Step Up

So you know how to use memory palaces, want to be good at it. In this series, we cover how to brainstorm potential palaces, choose strong loci, and convert information into images. Cathy and Alex (team #mullenmemory) put it all together by doing a real-life example in real time, showing how they each use these techniques to memorize the "acute pyelonephritis" section from the First Aid review book. A post with our top tips for using memory palaces effectively is below. This article summarizes our argument for using memory palaces (sparingly) during learning.



Alex discusses his suggestions for brainstorming potential memory palaces. Try this palace worksheet.


Do you get stuck when trying to create images? Here's how to create images quickly like a pro.


Alex discusses his suggestions for how you can fill your palaces with memorable, meaningful loci. 


Alex and Cathy separately memorize notes from First Aid about acute pyelonephritis in real time.


Our Top Tips for Applying Memory Palaces Effectively

You should now understand how to begin applying memory palaces to your learning goals. At this point, we encourage you to check out Our Top 3 Don't-Miss Memory Palace Tips. This post summarizes the three key roadblocks—and our solutions after experimentation—when it comes to using these tools sustainably.

memory technique and memory palace for medical students


Interested in a specific topic? Jump to memory palace examples in Medicine, Anatomy, Pharmacology, or Languages.

Check out more advanced FAQs below, or head over to Study Hacks, where we discuss other evidence-based learning strategies and how memory palaces can be integrated into a comprehensive learning approach.


Should I reuse palaces to learn new information?

Alex: For short-term information (e.g. day-to-day info), I do. I keep a few palaces on hand and "overwrite" my images again and again when I want to memorize something quickly. This is also what I do for competition (e.g. keep a handful of palaces I use over and over for different speed cards attempts).

When learning for the long-term, however (e.g. medicine, languages), I never reuse palaces I've already filled (unless I've decided I don't care to remember the original information). Ultimately, this comes down to personal preference, but here are my feelings. I want the information I encode at each locus to permanently remain there for review and retrieval. I like to imbue each location with the "feeling" of whatever material is stored there. For example, for me personally, my high school swimming locker room has become synonymous with interstitial pneumonia. It's my interstitial pneumonia headspace. The local university's music auditorium is my acute pyelonephritis headspace. Overwriting those places with different information--having both old and new images in the same loci--interferes with this process. In my experience, that's confusing and unnecessary. This method means I have to keep making new palaces, which isn't as hard as you might think. For tips on how and where you can create palaces, check out "Doesn't it take too much time to make all these memory palaces?"

I will, however, return to old palaces to add in newly learned info about old topics (e.g. I choose a new locus in the locker room to add a pathogen I've just learned also causes interstitial pneumonia).

If you're dead set on reusing an old palace, my recommendation would be to adjust each locus's vantage or focus point (e.g. if you focused on the TV's screen initially, focus on its top instead), so the loci feel different.

Doesn't it take too much time to make all these memory palaces? How can I make the palace creation process as efficient as possible?

Alex: When using memory palaces to learn (unlike for competitions), I don't "create" my palaces beforehand. I keep a running worksheet of potential palaces (check out Refine Your Technique #1 for a quick video tutorial for generating this), but I don't choose the individual loci within those palaces until I'm actually in the process of learning.

For example, let's say I'm learning about antibiotics, and at the start of the chapter I decided to use my elementary school as my palace. As I learned, I followed a path organically through the school, choosing rooms/areas to neatly house my images for each chunk of info. Let's say I've just used the school lobby as my dedicated area for the drug class tetracyclines. Now, the next chunk of info being about the drug chloramphenicol, I might choose the room just right of the lobby to house all my high-yield chloramphenicol images, selecting loci within that room to fit the facts.

As an additional example, I might select a particular area/room to hold all the info I want to memorize about the disorder hyperaldosteronism, and that area/room would be located within a larger palace about adrenal gland diseases. The next area along my path through that palace would, for example, house high-yield facts about Cushing syndrome, another adrenal pathology.

Alternatively, if I come across an out-of-context list I want to encode (eg, while working practice questions), I might choose the first space that springs to mind (eg, list of insulin upregulators → Turk and JD's apartment from the TV show Scrubs, since Turk has diabetes). The as-needed creation of palaces fits quite seamlessly with my own analysis of what's worth memorizing and how I should structure it, so I find it actually fosters critical thinking about the material.

This approach allows me to choose rooms/areas/loci that most appropriately match the material. It also permits me to easily go back and drop in new information. For example, if I encounter a new, image-worthy fact about chloramphenicol, I simply choose another locus within that original school room.

In this way, I don't spend time beforehand plotting out routes and selecting loci with no visible goal. The as-needed creation of palaces fits quite seamlessly with my own analysis of what's worth memorizing and how I should structure it, so I find it actually fosters critical thinking about the material.

For many, converting information into images is the more time-consuming process, but that's something that becomes effortless with practice. See Refine Your Technique #3 and "Should I try to encode or 'memorize' everything using memory techniques?" for my tips for doing that efficiently.

