Eight experiments examined whether certain human problem-solving mechanisms should be expected to represent probability information in terms of frequency. Findings are consistent with literature indicating that frequentist representations eliminate various cognitive biases, including overconfidence, the conjunction fallacy, and base rate neglect. Findings undermine the common conclusion that inductive reasoning mechanisms do not embody a calculus of probability. (HTH)