Spaced repetition: Forgetting what I have learned

I habe been using mnemonics for quite some time and I like building images in my head to remember things. However, there is one thing I struggle with a lot and I am curious about how you could solve this:

I started learning the capitals of all counties in the world by building images that link the county name to the capital. I use Anki to learn the country-city pairs and then repeat them like the spaced repetition schedule suggests. I get most pairs correct for the 3 first repetitions, but then when the time in between gets longer, I forget my links completely and the card returns to the first step and must be learned again.

In the beginning, I was very enthusiastic about being able to learn all these facts so fast but now I feel frustrated and have the feeling I’m doing something wrong with my learning.

So my question for you is: is this normal? Do you also forget a lot of cards when the reviewing intervals get longer?
What can I do to get better at the long term recall?

1 Like

I haven’t had this problem, but I have two suggestions:

  • Index duplicates of all your cards into Anki. I did this for learning my binary digits system, when I wanted very fast recall. Then on average the interval halves.
  • Be more conservative with telling Anki you know the card. If it’s very clear then say you know it. If you have to struggle a little to remember it, then rate it less highly. In fact, the first time you go through the cards, you could even rate them all as not well-known as a default policy.


  • Slovenia =
  • Ivory Coast =
  • Kyrgryzstan =

Hi, @Zebrain, I find that sometimes it’s the preparation of the image that causes my failure after a longer period where I don’t review instead of the review. Could you give an example of how you see a country and a capital visualized and associated?


Sometimes the default anki intervals need adjusted.


Are you using location, e.g., memory palace or journey? Even when I am using Anki I put items somewhere. I find it’s much more effective for me.

1 Like

Thank you for your comments. Maybe it is as simple as that and I just need to modify the default intervals like you proposed. Sometimes the obvious is not always so obvious :grinning_face_with_smiling_eyes:

@thinkaboutthebible An example for my kind of associations is: Ukraine - Kiew
Ukraine makes me think of the UK flag and a person I know called Rainer. So I see Rainer waving with an UK flag.
Kiew makes me think of “Kiefer”, the German word for jaw. So to connect the two I see Rainer, waving the flag like a torrero towards a giant jaw that is chasing the flag while snapping for it like and heading towards the flag like a bill in a bull fight.

Do you think I could improve my association in some way so that it sticks longer.

Also thanks for the suggestion @deeptravel. When revising your learning material, do you go through your palace then or do you just use your palace for putting the material in there when learning it for the first time?

1 Like

@deeptravel I just found the description of how you use anki in another thread, here

Very interesting, thank you. Do you have an example of how such an anki card looks like? I’m always curious to see how others use anki in combination with mnemonics.

1 Like

@Zebrain I think you have a good amount of strength in your images that allow you to create an association. But I think the association only has meaning for these two items if you don’t plan to expand your knowledge of them. If I were to diagram it with my SEA-IT data types which are:

Subject Action Item Terrain
subject enhancements action enhancements item enhancements terrain enhancements

Ukraine would look like:

Rainer waving a UK flag ?
as a torero ? ? in a bullring

and Kiev would be

ein Kiefer snapping ? ?
like a bull while running ? in a bullring

What you notice is that you have most of the basic four types covered when you create your merged image which is really a short narrative with the two. The enhancements to each data type are all there except for the item enhancement of the UK flag. Is it big, small, heavy, twinkling with lights, etc?

Also, you notice that the two form a scene where they share a terrain or location. That probably isn’t going to happen separately.

One thing I think you need in case you plan to associate more items to each country is a standard way to visualize a capital. That could be a city you see on top of Rainer’s head ( is that a Hauptstadt?) and for any subject that represents a country. Now you can see the bullring in the city where the Kiefer is running at a torero. The torero doesn’t have to be Rainer.

I also would create the SEA-IT images independently for the country and city so they can be reused. Add details to Ukraine that are appropriate to the Ukraine. That way when you go to Spain and want to create a bull fight scene, it won’t be a problem. I enjoy food. If Rainer is wearing his UK chef’s uniform, he might be stabbing a crispy brown Chicken Kiev and getting hot butter on it in his coastal restaurant.

As Kiev is the “mother of Rus cities,” I might use a mother wearing a headscarf putting in a shiny new pair of dentures in her large Kiefer in her simple bathroom. Now an association can be asked “why did the mother go to Rainer’s restaurant in the coastal city?” A: To eat a Chicken Kiev with her new dentures.

The association query that is the one for the captial of Ukraine would be “what is going on in the city on top of Rainer’s head?” A: a woman in a headscarf is eating Chicken Kiev with her dentures getting stuck in them while the butter runs down on Rainer’s chef’s uniform.

Detail is important for each visual image to help recall. But it has to be relevant to that image and then the image is much more easily associated with another image by a replacement or addition of an action in most cases or other types of interactions. This is where a scene type of association would be created when you merge the two to exist in the same location. To get the fullest visual sentence for each image, fill out the SEA-IT data for each country and city and then make an association.

This is just my system for creating memorable narratives and I hope you find it helpful. The data types and other terms aren’t used by other people yet as far as I know.


I found an interesting article about adjusting the interval:


Stack spaced repetition with the knowledge from the forgetting curve. Review the information that you learned in the previous day, for at least 10 minutes prior to 24 hours occuring.

1 Like

It’s pretty basic. I just orient to the locations or, if I have the loci/stations numbered, the numbers. If I have the stations numbered I will try and mix up the order.

So a card in a non-numbered might be “kitchen cabinet to desk.” That would prompt me to retrieve whatever is in the stations in between.

And a numbered card might be “1, 7, 23, 31.”

1 Like

Thank you all for your input, I tried to train with adjusted intervals over the last days and can already see how it makes a diffefence.
I’ll also try to play around with putting the images in a memory palace for the next subject that I’m going to learn.

I decided to practice names and faces a little more as remembering names is something quite useful in my job. But instead of learning random names that I just use for practice, I decided to get names and images from my network on linked in. That way, I can practice with real names and at the same time learn some names of people I might talk to in the future.

I will use anki again and put the face on one side and the name on the other side. I will also add a little background information about the person like their role as I found that this helps for long term memorization.

1 Like

An update from my side: I recently started reading the Memory Code, one of the best books I’ve ever read.
With this and your advice in mind I started putting my countries into a memory palace.
I just started two days ago but I can already see an astonishing difference compared to learning the counties and capitals just as pairs. The information feels so much more secured and I can easily review it while being on a walk, I enjoy these review sessions a lot and therefore review the countries much more often than I did when I just stored them as pairs without the palace.

For the profile picture - name combinations it works well without a palace and just st reviewing them trough Anki because I use some feature in the face as a peg here and therefore the information is not floating around in my head but has a dedicated peg. But maybe I’ll also try to store the names in a palace to see whether I’ll also experience a difference here.