Does this technique reliably increase your fluid intelligence?

I am passing this along without endorsing it (travel prevents me from going through the research):

The n-back task involves presenting a series of visual and/or auditory cues to a subject and asking the subject to respond if that cue has occurred, to start with, one time back. If the subject scores well, the number of times back is increased each round. The task can be done with dual auditory and visual cues, or with just one or the other.

A few years ago, Jonides and his colleagues Martin Buschkuehl, Susanne Jaeggi, and Walter Perrig demonstrated that dual n-back training increased performance on tests of fluid intelligence. But the current work extends that finding in several ways.

“These new studies demonstrate that the more training people have on the dual n-back task, the greater the improvement in fluid intelligence,” Jonides said. “It’s actually a dose-response effect. And we also demonstrate that the much simpler single n-back training using spatial cues has the same positive effect.”

In the so-called real world, who actually gets this kind of training?:

According to Jonides, the n-back task taps into a crucial brain function known as working memory—the ability to maintain information in an active, easily retrieved state, especially under conditions of distraction or interference. Working memory goes beyond mere storage to include processing information.

For the pointer I thank MR commentator JamieNYC.


I have not read this Jonides paper, but I do have collaborators working on this sort of training. N-back is one of their primary areas of interest.

There is some quiet muttering in their lab that people who improve performance on n-back are not doing so because their working memory is improved, but because they are getting better at correctly guessing by refining their internal estimation of the probability distribution of the correct answers.

I'm not saying Jonides et al. are wrong, just that our understanding of the transfer of training after doing this sort of training is still very murky.

> because they are getting better at correctly guessing by refining their internal estimation of the probability distribution of the correct answers.

I don't think that's possible as a complete explanation. The default base-rate of a 'match' either audio or visual in DNB (in Brain Workshop) is 20%, IIRC, and then you would have to guess whether to do audio, visual, or both. So a perfect mimicking of the internal probability distribution would give you.... what, 10% correct answers? The default settings under either Brain Workshop or Jaeggi-mode would kick you down a N level for performance that terrible.

That kind of an explanation is one of the more convoluted ones I've seen. In my experience, there is no "estimating" going on. You either remember the nth stimulus or your don't.

I have used brainworkshop (a free program that implements this task) for a while now and, while I've certainly improved at n-back type tasks, I can't say that I've noticed any improvement while handling real life problems. I think the effects do generalize - I'm quite good at highly g-loaded tasks like the PASAT now, even without much practice - but the range of tasks which are subject to improvement from n-backing seems limited. I'm better at tasks involving mental updating, but my short term memory has only slightly improved, if at all. I don't have an accurate way of measuring my change in Gf (or g), as most of the fluid reasoning tasks available online use the same/similar rule patterns or aren't accurately normed, but as I said before, my real life problem solving abilities have not subjectively improved. I expected the effects to be dramatic given the size of the effects reported in the initial study and the fact working memory and general intelligence are nearly isomorphic constructs, but it certainly hasn't been that way for me.

At the risk of self-promoting, I will point out that the DNB ML is one of the main discussion forums about DNB (

My FAQ summarizing a lot of group discussion:

Research supporting DNB:

Criticism and/or failure to replicate:

As for this article - we already know about & read material on the adult studies which quite surprisingly suggested that *Single* N-Back is better than Dual N-back (this would be Jaeggi 2010 in the FAQ: ), but the article/press release also mentions training & testing of children, which is something we hadn't heard about. I looked around a little today and couldn't find any existing publications or other discussion of it, so maybe it's still in progress.

Or you could just play video games.

"In the so-called real world, who actually gets this kind of training?"

Live poker players.

Or anyone who listens to music.

“In the so-called real world, who actually gets this kind of training?”

Computer programmers. The quality of code you write is generally a function of how much of the structure, and the details, you can keep in working storage in your head.

Anyone writing a paper using a mark-up language and the "ed" text editor (yes, I know this dates me). When efficient navigation through the document is limited to "Jump to where I used ", you have to hold a lot of details in working storage.

The example of computer programmers is to confound the issue slightly. The typical programmer has an arsenal of programming _languages_ at their disposal, and their extensive familiarity with any given language is a direct reflection on their Long-Term Memory. It's more likely that programming is like playing chess (or speaking language), where Long-Term Working Memory has a greater influence on the final resultant performance, i.e., infinite chunking of a program as one iteratively expresses long-term information through a working short-term filter.

There are two things, not just working memory, that has to be kept in mind there. At some point, WM capacity won't matter, but LTM always will. If anyone doesn't get my meaning, peruse this: .

From my career experience I agree, programmers need to do this sort of task a lot.

To the extent you find it easy to understand how all the pieces of a set of a code fit together, you can be a much more effective programmer -- and this is a real limiting factor for a lot of people in the field, perhaps even the one that washes out the most new entrants.

This seems especially relevant in a work environment in which there are constant distractions (such as this blog) and interference (multiple support calls, co-worker chat, etc).

Anyways, always good see my buddies at ONR. I'm really pulling for their Polywell project.

That should read, "Jump to where I used <phrase>"; either the server or my browser ate what looked like an HTML tag.

whew! At least you remembered!

There's a game on Wii Fit that's similar to the single n Back. Learning to quickly squat on the right answer is the trick.

It's helped by cardio fitness, but I can't recall if my memory is improved.

This would be a great product (n-back task) to market to the masses. Like "est training" seminars of past (now owned by Landmark Education). Fools could sign up for a premium price, take "seminars" and come out the other end feeling so much intelligent/superior then the rest of us. If anyone would like to get this rolling, I would like to assist with the marketing/sales end of this idea.

You guys are fucking pussies for censoring comments that aren't PC. Don't you realize you are arming the Sailer's of the world? They get to claim that your economic theories are what they are because you are afraid of being non-PC? So now you censor people like me and end up making the Sailers seem like they have a point? Well, fuck you.

I of course don't know what happened with your comment, but sometimes it does happen that their blogging software catches some comments and puts them up for moderation, for such reasons as malformed href's. Both these professors, I would expect, would be too busy to regularly check the moderation queue.

Dirk's follow-up comment leads me to wonder whether the blogging software might have been on to something. The "when my comments are moderated, Steven Sailer wins" argument definitely gets points for novelty, though.

I did this for about a month and the most notable change was a much greater consistency in my ability to remember my dreams.

I played around with this yesterday for a few minutes. Last night I ended up having a lucid dream. I have had lucid dreams before but, they aren't particularly common for me.
N = 1 but, might make for an interesting experiment to see if the n-back does effect dream recall.

Here's a card game that you can play with your kids that mimics 1-back:
Pull out the 10,J,Q,K,A in all suits. Flip them one at a time and the other person has to say if the card matches the previous suit or rank. Vary timing as needed.

This shows which they last very much lengthier and thus saving you income which could otherwise are actually utilized to purchase new ones.

Comments for this post are closed