Sensitivity reductions in false recognition: a measure of false memories with stronger theoretical implications.

Signal detection analyses of recognition memory indicate that a bias to respond "old" is large for critical words that are centrally related with previously encoded word lists, is small for words that are less related, and is not observed for unrelated words. Also, recognition sensitivity has not been previously shown to differ between those conditions, which has focused debate over how to explain false recognition on the bias differences. In 3 experiments, critical-word sensitivity was lower than sensitivity for other word types, but related-word sensitivity was not lower than sensitivity for unrelated words. Extant models that predict reduced critical-word sensitivity also predict lower sensitivity for related words than for unrelated words. These results provide crucial new constraints on theoretical explanations of false memories.