Comparing retrieval dynamics in recognition memory and lexical decision

The response-signal method was used to study memory retrieval in the lexical-decision (LD) and recognition-memory (RM) tasks. In Experiment 1 subjects studied mixed lists of words and nonwords and then were tested on old and new words and nonwords, under either RM or LD instructions. The earliest above-chance performance (the intercept) was shorter for LD than for RM. Repetition priming in LD appeared primarily in bias, and the only reliable RM difference between words and nonwords was a bias to respond ''old'' more often to nonwords. In Experiment 2, subjects made lexical decisions to high frequency (HF) and low frequency (LF) words. LF words had a later intercept than HF words, but repetition priming largely eliminated this difference. As in Experiment 1, priming also affected LB bias. The findings have implications for several issues concerning lexical and episodic retrieval and for the ways in which the LD and RM tasks are related.