Psychological reactivity to discrepant events: Support for the curvilinear hypothesis