Linda is not a bearded lady: Configural weighting and adding as the cause of extension errors.

Abstract

This article explores the configural weighted average (CWA) hypothesis suggesting that extension biases, like conjunction and disjunction errors, occur because people estimate compound probabilities by taking a CWA of the constituent probabilities. The hypothesis suggests a process consistent with well-known cognitive constraints, which nonetheless achieves high robustness and bounded rationality in noisy real-life environments. Predictions by the CWA hypothesis are that in error-free data, conjunction and disjunction errors should be the rule rather than the exception when pairs of statements are randomly sampled from an environment, the rate of extension errors should increase when noise in data is decreased, and that adding a likely component should increase the probability of a conjunction. Four experiments generally verify the predictions by the hypothesis, demonstrating that extension errors are frequent also when tasks are selected according to representative design.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 64 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.