- Scott H Young - https://www.scotthyoung.com/blog -

Is There Value in Ideas You Can’t Remember?

When I give learning advice, as a rule, I suggest active recall. That’s the process of giving the answer to the question without looking at the solution.

The best way to understand active recall is to look at more passive review strategies. Rereading notes, for example, is not active recall because you never need to produce the answer to a question without looking at it. Instead you are simply making the ideas increasingly familiar.

Students, for the most part, should focus on active recall strategies, not passive ones. The majority of testing methods for academic subjects focus on being able to retrieve an answer, from a question, without looking at your notes. Even open-book exams or assignments require recall, since you have to recall at least enough to know where to locate the answer in your notes.

For practical subjects, recall is also important. To speak a language, you need to be able to produce sentences without a translator. It doesn’t suffice to enter the sentence into Google Translate and feel like you knew it all along.

Given the prevailing importance of recall for both students and practical learners, it might be worth asking whether there is any point in learning something if you can’t recall it later. Despite the weaker value of unrecallable information, I believe there are still a lot of cases where it can be useful to have learned things you can’t recall.

What Does it Mean to Recall Information?

First, I think it makes sense to clarify a definition of recall. I would define recall as being able to retrieve a piece of information, given a particular prompt. The choice of prompt is important, however, as some information may be recallable under certain prompts but not others. (I’d add that I suspect that all information requires some prompt to be able to recall it, but that for really flexible knowledge the range of prompts is large enough that specifying it isn’t necessary.)

This point may be confusing, so I’ll give an example. I’ve spent the last two years learning Chinese characters, with around half of them having been learned during my stay in China [1]. According to Anki [2], I’ve learned roughly 2500 characters, and with a successful recall percentage of 85%, that means I probably “know” about 2100 of them.

However, in all of these cases, what I “know” is being able to correctly state the pronunciation and meaning of a character, given its image as a prompt. This is a skill very useful for reading and writing characters using alphabetic input.

I have learned how to handwrite some characters, but it is far less than 2100. If being able to successfully recall and handwrite a character counts as “knowing” it, I would guess I only “know” around 100 characters.

So which is it? Am I able to recall 100 characters or 2100?

It simply doesn’t make sense to talk about what information you can recall without also including the prompt. In this case, I can recall 2100 characters’ meaning and pronunciation, given the image, but I can only recall how to produce the image given the definition in a small fraction of those cases.

The same is true of learning other foreign vocabulary words. Often it is possible to recall the meaning of a foreign word, but not be able to produce the word, given the meaning. Saying that I only “recognize” the word in one case but “recall” it in another is incorrect. It’s more accurate to say that I am able to recall the meaning, given the word, but not recall the word given the meaning.

For information to be truly unrecallable, there would be no prompts that could bring the information up. However, as a practical measure, I’ll call information unrecallable if the only prompts that can bring the information up contain so much of the original information that they aren’t going to be useful in practice (for example, showing 95% of the image of a Chinese character isn’t a useful prompt for “recalling” the remaining 5%).

Is There Any Benefit to Unrecallable Information?

Given our definition—that recallable information is information that you can access, given a range of at least somewhat useful prompts, does unrecallable information have any value?

Maybe it will be helpful to reword this example more specifically: if you read a book which you can’t recall any of the contents, the thesis, ideas or suggestions, was there any point in reading the book?

Obviously a book you can’t recall anything from is less valuable than one you can. But I’d like to argue that, contrary to popular wisdom, there is still value in a book you can’t recall.

The value of unrecallable information could potentially come from two different places.

  1. Unrecallable information can be relearned faster.
  2. Unrecallable information can have unconscious influences on how you process future information.

Relearning Information You’ve Forgotten

The first value of unrecallable information is that it can be relearned faster. One way this might work is that recall requires a certain threshold of activation to retrieve. Any pattern of memory laid down that is below this threshold can’t be accessed directly.

However, just because it can’t be accessed doesn’t mean it doesn’t exist. If you were to relearn the information, those inaccessible patterns might get boosted back to threshold, and be retrievable again. Since the pattern was already there in a weakened form, it will require less practice than learning the first time.

Forgotten Ideas Influencing What You Learn Now

An alternative effect could be that forgotten information will influence future learning, even if it is no longer directly accessible. Paul Graham has an essay here where he muses about this possible value of forgotten ideas [3].

It’s hard to come up with examples of this effect because, in order to be a true example, we must necessarily be unaware of its impact. But I’ve had many experiences of information that was only weakly recallable influencing my beliefs, so it’s hard for me to believe that the effect would suddenly have zero impact once they go below threshold.

One pseudo-example of this type, was when I was reading an article that was supportive of group selection [4] in biology. At the time, I had not remembered reading a book arguing against group selection and the specific arguments made against it were completely lost to me. But I recall feeling skeptical of the new article I read, nonetheless. It was only later did I stumble upon the old book and realized that I had read an argument against group selection previously.

Similarly, I might forget how to say a particular word in Spanish, but if someone suggests an incorrect word to me, I may have a feeling it is wrong, even though I’m unable to recall the correct word.

Why Value Forgotten Ideas?

Obviously recall matters. And, as a rule, I still recommend practicing active recall when learning, with few exceptions. But I worry that by promoting active recall as a successful strategy, there is a tendency to believe that information which can’t be recalled might as well not be there in our brains.

I sometimes hear from people who stress because they feel they’ve forgotten most of the things they learned in school, books or past experiences. While this is unfortunate, the conclusion that they draw, that therefore there is no current value in any of these experiences is wrong. Information, even if it feels forgotten, can still have subtle, useful effects.

If I spend my life reading books, and I end up having read 5000 books, it would seem quite a shame if I couldn’t recall the main ideas and arguments of 3000 of them. But, I wouldn’t argue that I would be exactly as smart about all things had I only read the 2000 for which I can actively recall. The silent 3000 influence my ability to relearn and shape what I believe even if their traces in my brain are no longer directly accessible to me.