Recent demonstrations of statistical learning in language acquisition have pro- vided insight on possible connections between language and general cognition. Hu- man languages evolve both on a population and on a personal-psychological level, that of language acquisition. This article addresses the research surrounding learn- ing, working memory, and statistical learning from an evolutionary, neuroscienti c, and psycholinguistic perspective. I will argue that working memory restrictions af- fect the complexity of detecting grammatical irregularity, and that there is an inverse correlation between working memory resources and statistically accurate implicit learning. Used here was a semi-arti cial language learning methodology, exploring the correlation between the two aspects of working memory, namely spatial and lin- guistic, with statistical learning. The participants performed a computerized task, during which they were presented with examples of the language. A numerical recall or a spatial manipulation vigilance task was used as distractors in di erent groups. A nal testing phase examined the ability of participants to reproduce the statistical irregularity included in the language.