During the last two decades, there has been a widespread interest in working memory (WM) training among researchers and the public. As current WM training studies have been criticized for the lack of theory-based approaches explaining the mechanism that underlies WM and transfer, recently, Gathercole et al. (2019) presented a new framework conceptualizing training-induced changes in WM as the acquisition of novel cognitive routines similar to learning a new skill. From this new framework, we can infer that WM training might not always result in positive outcomes when considering the variability of newly formed routines and the impact that routines bring on one’s intelligence. Thus, the primary goal of this study is to demonstrate the negative effect of working memory training. We have already conducted Study 1 fully online using a two-phase training paradigm, which has developed by Susan Gathercole and Dennis Norris at Cambridge. The results indicated that training of circle span backward task transferred negatively to subsequent training of letter span backward task. It is inferred that the impaired performance of letter span backward training may reflect disruptions by incompatible cognitive routines generated in the former training phase. Using a reversed design corresponding to Study 1, Study 2 will be conducted online to further examine the above hypotheses emerged from the results of Study 1.