Review of working memory training programs finds no broader benefit

July, 2012

A meta-analysis of 23 studies has found no evidence that working memory training has wider cognitive benefits for normally developing children and healthy adults.

I have said before that there is little evidence that working memory training has any wider benefits than to the skills being practiced. Occasionally a study arises that gets everyone all excited, but by and large training only benefits the skill being practiced — despite the fact that working memory underlies so many cognitive tasks, and limited working memory capacity is thought to negatively affect performance on so many tasks. However, one area that does seem to have had some success is working memory training for those with ADHD, and researchers have certainly not given hope of finding evidence for wider transfer among other groups (such as older adults).

A recent review of the research to date has, sadly, concluded that the benefits of working memory training programs are limited. But this is not to say there are no benefits.

For a start, the meta-analysis (analyzing data across studies) found that working memory training produced large immediate benefits for verbal working memory. These benefits were greatest for children below the age of 10.

These benefits, however, were not maintained long-term (at an average of 9 months after training, there were no significant benefits) — although benefits were found in one study in which the verbal working memory task was very similar to the training task (indicating that the specific skill practiced did maintain some improvement long-term).

Visuospatial working memory also showed immediate benefits, and these did not vary across age groups. One factor that did make a difference was type of training: the CogMed training program produced greater improvement than the researcher-developed programs (the studies included 7 that used CogMed, 2 that used Jungle Memory, 2 Cognifit, 4 n-back, 1 Memory Booster, and 7 researcher-developed programs).

Interestingly, visuospatial working memory did show some long-term benefits, although it should be noted that the average follow-up was distinctly shorter than that for verbal working memory tasks (an average of 5 months post-training).

The burning question, of course, is how well this training transferred to dissimilar tasks. Here the evidence seems sadly clear — those using untreated control groups tended to find such transfer; those using treated control groups never did. Similarly, nonrandomized studies tended to find far transfer, but randomized studies did not.

In other words, when studies were properly designed (randomized trials with a control group that is given alternative treatment rather than no treatment), there was no evidence of transfer effects from working memory training to nonverbal ability. Moreover, even when found, these effects were only present immediately and not on follow-up.

Neither was there any evidence of transfer effects, either immediate or delayed, on verbal ability, word reading, or arithmetic. There was a small to moderate effect on training on attention (as measured by the Stroop test), but this only occurred immediately, and not on follow-up.

It seems clear from this review that there are few good, methodologically sound studies on this subject. But three very important caveats should be noted in connection with the researchers’ dispiriting conclusion.

First of all, because this is an analysis across all data, important differences between groups or individuals may be concealed. This is a common criticism of meta-analysis, and the researchers do try and answer it. Nevertheless, I think it is still a very real issue, especially in light of evidence that the benefit of training may depend on whether the challenge of the training is at the right level for the individual.

On the other hand, another recent study, that compared young adults who received 20 sessions of training on a dual n-back task or a visual search program, or received no training at all, did look for an individual-differences effect, and failed to find it. Participants were tested repeatedly on their fluid intelligence, multitasking ability, working memory capacity, crystallized intelligence, and perceptual speed. Although those taking part in the training programs improved their performance on the tasks they practiced, there was no transfer to any of the cognitive measures. When participants were analyzed separately on the basis of their improvement during training, there was still no evidence of transfer to broader cognitive abilities.

The second important challenge comes from the lack of skill consolidation — having a short training program followed by months of not practicing the skill is not something any of us would expect to produce long-term benefits.

The third point concerns a recent finding that multi-domain cognitive training produces longer-lasting benefits than single-domain training (the same study also showed the benefit of booster training). It seems quite likely that working memory training is a valuable part of a training program that also includes practice in real-world tasks that incorporate working memory.

I should emphasize that these results only apply to ‘normal’ children and adults. The question of training benefits for those with attention difficulties or early Alzheimer’s is a completely different issue. But for these healthy individuals, it has to be said that the weight of the evidence is against working memory training producing more general cognitive improvement. Nevertheless, I think it’s probably an important part of a cognitive training program — as long as the emphasis is on part.

Reference: 

Melby-Lervåg, M., & Hulme, C. (2012). Is Working Memory Training Effective? A Meta-Analytic Review. Developmental psychology. doi:10.1037/a0028228
Full text available at http://www.apa.org/pubs/journals/releases/dev-ofp-melby-lervag.pdf

[3012] Redick, T. S., Shipstead Z., Harrison T. L., Hicks K. L., Fried D. E., Hambrick D. Z., et al. (2012).  No Evidence of Intelligence Improvement After Working Memory Training: A Randomized, Placebo-Controlled Study.. Journal of Experimental Psychology: General.
Full text available at http://psychology.gatech.edu/renglelab/publications/2012/RedicketalJEPG.pdf
 

Recent posts at Mynd

A study involving 97 healthy older adults (65-89) has found that those with the “Alzheimer’s gene” (APOe4) who didn’t...

An Indian study involving 648 dementia patients, of whom 391 were bilingual, has found that, overall, bilingual patients developed dementia 4.5...

A study, involving 371 patients with mild cognitive impairment...

A study involving 206 spousal and adult children caregivers of dementia sufferers (mostly Alzheimer’s) has found that about 84% of...

A study involving 254 people with dementia living at home has found that 99% of people with dementia and 97% of their caregivers had one or more...

A new U.S. study suggests that Alzheimer's disease and other dementias are markedly under-reported on death certificates and medical records....

It’s often argued that telling people that they carry genes increasing their risk of Alzheimer’s will simply upset them to no purpose...

11 new genetic susceptibility factors for Alzheimer’s identified

The largest international study ever conducted on Alzheimer's...

Understanding a protein's role in familial Alzheimer's...

A brain imaging study of 162 healthy babies (2-25 months) has found that those who carried the ApoE4...

A gene linked to Alzheimer's has been linked to brain changes in childhood. This gene, SORL1, has two connections to Alzheimer’s: it...

Analysis of data from 237 patients with mild cognitive impairment...

Two studies indicate that young people carrying the “Alzheimer’s gene” (ApoE4...

Analysis of data from more than 8,000 people, most of them older than 60, has revealed that, among the 5,000 people initially tested cognitively...

Analysis of 700 subjects from the Alzheimer's Disease Neuroimaging Initiative has revealed a genetic mutation (rs4728029) that’s associated...

Analysis of brain scans and cognitive scores of 64 older adults from the NIA's Baltimore Longitudinal Study of Aging (average age 76) has found...