Call/WhatsApp: +1 914 416 5343

Does Working Memory Training Improve Intelligence?

Debate Introduction

This report is a critical analysis of the debate on whether working memory improves intelligence. It is also an in-depth critique of two antagonistic journal articles “Improving Fluid Intelligence with Training on Working Memory” by Jaeggi, Buschkuehl, Jonides, and Perrig, and “Working Memory Training does not Improve Intelligence in Healthy Young Adults” by Chooi and Thompson. Jaeggi et al. (2008) obtained a definite link between working memory training and intelligence. While Chooi and Thompson (2012) assert that memory training has no significant effect on fluid intelligence, Jaeggi et al. (2008) finds a significant correlation between repeated sessions of working memory training and fluid intelligence. It is, however, interesting that Choi and Thompson find their motivation for the study from the discoveries of Jaeggi et al. but reach contradicting results. From this critique, it is evident that while working memory training does not improve the arguably innate fluid intelligence, it acts as a remedial intervention tool that impacts training task outcomes in people with intellectual deficiencies.

Article Summary

Article 1. Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perrig, W. J. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences, 105(19), 6829-6833.
In a bid to establish the correlation between working memory training and fluid intelligence, Jaeggi et al. (2008) set an experiment-based study. The researchers conducted four studies, employing novel training paradigms that entailed an immensely challenging working memory task. The 70 participants were presented with two stimuli series synchronized to appear at the rate of 3 seconds per stimulus. The number of items in the task changed both with improved and worsened performance, making the tasks continuously challenging. The memory tasks engaged the learning route needed for the accomplishment of simultaneous tasks, discouraged automatic processes, and task-specific tactics and involved the processes necessary for individual tasks.
The study found an improvement in performance on working memory tasks among all the four participating groups. Additionally, the trained group registered a spectacular improvement in fluid intelligence tests, although the control group also had some improvement (attributed to retests). Interestingly, there was an improvement on fluid intelligence tests with an increased time exposure to the training. The longer the participants were exposed to the training, the better result they registered. It was also found that the improvement in fluid intelligence was purely training-related, suggesting that the impact of the training was not limited to people within particular range of cognitive ability.
Article 2. Chooi, W. T., & Thompson, L. A. (2012). Working memory training does not improve intelligence in healthy young adults. Intelligence, 40(6), 531-542.
In their study, Choi and Thompson (2012) aimed to replicate and broaden the findings of Jaeggi et al. (2008) in a more controlled experimental environment. The study mainly aimed at establishing the relationship between the working memory training and fluid intelligence. To do this, they set to work with a larger sample of 93 participants who were divided into three study groups: active control group, passive control group, and an experimental group. The tests were done in bunches of two periodic seasons to prove Jaeggi et al.’s assertions on mediating effect of training exposure time, with half of the participants taken through an 8-day and another half on a 20-day training sessions. It was found that as participants continued practicing, their performance on the working memory tasks improved, with the variance among the participants tending to escalate with an increasing number of training periods. Interestingly, some participants showed better results in task performance than others. The participants in the 20-day training period showed a 44% improvement against the 34% in those exposed to the 8-day training period. While there was an improvement in task performances, the study found no significant change in intelligence even after the long-term exposure to excruciating working memory tasks. The training neither improved the participants’ fluid intelligence or working memory.
The two groups of scholars adopted a similar research approach (quantitative) in determining the relationship between the identified constructs (working memory training and fluid intelligence). The approach is more relevant in cases where relationships between the study constructs are to be determined to justify its suitability as a research approach in the two studies. Among the various quantitative research designs (descriptive, correlation, and experimental), Jaeggi et al. and Chooi and Thompson adopted an experimental design. The purpose was to ascertain the cause-effect relationship between working memory training and fluid intelligence level. The processes of experimentation emphasize the control of all variables apart from the independent variable (working memory training) that is being manipulated to determine its effects on the dependent variable. Through subjecting the study participants to a pre- and post treatment measurement, the researcher can ascertain the cause and effect relationships between the dependent and the independent variable, further justifying the suitability of the experimental research design in determining relationships.
While the two groups of authors adopted a similar research approach and design in their studies, the treatment process of the study subjects varied. The participants in the two studies were university students; however, Chooi and Thompson only engaged students with a background in psychology, resulting in a stratified sample contrary to the random sample achieved by Jaeggi et al.. Though the simple random sampling technique enabled the researchers to eliminate possible bias in the study, the stratified sample applied in the case of Chooi and Thompson was more effective in receiving proper representation of the target population. The stratified sample provided better coverage of the population to enhance the reliability and generalizability of the findings, thus improving the overall quality of the findings.
The sample size adopted by the two groups of researchers also varied. While Chooi and Thompson attained a sample size of 93 participants, Jaeggi et al. (2008) only managed to engage 70 individuals. Quantitative research approach requires large sample size to attain a proper representation of the target population as well as to enhance the power of analysis. Small sample size evident in the two studies did not only restrict the range of intellectual ability analyzed but also affected the outcome of the test variables.
The group arrangement and treatment sessions adopted by the researchers also differed. Jaeggi et al. randomly distributed the study participants into two groups: control and experimental groups, while Chooi and Thompson had three groups: active control group, passive control group, and experimental groups. The use of the two control groups in the case of Chooi and Thompson (2012) ensured that the researchers were able to regulate simple practice and the treatment effect, thus, enhancing the quality of the findings. The treatment sessions also varied with Chooi and Thompson (2012) employing 8 and 20-day training sessions while Jaeggi et al. (2008) subjected the study participants to four different training sessions including 8, 12, 17 and 19 days training sessions. Even though the training sessions differed between the two groups of scholars, they reported an improvement in the training tasks performance with an increase in the number of training sessions. Therefore, the two groups of scholars reported that the change in task performance was attributed to the training dosage where an increase in the training dosage resulted in an enhanced level of memory task performance.
The data analysis technique adopted in the two studies was effective in determining any differences in the fluid intelligence levels between the control and the experimental groups. The use of ANOVA, t-Tests and ANCOVA analysis as evidence in the two studies did not only allow the researchers to ascertain the changes in fluid intelligence post-treatments but also enabled them to identify whether the changes were associated with an initial ability determined before the test. The effectiveness of data analysis approaches adopted in the two studies contributed to the quality findings attained. As observed, the two groups reported a significant improvement in the training task performance amongst the experimental groups, indicating that working memory training is associated with enhanced memory tasks performance.
Nevertheless, the conclusion drawn from the data collected and analyzed by the two groups of scholars differed. A significant improvement in the training tasks performance was reported amongst the study participants, though no change in their fluid intelligence level was evident, as presented by Chooi and Thompson. On the contrary, Jaeggi et al. combined the results for all the groups after the test to conclude that there is an improvement in the fluid intelligence level and memory capacity. The conclusion was misleading since the study participants were subjected to different tests; thus, the transfer effects varied. Also, the tests were undertaken under similar administration conditions with time constraints making it impossible for the study participants to respond to a similar number of tests effectively. Collapsing the post-test scores and using the results to conclude an improvement in the intelligence level was, thus, inaccurate.
The dubious conclusion drawn by Jaeggi et al. makes their contributions to the debate on working memory training and fluid intelligence improvement less convincing. Though the work of Chooi and Thompson provides satisfactory results on the debate, but more quality findings would have been attained with the use of a larger study sample. As observed, apart from the limitation of an option to switch groups allowed by the scholars, the limited sample size was another constraint that might have affected the quality of the findings.

