I continually train my ten-year-old son’s working memory, and urge parents of other young children to do likewise.  While I have succeeded in at least temporarily improving his working memory, I accept that this change might not be permanent and could end a few months after he stops training.  But I also believe that while his working memory is boosted so too is his learning capacity.    

I have a horrible working memory that greatly hindered my academic achievement.  I was so bad at spelling that they stopped counting it against me in school.  In technical classes I had trouble remembering what variables stood for.  My son, in contrast, has a fantastic memory.  He twice won his school’s spelling bee, and just recently I wrote twenty symbols (letters, numbers, and shapes) in rows of five.  After a few minutes he memorized the symbols and then (without looking) repeated them forward, backwards, forwards, and then by columns.    

My son and I have been learning different programming languages through Codecademy.  While I struggle to remember the required syntax of different languages, he quickly gets this and can focus on higher level understanding.  When we do math learning together his strong working memory also lets him concentrate on higher order issues then remembering the details of the problem and the relevant formulas.     

You can easily train a child’s working memory.  It requires just a few minutes of time a day, can be very low tech or done on a computer, can be optimized for your child to get him in flow, and easily lends itself to a reward system.  Here is some of the training we have done:     



  • I write down a sequence and have him repeat it.
  • I say a sequence and have him repeat it.
  • He repeats the sequence backwards.
  • He repeats the sequence with slight changes such as adding one to each number and “subtracting” one from each letter.
  • He repeats while doing some task like touching his head every time he says an even number and touching his knee every time he says an odd one.
  • Before repeating a memorized sequence he must play repeat after me where I say a random string.
  • I draw a picture and have him redraw it.
  • He plays N-back games.
  • He does mental math requiring keeping track of numbers (i.e. 42 times 37).
  • I assign numerical values to letters and ask him math operation questions (i.e. A*B+C).        



The key is to keep changing how you train your kid so you have more hope of improving general working memory rather than the very specific task you are doing.  So, for example, if you say a sequence and have your kid repeat it back to you, vary the speed at which you talk on different days and don’t just use one class of symbols in your exercises.



New to LessWrong?

New Comment
32 comments, sorted by Click to highlight new comments since: Today at 8:42 AM

As someone who worked in the area of Intelligence training I am very, very skeptical. For example, there was a burst of optimism about training working memory through Dual N-Back tasks, bought about by a revolutionary paper from Jaeggi et al. Then... not quite nothing but close.

I suspect there's a reason that no braining activity, ever, has been consistently shown to improve intelligence at the construct level. It may be that more specific capacities related to intelligence (like working memory) are improved, and that these affect life outcomes and practical capacity to grasp concepts etc, but this has yet to demonstrated to my satisfaction.

Given your prior knowledge you should be skeptical, but given his experience James Miller should also continue to do this training. It's seems low cost - no more stressful for Alex than other activities that parents often force their children to partake in and with a potential very large payoff.

Also, James and Alex are going much further than the constraints of what a study can realistically expect of people. To use an analogy. Suppose you did a study on whether a bench press can increase muscle mass. Suppose you didn't know that bodybuilders exist. Suppose you ran the study for 2 months, with twice per week bench press sessions. Even with a very large sample size, do you think your results would have supported the possibility of extreme body building? I doubt it. It's very hard to push the average person to the level of training needed to see substantial gains. But it doesn't tell us what is possible.

What type of learning do you estimate would have the highest expected payoff for a smart 10-year-old? Also, something is causing the Flynn effect so we know that environmental changes can boost IQ.

My working assumption is that the Flynn effect is mostly to do with improving the average eg by improved nutrition and wellness due to higher availability of food and health care. We're seeing improvements in the average because we're lifting up the bottom quartile, not the top.

But I think the performance on IQ tests for people at the top have been going up, at least until recently.  It could be, however, that you are right and the increase in IQ test performance reflects real gains in intelligence for those in the bottom, but fake "just getting better at taking tests because of repeated exposure to tests" gains for everyone else.

How much of the research is done on children? Did you train children?

What does working memory have to do with spelling?

I guess it is more long-term memory.

This reminds of Kvothe from "the Name of the Wind" (a magician prodigy who's mentor trained him very hard, emphasizing memory and concentration, allowing him to best every other magician).

Upvoted for:

1) Awesomeness 2) Actually putting something into practice and observing what happens instead of spending endless time theorizing and worrying about whether it the "best" use of time while getting nothing done (which I think is a fairly common problem in this community).

It sounds like Alex is benefitting a lot right now from having a large working memory. It is nice to have enough memory that you don't get left behind. That's usually not been a problem for me, but when it happens I can end up wasting an hour of potential learning. I don't know that many kids would benefit as much as Alex, I think most kids would get bored and frustrated more easily.

My only concern is whether Alex's gains in working memory are permanent and if not whether he can keep up the training needed to maintain it. If no to both, then he might be at a disadvantage later on, especially if he's surrounded by people who naturally have that level of memory. Although perhaps it would still beneficial overall.

Much of the effect you attribute to training is likely just maturation of the brain. There's going to be a lot of natural improvement in cognitive functions until your son reaches his 20s. Can't see harm in the exercises though, and I think you can assume by default that training will improve at least the specific functions trained.

