an additional tool to your kit.

Simple Guy.


The number of loci doesn’t matter very much (perhaps a minimum of 1,000 reusable loci is a good number to aim for), but FAST processing speed is essential. Can you jump to any place (locus) in a second or two, then retrieve the information (decode the mnemonic) almost immediately after that? This is how it’s supposed to work - and yes, it takes some practice and repetition to achieve real speed. But it will make your problem of “retrieving information stored in 5 different memory palaces” quite easy to do quickly.

You seem to be talking about two different things: 1) memorizing data, and 2) solving problems with this data.

Nothing is permanently “locked” - you can always rearrange anything you want. If you find it difficult to forget the associations you make, then congratulations, maybe you don’t need memory palaces at all. Just make the links and let them float around in “one BIG void,” as you said (although it’s not a void). There are entire mnemonic systems that don’t use loci at all - most famously the one by Richard Grey (died 1771), but also Harry Lorayne and many other mnemonists have ignored or played down loci techniques.

On the other hand, the BBC’s Sherlock Holmes uses his “brain attic” to file away tons of data, on criminals and all sorts of things, AND he uses his “mind palace” to solve problems. This program has brought a lot of attention to mnemonic techniques. I’ve watched the entire series so far, and have found the descriptions of mnemonics to be very disappointing, but the gist is correct: solving problems (meditating on them in this way), and storing information, are two of the traditional uses for memory palaces.

Since memory palaces have been used successfully for many centuries, any new discovery of a “major drawback” would be highly unlikely at this point. There are probably as many different ways to implement loci systems as there are people trying to do this. Mnemonic techniques are creative processes, so experimentation is always required. Or to put it another way, if a memory system isn’t functioning correctly, what can you do to change it, to make it work efficiently? Of course any memory system needs to be rock solid - not something you don’t trust or feel good about.

As a medical student, you are probably aware of first-letter mnemonics, used traditionally in medicine. (See books like Mnemonics, Rhetoric and Poetics for Medics, and similar.) This is a very fast way to tie together info that is scattered around in your brain-box. There is no good reason to limit yourself to just one memory system or technique.



Hi m1313s, your problem is definitely one I’ve run into personally. When I first began using memory techniques for school, I found them clunky to use and not nearly as effective as I wanted them to be. They in fact seemed to be hindering learning, as you describe. I was allowing the “memorization process” to distract me from truly understanding the material. I do, however, think these problems are solvable. I’m currently a medical student like you, entering my third year, and after a few years of experimentation I’ve gotten to a point where I now find memory techniques extremely valuable. Here are what I consider the key realizations I made along the way:

  1. Don’t encode everything using images: I don’t mnemonically encode most of what I can intuitively understand. The goal is to maintain an optimal mix of understanding/intuition coupled with memory techniques as a supplement. If I can understand the mechanisms underlying a process or disease, that’s what I want. And that even goes for terms or concepts I feel in the moment I can learn quickly by rote. Memory techniques are really just there to help me keep the vast swath of material straight, and to help me recall unintuitive yet important details. I try to ask myself: in a few weeks, when I’ve gotten a generally better handle on the material, which images will still be valuable to me? From the get-go, those are the only images I want to make. Another way to say this is that I “memorize” as little as possible. In this way, I minimize (that’s the goal anyway) my reliance on memorization to the detriment of my problem-solving ability. The bottom line is that effective learning requires critical thinking about the concepts at play. If you’re spending loads of time creating images and “memorizing” everything without critical thinking, it should come as no surprise that learning suffers.

  2. Choose loci as you learn & use the palace to help you structure the material: When using memory palaces to learn (unlike for competitions), I don’t “create” my palaces beforehand. I keep a running worksheet of potential palaces, but I don’t choose the individual loci within those palaces until I’m actually in the process of learning. That allows me to choose rooms/areas/loci that most appropriately match the material. In this way, the creation of palaces fits quite seamlessly with my own analysis of what’s worth memorizing and how I should structure it, so I find in fact that it aids in my critical thinking about the material. It also cuts down time spent on “palace building.”

  3. Use spaced repetition: There are definitely tricks for making stronger images (e.g. I shoot for ≤3 images per locus, which helps me), but fading images are an inevitable problem to some extent. No matter how great my images are, I generally find myself having to review the information a few times. I use spaced repetition using Anki to do that. After about 2-3 reviews, the info tends to stick well (although I still review it indefinitely as it pops up in Anki). I definitely also try to work in as many practice problems (from Qbanks like UWorld, etc.) as well. There’s no substitute for that.

Hopefully this all makes some sense. My key realization was definitely #1, and I think that it solved for me the issue that you seem to be running into, while still allowing memory techniques to “work their magic.” I’m curious to hear your thoughts.



Thank you for all the replies and sorry for taking so long to respond (as uni has been quite hectic this week). Btw, just a bit of background about myself, I am a final year medical student currently trying to use memory palaces to remember everything I need for the wards.

1000 loci might seems a lot if you are using memory palaces for other purposes, but unfortunately for the purpose of studying medicine that will be like “fitting an elephant into a sedan”. I estimate roughly 60000 loci are needed, so if one decide to use 1000 reusable loci, that will mean he/she has to fit condense 60 loci into a single loci. I doubt if anyone can recall without significant delay with such a crowded loci.