Also see: "What are the key strengths of using palaces as a learning tool? Why palaces and not just images?"

When using memory palaces, should I "double-encode" information that appears in different contexts?

Alex: I often run into scenarios in which the same piece of information pops up in different contexts. Let's say that--while looking over lists of drugs that cause specific toxicities (e.g. all drugs that cause hepatitis, ones that cause diarrhea, etc.)--I learn that valproic acid is one of a handful of substances that can cause liver necrosis. Since this isn't a particularly intuitive fact, I encode it in my palace containing lists of various toxicities. Later, while I'm studying valproic acid on its own, "liver toxicity" comes up as a side effect.

Since I'm now using a separate set of loci specific to valproic acid, should I double-encode this fact? That is, should I encode Val Kilmer (valproic acid) in my liver necrosis loci and Will Ferrell (liver) in my valproic acid loci?

For me personally, the answer is generally no. I find that doing this forces me to think more broadly, so my palaces act in a more coordinated, inter-connected way. Instead of just listing off the various side effects of valproic acid, I'm forced to scan across my palaces and think more critically about how valproic acid operates and the various places it appears. Of course, if you're keenly aware at every instance in which you double-encode, that problem is likewise solved. Ideally, that should be the case. You should be constantly thinking critically about how new info meshes with the material you've already learned. Personally, however, I've found there's a slight tendency to miss these connections when encoding everything afresh as it appears, so I usually avoid doing it.

This isn't a hard-and-fast rule, of course, just a tendency. For instance, I still encoded the sulfonamides-G6PD connection in both my sulfonamide and G6PD locations.

Should I write or draw out my mnemonics?

Alex: I do write out (generally don't draw) all my palace images so I can reference them later if I need to. It takes more time up front, but I think it's a worthwhile investment.

First, I keep a spreadsheet with all my potential palaces (not the individual loci), coloring in the ones I've already used and noting what I used them for; for example, "University Rec Center (Path: GI)".

Second, for my actual course notes, I keep everything in the spaced-repetition flashcard program [Anki][24]. Each card has a "Mnemonic" field where I write in my images (I also keep these in a consolidated word doc as well, so I can more quickly read through everything if need be). For example, that field might look something like: "[stair railing] EINSTEIN knocks over ROTTEN EGG." I capitalize the things that actually correspond to information, sometimes writing in the association if necessary; for instance, "[stair railing] EINSTEIN=epstein-barr-virus knocks over ROTTEN EGG." I keep the language as concise as possible. I used to write long descriptions of my stories, but I found that doing so took too much time. and was largely irrevelant after I'd spent some time with the material. Now, my description often looks as simple as "[locus] IMAGE."

I also tend to fill the Mnemonic field with all mnemonics relating to the relevant subtopic. For example, each of the x cards about Cortisol will contain all mnemonics relating to Cortisol.

Generally, I try to limit the number of IMAGES to 1-3 following each [locus]. In my experience, it's much better to create more loci with fewer images on each than to stuff many images into fewer loci.

Also see: "What does your Anki (spaced repetition software) setup look like?"

My images generally get hazy after a few days. What can I do to combat this?

There are definitely tricks for making stronger images, but this is 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][24] to do that. The good news is that after 2-3 reviews, the images and ideas tend to stick really well. Here a few tips for making stronger images: Use images/loci that have special significance for you. Try to tie each image strongly into its specific locus. Avoid trying to stuff too many images into one locus (keep it from 1-3). In my experience, it's much better to create more loci with fewer images on each than to stuff many images into fewer loci. Think of a narrative for each locus. Mentally see each locus from multiple angles. When I'm dealing with a particular area of a palace, I try every now and then to "step back" and see all that area's loci together, so as to keep the structure clear in my mind. Also, generally speaking, don't convert everything into images; you should be "memorizing" as little as possible. This of course reduces your number of images. See "Should I try to encode or 'memorize' everything using memory techniques?" for more about that.

Do I need to train regularly (ie. memorize lots of numbers or decks of cards) to become proficient?

It just depends on your goals. If you're looking to become a top competitive memorizer, I'd say 30-60 minutes per day is a minimum. If your primary goal is to get your feet wet with mnemonics and begin applying them to learning, I don't think that's necessary at all. So in that sense, none of the events is "required" training. I do, however, think memorizing a deck of cards regularly (for instance) is a great way to get a feel for the techniques and stay sharp.

I struggle to convert information into images quickly. How can I improve?

This is really one of those areas that gets much better with practice, but check out Refine Your Technique #3 and "Should I try to encode or 'memorize' everything using memory techniques?" for our tips.

