PROJECT SUMMARY/ABSTRACT Lesion symptom mapping (LSM) is a crucial tool used to make causal inferences about behavior from neuroimaging data. Recent work has suggested that structural white matter (WM) and functional connectivity between cortical regions play an important role in supporting healthy language function. However, any causal role of connectivity in language remains unclear, due to both intrinsic limitations of the cohorts typically studied with LSM and often discordant findings across patients and healthy controls. To shed light on this problem, the proposed project will use a multimodal approach to examine connectivity and language in a large and still- growing dataset of patients undergoing resective neurosurgery. This population (a) regularly experiences transient, site-specific aphasias in the acute period following surgery, (b) is not subject to the same confounds of populations typically studied in LSM, and (c) can be studied using electrocorticography (ECoG) prior to resection, allowing both healthy and aphasic language to be neurally characterized within the same individuals. The central hypothesis is that the neurosurgical cohort will reveal classical language syndromes to be a function of disconnection rather than modular damage, with marked deficits in language arising primarily from lesions to WM bottlenecks supporting functional connectivity within the broader language network. The rationale is that this unique approach will contribute a new and clarifying perspective on language and the brain, allowing us to directly examine the extent to which connectivity is necessary for versus simply involved in healthy language processing. The central hypothesis will be investigated via two specific aims: (1) to use multivariate LSM (MLSM) to determine the extent to which the structural integrity of white matter (WM) tracts predicts fluency and comprehension in the acute period following resective neurosurgery over and above what is predicted by the integrity of classical, cortical language regions alone, and (2) to use network analysis of ECoG to determine whether the resection of tissue that exhibits strong functional connectivity prior to surgery predicts poorer fluency and comprehension outcomes post-surgery. In the first aim, MLSM models based on cortical and WM ROIs will be statistically compared to determine which provide the most accurate predictions of language outcomes. In the second aim, functional connectivity of ECoG from later-resected tissue will be analyzed to determine whether pre-surgical measures of connectivity lead to better predictions of language outcomes. The research proposed here will provide the first multimodal study of language including both MLSM and ECoG, with a distinct focus on the causal role of connectivity in language. This work is innovative because it will make use of a rare cohort, sophisticated multivariate and network-based analyses, and an unusually large dataset to predict language outcome...