Adaptive training leads to sustained enhancement of poor working memory in children

Institution: University of York

Title: Adaptive training leads to sustained enhancement of poor working memory in children

Researcher(s): J. Holmes. D. Dunning, S. Gathercole.

Program: Cogmed RM

Published: Developmental Science, April 2009

Funding: This research was supported by a project grant from the Economic and Social Research Council of Great Britain. Training program, investigator training, and technical support provided by Cogmed. No funding provided by Cogmed.

View a presentation about this study by Dr. Joni Holmes.

Working memory (WM) is the capacity to store and manipulate information for brief periods of time. According to researchers at the University of York, “It provides a mental workspace that is used in many important activities in learning…and is a pure measure of a child’s learning potential.” WM deficits contribute to difficulties with attention and learning for many children with ADHD and individuals with poor WM typically make poor academic progress during the school years. In fact, “…of those children whose WM abilities fall in the bottom 10%, over 80% have substantial problems in either reading or mathematics or, most commonly, in both (Gathercole & Allow, 2008). You can find additional information on Working Memory and its role in learning and attention at

Until recently, WM capacity has been believed to be fixed and it seemed unlikely that the adverse effects of poor WM on learning could be overcome. However, several recent reports indicate that intensive training of WM can enhance WM capacity in many individuals and that this leads to better attention and improvements in daily cognitive functioning. This research has included randomized- controlled trials of WM training for children with ADHD, normally developing preschoolers, adult stroke victims, and healthy younger and older adults. (The study of WM training for children with ADHD is reviewed at In an especially intriguing study, WM training was found to improve fluid intelligence in normal adults – the text of this study can be viewed online at

The authors of the current study reviewed note that “while these early findings look promising, the educational significance of WM training is as yet untested.” In addition, prior studies of school- age children were conducted with children diagnosed with ADHD and did not include youth with poor WM who did not also have ADHD. Thus, the goals of their study [Holmes. J., Gathercole, S.E., & Dunning, D.L. (2009). Adaptive training leads to sustained enhancement of poor working memory in children. Developmental Science] was test whether WM training helps children with documented WM deficits and whether the benefits are sufficient to overcome learning difficulties associated with poor WM.

Participants were 42 eight- to eleven-year-old children attending six schools in the North-East of England who were selected based on scoring in the bottom 15% on a validated test of WM. Children were randomly assigned to one of two conditions – a high intensity (HI) WM training condition and a low intensity (LI) WM training condition that served as the control condition. (Note – Children attending the same school all received the same type of training.)

The HI treatment consisted of performing WM tasks via a computer program developed called Cogmed Working Memory Training. The WM tasks visuospatial tasks – remembering the position of objects on the screen – as well as verbal tasks – remembering sequences of letters, sounds, and digits. In all cases, children responded to the WM task by clicking on various choices with the computer mouse.

Each training session provided exposure to 115 WM trials and required about 35 minutes to complete; training occurred at school during the regular school day. In the HI condition, the difficulty level of the WM trials was adjusted to match the WM ability of the child on a trial-by-trial basis. For example, if a child successfully recalled three digits in reverse order, on the next trial he had to recall four. When a trial was failed, the next trial was made easier by reducing the number of items to be recalled By this method of ‘adaptive training’, children were challenged to work at a level that closely matched their ability and to stretch their WM capacity by presenting more difficult tasks after easier ones were successfully passed.

The LI condition was identical to that described above except that the difficulty of the WM trials remained at a low level throughout, i.e., the number of items children were required to recall never increased beyond two. Thus, these children had the same experience as children in the HI group, i.e., they spent the same amount of time engaging in computerized WM tasks, but they were not challenged to improve. As a result, they were not expected to show the same improvement in WM as children who received HI treatment.

Both groups of children completed a minimum of 20 training sessions spread over five to seven weeks.

Measuring the impact of training
The impact of WM training was assessed in several ways. First, children completed a computerized assessment of WM immediately before and immediately after training was completed using tasks that differed from those on which they trained. Prior research has demonstrated that performance on this WM assessment does not improve the second time it is taken. Therefore, any gains associated with HI training would reflect actual gains in WM rather than mere practice effects.

