Cognition and Inquiry: The Pragmatics of Conditional Reasoning
Oaksford, Michael Robert
This thesis reports the results of both normative and empirical investigations into human conditional reasoning, i.e. reasoning using if ... then and related constructions. Previous empirical investigations have concentrated on experimental paradigms like Wason's Selection Task, where subjects must assess evidence relevant to the truth or falsity of a conditional rule. Popperian falsification provided the normative theory by which to assess errorful behaviour on these tasks. However, it is doubtful whether this is an appropriate normative theory from which to derive a competence model of human reasoning abilities. The relationship between normative theory and competence model need not be direct, no more than the relationship between competence model and performance needs to be. However, research in this area has imported a theory directly into individual psychology from the philosophy of science. On the apparently orthodox assumption of directness, continued adherence to this import may stand in need of re-assessment in the light of the quite radical descriptive inadequacy of falsification as a model of rational scientific inquiry. However, this model also possesses the virtue of relating the interpretation of the rule directly to the normative task strategy. Hence, this thesis has two aims: first, to retain the virtue of a direct relation between normative task strategy and interpretation while simultaneously offering a competence model which is consistent with more recent and descriptively adequate accounts of the process of scientific inquiry. In Part I, this will involve introducing a semantic theory (situation semantics) and showing that the process of inquiry implicit in this semantic theory is consistent with recent normative conceptions in the philosophy of science. The second aim is to show that the competence model derived in Part I can provide a sound rational basis for subjects' observed patterns of reasoning in conditional reasoning tasks. In Part II, chapter 5, the data obtained from the Wason Selection Task using only affirmative rules is discussed and the behaviour observed rationally reconstructed in terms of the competence model of Part I. A central concept of that model is partial interpretation (motivated by concerns of context sensitivity). Prima facie evidence for partial interpretation is provided by the observation of defective truth tables. However, in conditional reasoning experiments using negated constituents, this evidence has been interpreted differently. A subsidiary aim of Part II (which will constitute the largest section of this thesis) therefore concerns the empirical demonstration of the consistency of this data with the competence model.