Further Studies

With studies containing different results, the debate on the effects of working memory training on fluid intelligence is expected to continue. For the future studies, it would be vital to establish the relationship between working memory test performance and fluid intelligence, which would help to clarify transferability of effects from working memory training to fluid intelligence. Since both studies agree that the impact of the working memory training on test performance is dosage based, it will be essential to carry out a future longitudinal study on whether the improvement in test performance stagnates after some period or continue to improve indefinitely. If it continues to improve, it will be important to note whether, at some point in the improvement graph, it impacts fluid intelligence and working memory.


The articles by Chooi and Thompson and Jaeggi et al. present contradicting views on the effects of working memory training on fluid intelligence. A critical analysis of the methodological approaches and the resulting findings indicates that Chooi and Thompson obtained more convincing results. While both studies adopted a quantitative experimental research design in determining the causal relationship between the study constructs, the test administration process and conditions differed, contributing to the varying results they reported. The test administration process adopted by Jaeggi et al. was less effective, thus making their conclusion unsatisfactory. At the same time, the findings of Chooi and Thompson that working memory training enhances the test performance with no effects on fluid intelligence is more convincing. Nevertheless, future studies should focus on determining the link between memory test performance and fluid intelligence to resolve the conflicting ideas presented by different scholars.

Leave a Reply