Working memory can be improved by training

In experimental subjects, which are usually undergraduates majoring in psychology, working memory can be improved (emphases mine) ...

Memory Training Shown to Turn Up Brainpower - New York Times

... The key, researchers found, was carefully structured training in working memory — the kind that allows memorization of a telephone number just long enough to dial it. This type of memory is closely related to fluid intelligence, according to background information in the article, and appears to rely on the same brain circuitry. So the researchers reasoned that improving it might lead to improvements in fluid intelligence.

First they measured the fluid intelligence of four groups of volunteers using standard tests. Then they trained each in a complicated memory task, an elaborate variation on Concentration, the child’s card game, in which they memorized simultaneously presented auditory and visual stimuli that they had to recall later.

The game was set up so that as the participants succeeded, the tasks became harder, and as they failed, the tasks became easier. This assured a high level of difficulty, adjusted individually for each participant, but not so high as to destroy motivation to keep working. The four groups underwent a half-hour of training daily for 8, 12, 17 and 19 days, respectively. At the end of each training, researchers tested the participants’ fluid intelligence again. To make sure they were not just improving their test-taking skills, the researchers compared them with control groups that took the tests without the training.

The results, published Monday in The Proceedings of the National Academy of Sciences, were striking. Although the control groups also made gains, presumably because they had practice with the fluid intelligence tests, improvement in the trained groups was substantially greater. Moreover, the longer they trained, the higher their scores were. All performers, from the weakest to the strongest, showed significant improvement...

... No one knows how long the gains will last after training stops, Dr. Jaeggi said, and the experiment’s design did not allow the researchers to determine whether more training would continue to produce further gains.

Fluid intelligence is a part of what determines IQ scores on tests.

I reviewed the PNAS abstract, but there wasn't much there (psychologists don't do structured abstracts, alas). The PDF costs money -- PNAS isn't part of the "open access momvement". I'd like to know how big the improvements really were.

One of my children's working memory is very limited. A 10 to 15% improvement might make a big difference in what he's able to do. If other studies replicate these results we might hope to see a training program developed.

Update 7/5/08: Hoisted from comments, by Martin of MindEvolve:

I was so impressed that I splurged for the full study. The improvements weren't subtle. Over the course of 19 days, the number of objects held in working memory just about doubled, and scores of raw intelligence increased by as much as fifty percent.

I was so impressed that I contacted Jaeggi and her team and developed a software program using the same method so that anyone can achieve these improvements at home. (For just $9 more than purchasing the study!) ...

Martin is marketing a product he's developed here, but he's been at this for a while. So it's intriguing ...

Update 12/14/2008: The full text of the article is now available for free online. See my Dec 2008 f/u essay.