Much of the effect you attribute to training is likely just maturation of the brain.

Certainly possible.

He does mental math requiring keeping track of numbers (i.e. 42 times 37).

I notice that being able to enjoy juggling large amounts of digits requires a certain predisposition.

I do lots of math with my 9 year old and he likes to do (and discover) tricky (at his level) properties of numbers like noticing that 3 cancels in 3/2 times 4/3 (he doesn't know the 'rules' governing multiplication of fractions but somehow juggles the numbers nonetheless - in this case he notices that 3/2 is 1,5 is one and a half...). I'm quite sure he enjoys it as I talked about this exercise with his older brother when he chimed in. But he doesn't like mentally multiplying two-digit numbers (though he can). He rather does rough estimates.

Why am I telling this? There are differnt kinds of working memory and other brain functions. And doing multiplication exercises can be fun but need not. Why do it? Could be that it's easy/natural. Could be that it is fun. Could be that it is challenge. Could be that it is hard work but wanting to conform to expectations may do the trick. What's it in your (sons) case?

My son is mathematically inclined in terms of skill and interest.

This is a very interesting suggestion that I may try out. To make sure I see the effect hope to do a before and after test. And do it with all my four sons.

But first I have some questions:

  • It could be that he just natively has a good working memory. I was always surprised by the very good episodic memory of my oldest son. And the rote memory of my second oldest. And the motor memory of my third. All untrained.

  • Are you sure this provides a benefit/is wirth the effort? We live in a digital age where we do have access to lots of short-term memory aids. - This is a genuine question I'm not sure on myself.

My personal guess would be the training itself doesn't help much. But the memory techniques do. Why? Because I playd lots of Concentration Game with all my sons and it is very difficult for all of them. But after I introduced building stories around the visible cards they were able to pick up the technique and at least improve somewhat.

I'm sure genetics plays a big role my son's strong working memory, but his progress while training has been fast enough that I'm more than 90% confident that the training has significantly helped. Memory aids (like spell check) do reduce the value of having a good working memory, but don't come close to negating it especially for children who will be taking many closed book exams, and because having to keep looking up things (like the syntax for a for loop) costs time, attention, and flow.

When did he start this working memory training? Were there pre-cursors?

I don't remember exactly when he started, but I know we were doing memory training when he was 6. The first thing we did was him repeating back a numerical sequence he heard from me.

I started doing comparable games (repeat long words, non-sense words and sentences) when they were quite young (basically growing out of talking with them when they started learning to talk). With the older trying too. But these games only confirm the same thing: Repeating a sequence by rote is hard (for them). Repeating whole sentences is much easier - but it likely happens that only the meaning is transported not the literal form. Songs are even more easy - but I'd guess that that's due to other types of memory (rhythm?) cueing in.


Re: nonsense words - I use it to teach English to my 4-year old when we go outside. He sees a car - I tell him that it's an 'elephant's car', and say 'elephant' in English - he returns 'elephant-mobile':)

Interesting. Number sequences consist of objects that are all of the same type, they are all on the same level of abstraction. What if you asked him to remember not only numbers themselves, but also things that are on a different level of abstraction, e.g. the structure in which they are arranged? For example, if you made him memorize a tree, you could ask him to traverse it in some order.

Or perhaps you could say a short number sequence that has some non-obvious hidden pattern and ask him to discover that pattern and tell you the next number in the sequence.

Excellent ideas!

Have you thought about teaching him to have a Memory Palace?

Yes, but I haven't done it yet.

I think it's great you do fun stuff with your kid that also helps him learn. Have you tried doing the exercises yourself, to improve your own memory too?

Yes, and I often play lumosity.com.

Does it work? I'd love to get better at remembering people's names.

You get better at the games. I'm uncertain as to the transfer effect.

I often debate whether increasing types of working memory - i.e. strings of numbers, is actually helpful to real life. If it were, then I would be congratulating you on this training. I would be interested if you are aware of proof of its usefulness?

in a related example - the brain-training Nintendo DS games came out; people were able to show improvements in number tasks or sudoku's, but when evaluated scientifically (; at a later time), the improved skill was shown to not be applicable to other areas of life, rather; just good for in-game skill level. (I don't know where the sources for this information are)

I have no proof that what I'm doing is transferable. But in my defense I'm almost certainly doing more general and longer-term working memory training than any video game does, and there would also be a lack of evidence for the effectiveness of (I think) nearly any other type of learning I could do with my son.

The study you mention isn't directly applicable because, as I interpret your comment, the evaluation occurred after people stopped playing and I'm claiming my son will probably benefit for as long as we continue the training.

I don't mean to critique specifically; I like the idea of being able to train memory skill; but on a meta level - maybe an invested approach in searching for or finding the effective methods of memory upskilling would be more valuable than just trying this.

For a baseline maybe include a testing process, or tracking system so that you can evaluate the ability to keep making progress in the area. Also make graphs. Try to add notes that include what might recently affect ability. (some things that I can think of right now include - time of day, sleep, coffee or sugar consumed, prior or post exercise)


It should at least be useful in debates, if the other party begins their argument with 'the worst argument in the world'.

The title of this post wins the "If I Had Anti-Kibitzer On I'd Assume This Is Spam" Prize.