Bayesian and Heuristic Models of Human Causal Inference

A longstanding area of enquiry concerns how humans use covariation to learn about causation. An influential research program in psychology claims that people are essentially Bayesian in their causal inference, as well as in their cognitive behavior more generally. I explore this hypothesis as it pertains to how people learn about the relationship between a single cause and effect. I challenge the Bayesian view with a simple heuristic model, which mirrors the Bayesian solution in certain key respects, while at the same time is more parsimonious since it makes fewer representational commitments. Finally, I argue that the heuristic model is empirically preferred since it predicts patterns displayed in human judgments that are strongly inconsistent with Bayesian models.