Comprehending conditionals

• Main Author: Haigh, Matthew
• Published: University of Manchester 2010
• Subjects: 150.724
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.525643

Conditional statements of the form if p then q and q if p invite a reader to think hypothetically about the world as it is, how it could be, or how it could have been. The way we reason with conditional information has traditionally been studied using highly analytical deduction tasks. However, in everyday discourse readers must rapidly and efficiently comprehend conditional statements about real world events, in real time, to maintain coherence with the wider narrative. To date, there has been little attention paid to understanding the cognitive processes that contribute to this type of fast-acting and intuitive comprehension. In the series of experiments reported in this thesis, a temporally fine grained reading time measure was employed to track the real-time processing load associated with the pragmatic and probabilistic factors that constrain the interpretation of everyday conditional statements. These experiments revealed that (1) the scope of hypothetical thinking is differentially constrained for indicative and counterfactual conditionals (2) these constraints differ as a function of clause order (3) readers mentally represent the speech act communicated by a conditional (4) the initial representation of an indicative conditional captures the conjunctive probability of the events described by p and q occurring together i.e., P(pq). Conditionals have traditionally been examined as disinterested arbitrary propositions. Recent advances have extended this analysis to the types of conditional we encounter on a daily basis. The findings presented provide a further advance, showing that the incremental real time interpretation of a conditional statement entails multiple levels of pragmatic and probabilistic representation. If current theories of conditionals are to capture the processes involved in everyday human reasoning, they must be refined by testing their implicit processing assumptions, and ultimately used to generate explicit processing predictions that can guide future empirical work.