Abstract:  While some previous studies have found significant transfer effect from working memory to fluid intelligence, other studies have failed. The discrepancy may due to individual difference. One type of individual differences is the working memory training improvement. It was found that, transfer effect was found only in subjects who showed significant training improvement. Another type of individual differences is the cognitive ability at baseline, such as baseline fluid intelligence. It remains unclear how such individual differences modulate transfer effect in working memory training. Specifically, the aim of the present study was to investigate how the individual fluid intelligence at baseline modulates the working memory transfer effect. In total, 40 college students were recruited and randomly assigned into active control group (N = 19, 8 males/ 11 females) and training group (N = 21, 9 males /12 females). The training group was asked to complete a dual n-back task. The participants were asked to perform the training 25 minutes a day, 5 days per week in four weeks. The dual n-back task was computerized, in which participants were required to determine if the stimulus position and voice in the current trial were the same as that in the previous n-1 trial. The n was adaptively changed according to the participants’ performances. Meanwhile, the active control group received a scientific knowledge reading training. To make sure the participants’ engaged in the task, the reading material was different for each time. The training time setting in the active control group was the same as that in the training group. All participants were tested by the Raven’s Standard Progressive Matrices (RSPM) before and after the training. In order to avoid the impact of repeated measures, the RSPM were divided into two parallel tests and were counterbalanced across groups and test sessions. The training group showed significant improvement in the dual n-back task, with an average maximum n = 4.86 and mean improved n = 2.51 after 20 days training. In addition, the results have revealed three key findings. First, we found significant group by test session interaction. Specifically, while the RSPM scores were comparable across test sessions in the active control group, the score of RSPM was significantly improved in the post-test than the baseline in the training group. Secondly, to reveal the potential interaction between baseline intelligence and training score improvement on the transfer effect, comparison of moderators was performed by using hierarchy regression. The results revealed that, intelligence improvement was positively correlated with working memory training improvement and negatively correlated with baseline intelligence performance. Additionally, the interaction term of training improvement and baseline intelligence performance positively correlated with intelligence gain. The interaction suggests that, a person can gain the best if he/she showed highest baseline intelligence performance and highest working memory training improvement. In summary, the current study confirmed that working memory training can improve fluid intelligence. More importantly, the results demonstrated that individual difference, i.e. the baseline level of the fluid intelligence in the current project, has modulated the transfer effect from working memory training to fluid intelligence. The results thus suggested that, future studies should pay more attention on individual difference, to reveal the trainability or transfer gain variance across participants.

Key words:  working memory training, fluid intelligence, individual difference, transfer effect