Research into the roots of ideological extremism has traditionally focused on the social, economic, and demographic factors that make people vulnerable to adopting hostile attitudes toward outgroups. However, there is insufficient empirical work on individual differences in implicit cognition and information processing styles that amplify an individual's susceptibility to endorsing violence to protect an ideological cause or group. Here we present original evidence that objectively assessed cognitive inflexibility predicts extremist attitudes, including a willingness to harm others, and sacrifice one's life for the group. Across two samples (N = 1,047) from the United Kingdom and United States, structural equation models demonstrated that cognitive inflexibility predicted endorsement of violence to protect the national ingroup, which in turn predicted a willingness to die for the group. These statistical models accounted for an average of 31.4% of the variance in willingness to die for the group, after accounting for demographic variables. Furthermore, cognitive inflexibility was related to greater confidence in the decision to sacrifice one's life in an ingroup trolley problem scenario. Analysis of participants' performance on the cognitive tasks revealed that cognitive rigidity - distinctly from other aspects of cognition - was specifically implicated as a cognitive antecedent of extremist attitudes. Implications for the study of radicalization and identity fusion through a neurocognitive lens are discussed.