Abstract: Working memory training (WMT) induces neuroplasticity, but its specific mechanism remains unclear. In order to explore the spatio-temporal characteristics of brain function changes induced by WMT, the findings of 37 fMRI papers on WMT in normal population over the last 20 years were investigated with a step-by-step approach, including 5 methods and the following 6 steps, as well as with two assumptions based on the two theories— “Extended Parieto-Frontal Integration Theory” (ExtPFIT) and “neuro efficiency hypothesis”. Step 1, the changes of activation pattern and function connectivity in various brain regions before and after WMT among the 37 papers were compared using descriptive review, frequency analysis and Chi-square test. It was found that 5 association areas, 7 macro regions and 3 sub-regions of the brain were changed by WMT. Especially, the number of articles reporting decreased activation in 3 sub-regions—superior frontal gyrus, inferior parietal lobule and cingulate gyrus was more than that reporting increased activation in them, and this difference was statistically significant. Step 2, 26 out of the 37 papers were used to conduct meta-analysis with activation likelihood estimation method, and it showed that 3 sub-regions with the statistically significant effects in their decreased activation were middle frontal gyrus (BA6 and 8), superior frontal gyrus (BA6) and anterior cingulate gyrus (BA24 and 32). Step 3, based on the comprehensive results of qualitative and quantitative research analyses, a spatio-temporal model of decreased distribution in brain regions of WMT was proposed, and there were 5 results and their discussions obtained with the said model. Step 4, 31 out of the 37 papers were analyzed using non-parametric tests to check the factors that possibly regulated the training effect of WMT, which showed that the respective effect of task type and time of the training on brain activation had statistically significant. Step 5, according to the above analyses, 3 conclusions were drawn regarding the spatio-temporal characteristics of the neuroplasticity induced by WMT in normal population, i.e. 1) WMT can either weaken or enhance the brain activities, and the weakening effect appears more prominent. In addition, updating and shorter time of WMT tend to induce more weakening effects; 2) Such changes occur mostly in the frontal-parietal association area, but also possibly in the temporal lobe, occipital lobe, cingulate gyrus, striatum and other association areas, which reflects the whole brain functional association within a certain range. It highlights the spatial characteristics of the neuroplasticity, which fits in easily with ExtPFIT; 3) The 4 sub-regions of the middle frontal gyrus, superior frontal gyrus, inferior parietal lobule and cingulate gyrus (anterior in particular) in decreased activation mainly reflect the spatial characteristics of the neuroplasticity which fits in easily with “neural efficiency hypothesis” and “smarter brains may be slightly lazier”. Step 6, it was proposed that further research of the neuroplasticity induced by WMT was to possibly focus on low brain activity on its plasticity, distinguish the temporal characteristics of the training effect reflected in 4 sub-regions of middle frontal gyrus, superior frontal gyrus, inferior parietal lobule and cingulate gyrus (anterior in particular), as well as to probe the comprehensive influencing factors that increase or decrease the brain activities.

Key words: working memory training, neuroplasticity, a spatio-temporal model of decreased distribution in brain regions, middle frontal gyrus, superior frontal gyrus, anterior cingulate gyrus