FIGURE 3.

Individual performance enhancement in working memory training is associated with level of impulsivity reduction (experiment 1). (A) Change over time in working memory task accuracy was measured to examine whether or not the individuals showed performance enhancement. Average accuracy within each 10-trial sliding window (with 5-trial overlap) was calculated per participant. Indeed, the participants showed a significant performance improvement over time (Pearson’s correlation r = 0.64, P = 2.97e-5). Each dot represents mean accuracy within each window across participants, and error bars indicate s.e.m. (B) Although statistically not significant, individuals’ overall performance (i.e., accuracy) in the working memory task had a trend of positive correlation with the changes in their delay discounting rates (r = 0.32, P = 0.16, bootstrapping P = 0.16). Specifically, individuals who had the highest (lowest) accuracy in the working memory task showed a relatively larger (smaller) reduction in their estimated impulsivity (positive Δlog k indicates one’s impulsivity reduction). (C) Independent of individuals’ initial accuracy level, individuals’ speed of working memory performance enhancement (performance slope) was significantly correlated with level of impulsivity reduction (r = 0.46, P = 0.036, bootstrapping P = 0.034). Each dot represents an individual, and solid red lines are robust regression lines. (D) By dividing the working memory task into five bins (bin size = 1/3 of the entire task length, sliding window with 1/6 overlap), a stark difference was observed between bins in the association between individuals’ working memory accuracy and their impulsivity reduction levels. Particularly, individuals’ working memory task performance in the later period (fourth and fifth bins) showed strong associations with the extent to which they make less impulsive choices in the post-intervention ITC task. *P < 0.05.