Miller (1990) suggests that communication between linguistics and psychology is hampered for essentially cognitive reasons: linguists favor simplifying explanations while psychologists favor causal explanations. This paper reformulates this suggestion as three testable hypotheses. First, scientists vary in cognitive style from rationalist/nomological to empiricist/mechanistic. Second, linguistics is primarily a rationalist/nomological science, while psychology is primarily an empiricist/mechanistic one. Strikingly, even among nativists, linguistic and psychological research still contrast along the rationalist/empiricist dimension. Third, cognitive styles are relatively intractable, as suggested by empirical evidence that they are associated with intrinsic individual differences and by formal arguments that highlight their self-isolating nature.