What is the chance that preschool children inferring the causes of events are more rational in a fundamental way than current medical researchers using statistics to test causal hypotheses involving a binary outcome variable (e.g., reduced sugar intake lowers the risk of death by coronary heart disease)? You would likely say, a big zero. When we think of intuitive reasoning, we think of the myriad heuristics and biases to which we are prone, and we look to science for rational answers. The opposite of this expectation happens in the case of causal induction. This talk explains why in this case, Nature is wiser than the human scientific method. It shows that intuitive causal inference, even in preschoolers, is more coherent than the current experimental-science framework for causal hypothesis testing. This framework is associative in the sense that statistical measures do not make use of assumptions about causation. The crucial difference lies in the use of the concept of causal invariance—the sameness of the influence of a cause on an outcome across contexts in which the occurrence of other causes of the outcome may vary. Our analysis calls into question an assumption in the scientific method: the assumption that, when augmented with the principles of experimental design, associative statistics allows causal inference. It also suggests a new causal statistics.
This research was done in collaboration with Dr. Mimi Liljeholm (UC Irvine) and Dr. Catherine Sandhofer (UCLA).