The answer to that question will be a certain “YES”. I have no issues with speed, I can recall almost instantly. Two seconds will be too much. To survive in medicine using memory palace, one has to able to jump to any loci and retrieve information as fluently as if you were being asked the question “What is your name?”. The key to that is what AlexM has mentioned, one needs to fit memory palace to the information, not the vice versa. That is, the memory palace chosen has to be compatible with the information that one is trying to remember.

Just a word on that, I am strongly against over-relying first-letter mnemonics. It fades really quickly in comparing to other memory methods and recall is slow once you have too many of them (given one has not yet sync the information to their natural memory).

Yes, I have lots of experience with picmonic. It’s a good website but there are some issues with it besides the “problem solving issue”. Firstly, the information that is in the picmonic cards do not coincide with information that I find to be important for the wards. Also, the cards seem rather crowded, I find it hard to add even an extra 4 - 5 pieces of information to the card, while still keeping the items in the card in harmony with each other.

The cards are less reliant on loci, alot of times the items seem scattered and they seem to rely more on having a story to tie the items together. The biggest issue with this will be recall speed, since I would like to be able to visualize and extract the information straight away without slowing myself down to think about the story. The other issue with this is I find information fades faster without a loci to anchor to.

Sorry, I forgot to mention that I always make sure I fully understand the material before putting it into my memory palaces. Because otherwise, that would be like having a book with you that you have never read. The experiment that I was doing in the past 12 months was “completely relying on memory palaces to memorize materials (after I have understood the material) and see how that compares with friends who were previously on a similar level (i.e. with similar exam scores) but study via the conventional method”.

Finally, many thanks to suggestions given by Tarnation and AlexM which I find very helpful and their ideas have enable me to come out with a solution to the problem.


The 1,000 loci (or a few thousand, if you like) are supposed to be dynamic. That’s why I said “reusable” loci, and assumed the following:

  • the things to be remembered are first grouped logically (even just into very broad categories)
  • many facts can be grouped (logically) within a single locus, by simple association
  • a truly overcrowded locus would evoke a "room" somewhere else, or a mini-journey, etc. (i.e., improvise memory spaces as needed, related to a specific locus)
  • the context of the memory search is very important! Since you always know the category or general "target" this will greatly reduce any interference from irrelevant data, within a locus.
  • the mnemonics are reviewed a few times, over weeks or months
  • much of the info is eventually recalled without loci (through further review and/or use)
  • one doesn't try to "completely rely on memory palaces in learning everything" as you did. This would be an enormous waste of mental resources, IMO. (Although I guess it might work for some people.)

But you were complaining about slow speed in your original post:

I don’t think it’s fair or useful to draw sweeping conclusions from just one year of personal experimentation, about memory systems that have been used since antiquity. Memory palaces aren’t for everyone. There’s plenty of hype from certain quarters: anyone can do this, everyone should do this, it’s all extremely easy, it works well for everything, and so on. This is all wrong, but it doesn’t follow that there’s any “major drawback” to using memory palaces generally.

Great, so what was the solution? Better organization?


1 Like

Hi Sheldon,
Sorry if I haven’t made myself clearer. What I meant slow is that to access multiple memory palaces to answer a question. Not about recalling about any particular loci.
And about the drawback, it is about applying memory palace to learning in areas that require problem solving. What I said was just my personal experience, that’s why I made this post to ask you guys for your opinions.

Regarding the solution: Previously, I was revising my memory palaces mainly using Anki and this involves revising each memory palace separately. And I believe the solution to the problem is to build “Bridges” between memory palaces. That is, instead of using the majority of time to revise each memory palace on Anki, I should use the time to tackle problems using memory palaces. That way, not only the memory palaces are revised, the brain can also start to build crosslinks/bridges between the memory palaces to enable faster retrieval of information from multiple palaces for problem solving. In more detail, i will be using memory palaces to slowly solve problems, during which i would have revised the memory palaces but more importantly discovering important links between the palaces that enabled me to solve the problem. I will then use mnemonics to remember those links. In this way, each memory palace is still self contained, but there is a link between them. Hopefully this will allow the brain to simultaneously access more memory palaces, thus facilitating problem solve.

The reason why I haven’t posted the solution out was because I wanted to test out whether it works first before posting the solution.


i personnally think it is a matter of speed, when you first memorise anything even with the classical method you’re always gonna struggle trying to retrieve the information fast; but as you do it certain number of time your speed increases significantly. And more, if you’re using a mind palace, as you get used to retrieving from a loci, it becomes useless to travel throughout your whole palace before finding what you need, at that point a simple mental jump is required to retrieve and i personnally find it interesting when it comes to problem solving. In fact, before solving a problem requiering long term memory, you first need to bring the information involved on you’re working memory, and with enough speed, that is done within seconds, almost instantly, and then you can solve your problem, wich require another brain process. However, a thing one should not do is to memorise things just for the sake of doing so, without trying to put on a bit of understing. Understanding could be like the wires linking the palaces datas among them and to the “big void”. That said, everything is a matter of training.


There were great insights from all of your posts, I am in stage where I can memorize lists, numbers and dates, I don’t find any difficulty in using a peg, major or a loci systems, but when it comes memorizing book for competitive exam I find it extremely difficult in using the above memory techniques. So I have arrived at solution that to use one other memory system alongside with loci system, say for example placing mind maps in the specific loci please help with some analogy or illustration to make me understand better, kindly help me in the above, thank you in advance for your help

1 Like

It is now about 3 years since I have dedicated to applying memory palace to learning. I have provided an update to my experience here for those interested: Applying Memory Palace to Learning (Update)