In Defense of Working Memory Training

One minute we're being told that brain training makes you smarter, and the next minute we're told it's all bogus. Confused? I don't blame you. The research literature on brain training is confusing and even sometimes contradictory. This is the way of science. I believe, however, that there is hope in making sense of things if the field and the media can move beyond broad conclusions to look at more nuanced effects.

In his recent New Yorker piece, Pulitzer Prize winning journalist Gareth Cook concluded that working memory training will not make you smarter. According to Gareth, "Playing the games makes you better at the game, in other words, but not at anything anyone might care about in real life." But is this really the most informative conclusion we can draw from the data?

Cook based his conclusion on a recent review by Monica Melby-Lervag and Charles Hulme (download paper). Partly consistent with Cook's statement, the researchers saw little evidence for the generalization of working memory training to other mental skills such as reading comprehension, word decoding, and arithmetic. But that's not all they found. The researchers also found that working memory training programs do produce reliable short-term improvements in both verbal and visuospatial working memory skills. On average, the effect sizes were moderate, but in some cases the effects were actually quite large and rather impressive for brief training regimes which only lasted 12 hours (on average). In fact, the largest effects were found for Cogmed working memory training programs. So while Cook is quite right to criticize Cogmed for claiming too much in their promotional materials, I believe he is too quick to dismiss large and reliable short-term improvements in working memory as meaningless in the real world.

Working memory involves the ability to maintain and manipulate information in one's mind while ignoring irrelevant distractions and intruding thoughts. Working memory skills are essential for everyday intellectual functioning. Multiple research studies show that the inability to control one's train of thought has important real world consequences, from poor reading comprehension to unhappiness. Therefore, just because a working memory training program doesn't generalize to other cognitive functions doesn't mean that the program doesn't make people "smarter," at least in respect to one key aspect of intellectual functioning. After all, intelligence isn't a single ability. There is an emerging consensus among intelligence researchers that general cognitive ability is comprised of multiple interacting cognitive functions, and working memory is one of those crucial intellectual functions.

A telling finding of the Melby-Larvag and Hulme review (but not mentioned by Cook) is that age was a significant moderating variable on strength of training. Younger children (below the age of 10 years) showed significantly larger benefits from verbal working memory training than older children (11-18 years of age). This suggests that lumping people together who are at different stages in their cognitive development can obscure some truly meaningful effects. I imagine if the researchers also included elderly individuals in their analysis (e.g., those above the age of 60), they would have found effect sizes resembling what they found for the youngest age groups (alas, such cognitive decline is part of life).

But equally as troublesome, the researchers didn't look at other crucial moderating variables such as personality, motivation, learning disabilities, mental illness, and socioeconomic status. I believe each and every one of these variables also matter, and ought to be considered in grand reviews of the literature.

Take personality. In one study, participants scoring higher in conscientiousness showed greater improvement in working memory during working memory training compared to less conscientious participants, but they showed less transfer to a measure of fluid reasoning. This suggests that highly conscientious individuals may develop task-specific strategies that prevent generalizing effects beyond the specific skills that are trained. What's more, participants scoring higher in neuroticism (a proxy for anxiety) showed greater improvement on an easier version of a working memory task compared to participants lower in neuroticism, but displayed lower training scores on a difficult version of the task. It seems that in the easier version, higher levels of emotion may have been an advantage, allowing highly neurotic individuals to maintain their concentration and vigilance, whereas on the more difficult version of the task they became overwhelmed by their anxiety. These nuanced effects suggest that personal characteristics should be taken into account when considering the effectiveness of cognitive training.

A related consideration are individual differences in neurotransmitter functioning. Tantalizing recent research suggests that individual differences in the neurotransmitter dopamine, which plays a crucial role in motivation and working memory, is also related to working memory training improvement. In one study, Stina Soderqvist and colleagues found that variation in a gene that codes for dopamine transportation was related to improvements in working memory and fluid reasoning in preschool children following training. While the results certainly require replication (the sample size was small, and the effects did not remain significant after correcting for multiple comparisons), and the effects of a single gene on behavior tend to be vanishingly small, the results do suggest that differences in dopaminergic production may play a role in training outcomes.

Next, it's important to consider that working memory training is most helpful for those who need it the most. A number of interventions targeting specific cognitive domains have been found to significantly improve the cognitive functioning of individuals affected by Traumatic Brain Injury. A particularly promising training technique for improving focus is the Interactive Metronome, a program that involves synchronizing bodily movements to a precise computer-generated beat. Research suggests that this program is beneficial for those with neurological conditions such as brain injury, ADHD, stuttering, and stroke.

