Neural (re)organisation of language and memory: implications for neuroplasticity and cognition

Background In the presence of neurological insult, how language and memory networks jointly reorganise provides insights into mechanisms of neuroplasticity and can inform presurgical planning. As (re)organisation is often studied within a single cognitive modality, how language and memory interact during (re)organisation in response to epilepsy and the implications for memory outcomes is less clear. We investigated (1) the rates and patterns of joint (re)organisation and (2) their associations with pre- and postsurgical memory function. Methods Individuals with epilepsy (n=162) from three neurosurgical centres underwent the Wada procedure. We examined colateralisation patterns (ie, concordance/discordance) between language and both global and verbal memory (n=34), and associations with clinical characteristics and preoperative and postoperative memory outcomes. Results Overall concordance between language and memory colateralisation was minimal-to-weak across both global memory and verbal memory (kappa=0.28-0.44). Discordance was primarily observed in individuals with left-lateralised language, of whom 52% and 32% showed discordance in global and verbal memory, respectively. Discordance was most pronounced in left hemisphere epilepsy and mesial temporal sclerosis. Conversely, right-lateralised language consistently predicted right-lateralised memory (95%-100%), regardless of seizure laterality or memory type. While discordance was not associated with presurgical memory function, discordance predicted superior postsurgical memory outcomes following surgery in the language-dominant hemisphere (p<0.05; eta(2)(p)=0.30). Conclusions When language dominance is atypical, memory tends to colateralise. However, when language remains typical, concordance with memory is weak, particularly for left hemisphere seizure onset. An interhemispheric shift in language may trigger a shift in memory, possibly to maintain efficient communication between medial temporal and neocortical language networks. In contrast, memory appears able to reorganise in isolation, with discordance predicting better postsurgical memory outcomes without detriment to presurgical function. Our findings support the continued need for separate presurgical mapping of language and memory lateralisation, particularly in the case of typical language dominance and left hemisphere seizures.