A Causal Power Approach to Learning with Rates

Current models of causal induction are seriously compromised because they cannot represent variations in cause-effect timing. Theoretical considerations and empirical evidence converge, suggesting that cause-effect timing influences induction beyond mere interference, in line with predictions of psychophysical models of rate comparison. Rather than accepting two distinct cognitive processes for causal induction from rate vs. probability data, this paper shows that a current normative theory of probabilistic causality (Cheng, 1997) can be extended to encompass rate data. Causal induction in “experienced vs. described” situations (Shanks, 1991) may be rooted in a unified process after all.