Reading focus
Syllabus: How Do We Judge and Decide?
Trace the search and stopping rules and inspect what the simulated decision maker already knows. Compare performance under the same informational conditions.
Why you're reading it. Make a heuristic precise enough to evaluate its strengths, costs, and possible failure conditions.
A question to consider. If one cue predicts well in one population, what comparison would reveal whether stopping at that cue still works after the environment changes?
Why this reading
Using less information need not mean making a worse prediction. This paper specifies simple decision procedures and compares their performance with more integrative alternatives. It shifts attention from whether a strategy looks sophisticated to what it accomplishes in a defined environment.
Before you read
Bounded rationality studies judgment under limits on time, information, and computation. A cue is a feature used to predict a criterion. Cue validity is accuracy when a cue discriminates between alternatives. A stopping rule says when to stop searching. A noncompensatory rule lets one cue determine a choice without trading it against all other cues.
Read for
- Reconstruct the search, stopping, and decision steps of Take The Best. State what knowledge it requires before it starts.
- Inspect the city-population environment, recognition assumptions, and competing algorithms. Ask whether their information and evaluation conditions are comparable.
- Separate evidence that a procedure can predict accurately from evidence that people actually use it. Identify environmental changes that might reverse the comparison.
Consider afterwards
- Design an environment in which the best single cue is highly reliable and another in which several weak independent cues should help. How would you compare strategies fairly on cases they have not already encountered?
Editorial context
The central evidence is a simulation comparison in a specified environment. It is not a general demonstration that ignoring information is optimal or that people always stop at one cue. The appendix discloses two cue-value transcription errors used in the simulations; the authors judge them immaterial, but this remains part of the evidence record. Gigerenzer and Gaissmaier's later review develops the importance of matching strategies to environments. Process evidence and independent prediction remain separate requirements.