Semantic Cognition: A Re-examination of the Recurrent Network “Hub” Model

This paper explores a model of “semantic cognition” first described in Rogers et al. (2004). This model was shown to reproduce the behaviour of neurological patients who perform poorly on a variety of tests of semantic knowledge; thus purporting to provide a comprehensive explanation for semantic deficits as found in patients with semantic dementia and, as extended in Lambon Ralph, Lowe, and Rogers (2007), individuals with herpes simplex virus encephalitis. Therefore, not only does the model emulate these semantic impairments, it also underpins a theoretical account of such memory disturbances. We report preliminary results arising from an attempted reimplementation of the Rogers et al. model. Specifically, while we were able to successfully reimplement the fully-functioning model and recreate “normal” behaviour, our attempts to replicate the behaviour of semantically impaired patients by lesioning the model were mixed. Our results suggest that while semantic impairments reminiscent of patients may arise when the Rogers et al. model is lesioned, such impairments are not a necessary consequence of the model. We discuss the implications of these apparently negative results for the Rogers et al. account of semantic cognition.