Can intelligence be increased by training on a task of working memory?

Abstract

Claims that fluid intelligence can be significantly increased through working memory training are unsupported by the underlying experimental evidence. While specific cohorts demonstrate modest performance gains on the Bochumer Matrices Test (BOMAT) following training, the non-standardized administration of this assessment undermines the validity of the results. By reducing the allotted testing time from 45 minutes to 10, the evaluation was effectively transformed from a measure of fluid reasoning into a speeded test of the ability to solve simple visual analogies. This time constraint prevented participants from reaching the more difficult, progressive items required to assess high-level fluid intelligence. Furthermore, the lack of improvement on the well-established Raven’s Advanced Progressive Matrices suggests that observed gains were test-specific rather than representative of general cognitive enhancement. The visual-spatial nature of the training tasks closely mirrors the format of the BOMAT, indicating that performance increases likely reflect task-specific practice effects or improved test-taking strategies rather than a fundamental change in intelligence. The reliance on undocumented and potentially irrelevant historical data to justify these procedural deviations further weakens the conclusion that working memory exercises can expand innate cognitive capacity. – AI-generated abstract.