The field is grappling with whether to modify the long-standing theory of reward prediction error—or abandon it entirely.