Standardized measures of IQ and of reading and math were also completed pre- and post training. The reading test was a measure of single word reading which does not place demands on WM in the same way that reading comprehension does. The math test was an assessment of mathematical reasoning, a task for which WM is more important. For students who received HI training, a six-month follow-up was also conducted so that the maintenance of any training related gains could be determined.

Finally, the researchers devised a following directions task that represents a practical assessment of WM that is closely linked to what happens in the classroom. On this task, children listened to an increasingly long set of directions, e.g., “touch the yellow pencil and then put the blue ruler in the red folder…” and then had to perform that designated actions. When they passed a trial, the number of directions to be followed increased on the next trial. The test continued until the child was unable to produce the behaviors in the correct sequence with the number of trials passed to that point serving as each child’s score.


Impact on WM
Children completing the HI training showed significant improvements in all aspects of WM – both verbal WM and visuo-spatial WM. The magnitude of their improvement would be considered large by conventional standards, as the effect sizes were greater than 1.0. In contrast, control subjects who received LI training did not make significant gains in either area. Furthermore, 68% of children receiving HI training had WM scores at post-test that fell in the normative range compared to only 25% of children in the LI group.

WM gains for the HI group remained significant at the six-month follow-up. Although the magnitude of the improvements had diminished slightly, the effect sizes remained in a range that is considered large by conventional standards.

Listening Test
Similar results were obtained on the task that measured children’s ability to follow spoken directions. Children in the HI group showed significant improvement at post-test on this task, i.e., they were able to correctly remember a longer string of instructions, while those in the control condition were not. The magnitude of the gains made by HI children would be considered large and remained evident at the six-month follow-up.

IQ and Academic Achievement
No significant improvements in IQ were found. In addition, no significant gains in reading or math were evident at post-test. However, at the six-month follow-up, children who received HI training showed a significant gain in their mathematical reasoning scores compared with pre-training baseline levels. The magnitude of this effect would be considered moderate, i.e., an effect size of .49.

Summary and Implications
The authors begin their discussion by noting that in a classroom of 30 children, there will typically be four to five who have low WM abilities that hamper their academic progress. They go on to note that their study demonstrates that WM deficits and associated learning difficulties can “…be ameliorated, and possibly even overcome, by intensive training over a relatively short period.” In fact, the majority of children completing the intensive training improved their WM substantially and these gains were maintained six months out.

It should be noted that not all outcomes showed improvement following training. Thus, there were no gains in IQ and gains in word reading and mathematics reasoning were not evident immediately after training. The authors suggest the absence of IQ gains indicates that although IQ and WM are related, the contribution of WM to learning is not directly linked to IQ. They note that gains in academic achievement were not anticipated immediately post- training because achievement gains associated with better WM would be expected to take time to develop. Indeed, this was found for children’s mathematics reasoning as their achievement scores showed significant improvement six months after training ended. The absence of benefits on the reading measure may reflect the fact that the test used – a basic word reading test – does not depend heavily on WM skills. Had a reading measure that is more WM dependent been used, i.e., a comprehension task, perhaps such gains would have emerged.This is only speculation, however, and would need to be documented in subsequent work.

Although these are encouraging results, there are limitations of the study that should be noted. First, the sample size – 42 children – is relatively small and replicating these results with a larger sample would be important. Second, six month follow-up data was not collected on children in the control condition, and one cannot be certain that achievement gains would not have also emerged for this group. This seems unlikely given that no improvement in WM was found for control children at the conclusion of training, but cannot be ruled out entirely.

The study would have also been strengthened by the inclusion of teacher ratings of children’s behavior and academic performance. In a prior trial of Cogmed WM training in children with ADHD, although significant improvement in parents’ ratings of children’s inattentive behavior was found, similar gains were not reported by teachers. Thus, it would have been helpful to document whether teachers of children in this study observed them to show better attention and academic performance in the classroom.

Despite these limitations, results from this study add to the research indicating that intensive training of WM can yield important benefits for children and adults who struggle because of WM deficits.