We want to re-emphasize that creating images is really just about creating triggers. You don't need to encode the entirety of a term into its image. For example, I use a "crutch" to remember "Krukenberg tumor." Those words are only slightly similar, but the association becomes strong after a few repetitions. Contrary to the examples of "papule" and "pheochromocytoma" we gave in our [Getting Started series][26], we rarely--in practice--break the word down and encode every little piece, because we'll become familiar with the term soon enough. A simple analogy: on your own, you might not be able to recite a particular verse of a song, even if you "know" the song. Given its first few words, however, you can. At that point, there's no real sense in brute-memorizing the rest of the verse with mnemonics. It's the trigger--the first few words--that counts. We use images in a similar way; they need only be cues that bring the relevant info to mind.

Also see: "I'm having problems converting medical terms..."

How clear should my visualizations be?

Alex: Here's my experience, and this goes for both learning applications and memory sports: A key thing to realize is that the visuals themselves aren't always that important, so I don't worry if my images aren't clear. They’re often just fuzzy impressions. The memorability often comes from the story or narrative, simply the idea that image A is interacting with image B (or a particular location) in a semantically interesting way. [I blogged about this following my 2015 USAMC experience][27], and for those interested in digging a bit deeper, that post explains the idea more clearly. When it comes to long-term recall, this idea becomes even plainer. If I haven't recalled something in weeks (or especially months), my image has usually been reduced to its most basic form: an idea in a location. For example, when recalling that toxoplasmosis is one of the clinical uses of the drug combo TMP-SMX, I just have a notion that there is a cat (toxo) on my elementary school's stage (a locus in my TMP-SMX area). No salient sensory impressions, generally--although of course some unique visuals or tactile impressions may stick here or there. I just know there's a cat on that stage, and that's all I need.

What if I run out of loci before I've finished memorizing a section of my notes?

If I run out of loci before finishing, I'll just choose another location to continue my "palace." As an example, let's say I am learning antibiotics and select my college rec center as my palace. I'll place information about each new antibiotic until I’ve used all loci in the rec center. To continue memorizing, I could then move to the college cafeteria next door. For all practical purposes, the rec center and the college cafetria are part of the same "palace" I use to store antibiotics. Conversely, if I finish memorizing antibiotics but find I still have plenty of potential loci in the rec center, I can begin learning my next section of notes, e.g. anti-viral drugs.

What kind of memory palace (real, virtual, etc.) works best? Should I use places I’ve actually been to?

Personally, I tend to use places I’ve actually been to (notable exceptions: the offices from “The Office” and “Parks and Recreation,” to name a few). That said, I’ve seen all different kinds of palaces work for different people. For instance, I’m not a big gamer, so I’ve never drawn palaces from game environments, but plenty do to great effect. The bottom line: use any environment you’re comfortable using.

After you’ve spent time with the material, you’ll find that the details of the palace (e.g. how true to life it is) matter very little. I do try to use palaces in which I can structure the information, so places with distinct areas (e.g. rooms of a building) work better. That way, everything about drug x can go into area y, for instance. As always, two to three reviews will make that structure stick well.

For tips on brainstorming palaces and choosing good loci, see the “Refine Your Technique” series.

Also see: “Doesn’t it take too much time to make all these memory palaces?” for a more holistic approach to using palaces efficiently.

Do you ever erase palaces or actively clean them of old images?

Alex: You might consider making an active effort to erase "ghost images"--images on loci you want to reuse. The short answer is that I don't actively erase old palaces whose material I've decided isn't worth remembering. I believe the alternative strategies--brainstorming and using a new palace, or using a palace so old that its images have faded organically with time--are better. (Using new palaces needn't be painful. You can find our tips here.) If creating a new, ghost-image-free palace is easy, why spend time actively erasing an old one and risk confusing yourself anyway? It's also worth noting that I try to be very selective about what I encode using memory palaces. Generally, I only encode information I'm quite sure I'll want to remember indefinitely (e.g. most high-yield medical information, language vocabulary), so having a palace I want to overwrite isn't usually an issue anyway. If, however, you're dead set on reusing an old palace, my recommendation would be to adjust each locus's vantage or focus point (e.g. if you focused on the TV's screen initially, focus on its top instead), so the loci feel different.

This is all not to definitively say it's never worthwhile to erase palaces. It's just not something I've found necessary, either in competition or learning.

Also see: "Should I reuse palaces to learn new information?"

You recommend “encoding as little as possible.” What if I need to know something verbatim?

I'd recommending reading “Do Memory Palaces Hinder Learning?” if you haven’t already.

It will of course depend on your goals. For instance, if your course necessitates you learn every symptom of pneumonia to be recalled exactly, then yes, I'd absolutely encode every symptom. The two main reasons I try to "encode less" are simply: 1. It helps prevent me from brute-memorizing everything without proper thought, 2. It prevents me from having to wade through a “sea” of images, only 50% (for instance) of which may actually be useful once I've spent a few weeks on a topic. Of course, that's just in service of the goal of learning everything as "best" I can. However, sometimes you just need to know something verbatim (or close to it), so in those cases I'd encode more than I usually would.