We used ERP methodology to investigate how readers validate discourse concepts and update situation models when those concepts followed factive (e.g., knew) and nonfactive (e.g., "guessed") verbs, and also when they were true, false, or indeterminate with reference to previous discourse. Following factive verbs, early (P2) and later brain components (N400 and late frontal positivity) revealed that relative to true concepts, both false and indeterminate concepts were more difficult to validate, and only indeterminate concepts were ultimately updated into the situation model. Following nonfactive verbs, there was no evidence of situational model updating for any condition. However, there was a clear N400 gradient that suggests the lower commitment of nonfactive verbs leads to less incongruence with discourse context for the indeterminate condition than the false condition. These results provide novel insight into how pragmatic constraints afforded by verbs influence discourse validation and the updating of situation models. (Contains 2 tables, 3 figures, and 2 footnotes.)