Post-stroke lesion correlates of errors in verbal and spatial production tasks

Introduction: Traditional lateralization models assign post-stroke verbal impairments to the left hemisphere and spatial impairments to the right hemisphere. When considering error measures, this dichotomy may be too simplistic, as performance monitoring may involve domain-general and domain-specific components. Furthermore, the error-monitoring hypothesis predicts domain-incongruent specialization, with left hemisphere dominance for spatial and right hemisphere dominance for verbal errors. Methods: We performed voxel-based lesion-behavior mapping in N = 110 acute stroke patients who completed a cognitively demanding, error-prone, five-point spatial design fluency task and a verbal word-fragment completion task. Results: Significant associations were found between lesion location and error rates in both tasks, spatial fluency (correlation = 0.36, p < 0.001) and verbal completion (correlation = 0.31, p = 0.001). Right inferior frontal lesions correlated with errors in both tasks. In addition, left frontal white matter (WM) lesions were associated with spatial errors, whereas right frontal WM lesions were associated with verbal errors. After adjusting for demographics, the left WM cluster remained significant for spatial errors and the right WM cluster for verbal errors, while the right inferior frontal association with spatial errors was no longer significant. Discussion: Post-stroke performance monitoring involves two distinct neural systems. One is a domain-general system, probably centered in the right inferior frontal region, that supports overall accuracy. The other is a widely distributed, reverse lateralized system, with left lesions associated with spatial accuracy and right lesions associated with verbal accuracy. This suggests that performance monitoring relies on more complex hemispheric interactions than traditional models suggest.