Gaze behavior during predictive and diagnostic reasoning

Abstract

Many current theories of causal learning and reasoning offer causal structure as the answer to a long-standing question: what allows people to distinguish causal relationships from non-causal relationships? One component of causal structure is causal priority – the idea that causes precede effects – and that this implicit understanding of temporal order is integral to peoples’ ability to reason about causal relationships. In experimental settings, causal order can be manipulated by presenting the cause and effect in two orders: a predictive order where cause antecedes the effect, and a diagnostic order where effect antecedes the cause. Manipulating the causal order often changes the properties of both learning and making inferences about causal relationships. The purpose of this dissertation is to investigate differences in reasoning in both predictive (cause-effect order) and diagnostic (effect-cause order) causal relationships. In two experiments, gaze behavior measured via eye tracking was used as a metric of processing during causal reasoning. Gaze behavior was selected specifically to build upon prior investigations of causal order effects that examined measures of response speed and difficulty. In the first experiment, human participants completed a causal judgement task on an eye tracker. The second experiment had participants complete a similar causal judgement task, however, some of the information on screen was obscured during judgement making. In Experiment 1, it was observed that participants engaged in more processing during causal diagnostic trials than during causal predictive trials. The results of Experiment 2 are mixed, but offer interesting possibilities for the future of causal order research.