In an educational setting, helping students overcome working memory burdens can be particularly helpful. Over the past decade John Sweller and colleagues have designed instructional techniques that relieve working memory burdens on students and increase learning and interest. Drawing on both the expertise and working memory literatures, they match the complexity of learning situations to the learner, attempting to reduce unnecessary working memory loads that may interfere with reasoning and learning, and optimize cognitive processes most relevant to learning. Cognitive Load Theory can be particularly useful for students with working memory deficits who are otherwise extremely intelligent and competent as it allows them to more easily demonstrate their brilliance.

Finally, poverty matters. Elliot Tucker-Drob and colleagues have conducted some excellent studies on how socioeconomic status interacts with the development of cognitive ability. In one study, they found that at the age of 10 months, the home environment was the key variable explaining differences in cognitive ability across different levels of socioeconomic status. The story changed considerably, however, as the children got older and differences in educational enrichment became more pronounced. For the 2-year-olds living in poorer households, the home environment mattered the most, accounting for about 80 percent of the variation in cognitive ability. In wealthy households, however, genetics explained more of the differences in performance, accounting for nearly 50 percent of the variation in cognitive ability. This suggests that socioeconomic status is also an important moderating variable on the strength of the effect of intellectual enrichment.

***

While I believe multiple factors are important when considering the effectiveness of working memory training, I want to be clear that I am not saying that the kind of brain games that were included in Melby-Lervag and Hulme's review (and were the focus of Cook's article) is the best method to improve working memory. The evidence suggests that the activities that show the strongest and most widespread effects on cognitive functioning are those that target the "whole person," such as traditional martial arts training and enriched school curricula. I think we often underestimate the extent to which multiple aspects of development-- cognitive, physical, social, and emotional-- all feed off each other.

Also, while I've already mentioned the promise of Interactive Metronome (particularly for people with disabilities), I also see great promise in meditation techniques for increasing working memory skills in all people. Recent research suggests that mindfulness meditation training can improve working memory and reasoning while reducing mind wandering.

Regardless of the method, however, it's important to be realistic about the limits of brain training. Cook rightly notes that the effects of the large majority of working memory training programs don't generalize well beyond the specific skills that are targeted. This is important to keep in mind. Working memory interventions may improve focus and attention to a large and meaningful degree, but they shouldn't be expected to increase high-level critical reasoning skills or magically alleviate all of the symptoms of a learning disability. To improve logical and critical reasoning, one ought to actually engage in reasoning training. To reduce the symptoms of a learning disability, it's important to engage in comprehensive interventions that specifically targets the symptoms (e.g., phonological decoding interventions for people with dyslexia).

It's also important to keep in mind that regardless of the method, working memory improvements are transient. Repeated practice and challenge is essential to maintaining improvements in any kind of cognitive training or else they'll very likely decline rapidly. This shouldn't be shocking. You wouldn't expect to go on a diet for 12 hours and maintain the one pound you lost the rest of your life. To keep growing and improving intellectually requires constant engagement in intellectually challenging material. To maintain improvements in focus and concentration requires getting in the habit of concentrating and manipulating complex material in your mind.

I don't think any of these caveats, however, should lead us to give up on our search for long-term working memory interventions that provide reliable improvements in working memory across the lifespan. Let's not downplay the importance of working memory improvements, even if they don't generalize to other cognitive skills. I think we owe it to those who are truly suffering day in and day out from the inability to control their train of thought, as well as those who are growing up in intellectually impoverished conditions.

A recent survey found that from 2000-2012, about 250,330 military troops experienced Traumatic Brain Injury in some form. According to the National Center for Children in Poverty, nearly 16 million children living in the United States live in families with incomes below the federal poverty line. As they note, "Poverty can impede children's ability to learn and contribute to social, emotional, and behavioral problems." The potential payoff of working memory training for helping those who could truly benefit from it is too great to not at least attempt to get this right, and take a more nuanced approach that takes into account multiple factors.

If you’d like to learn more about the development of intelligence and the many paths to greatness, you may be interested in my forthcoming book “Ungifted: Intelligence Redefined, coming out this summer from Basic Books.

© 2013 Scott Barry Kaufman, All Rights Reserved

Note: I do not work for any brain training companies and have no horse in this race. I'm just as interested in the data as you are. Thanks to Silvia Bunge for her assistance with this post, and for providing me with some crucial references.