Reading focus
Syllabus: How Do We Judge and Decide?
Compare equivalent information in probability and natural-frequency formats. Track both accuracy and the proposed calculation, including how subsets enter the denominator.
Why you're reading it. Test whether changing the task can improve a judgment without assuming that all errors reflect a fixed limitation.
A question to consider. If a frequency display improves performance, what further manipulation could distinguish computational simplification from merely drawing attention to previously overlooked information?
Why this reading
The way information is represented can change the work required to reason with it. Gigerenzer and Hoffrage compare probability and frequency formats while making predictions about the calculations people use. The experiments connect a proposed computational advantage with observable performance.
Before you read
Bayesian inference combines prior frequencies with how likely evidence is under competing possibilities. The chance of a signal given a fault differs from the chance of a fault given that signal. Natural frequencies retain counts from a common sampled population and its subsets. Simply replacing every percentage with a count out of 100 does not necessarily preserve that structure.
Read for
- Translate one problem between formats while preserving its population and subsets. Identify which arithmetic operations each representation requires.
- Follow the format-by-menu comparison and the criteria used to classify a response as Bayesian. Separate a correct numerical answer from evidence about the procedure.
- Examine the differences among participants and problems. Ask what the experiments establish about representation and what remains uncertain about a general cognitive architecture.
Consider afterwards
- Of 1,000 components, 20 are faulty; 16 faulty and 49 nonfaulty components trigger an alarm. How would you display these counts so a reader can judge what an alarm means, and how would you assess whether the display helps?
Editorial context
The contribution is a testable account of how information format can facilitate inference, not a claim that frequencies eliminate all errors. McDowell and Jacobs' 2017 meta-analysis supports a facilitation effect while identifying effects of visual aids, task complexity, and information structure. It also discusses unresolved differences between ecological and nested-set explanations. Better performance alone does not uniquely establish an evolved frequency-specific mechanism.