The development of multi-resolutional systems for language models, with varying focuses on different segments of reading and learning, could lead to gaps in knowledge being filled and a more comprehensive understanding of information. A fully recurrent model with more degrees of freedom would be necessary for accurate predictions.