CAN RECOLLECTION EXPLAIN MIRROR EFFECTS IN ITEM RECOGNITION? EVIDENCE FROM HUMANS AND A COMPUTATIONAL MODEL
Abstract
This thesis examines the role of recollection in the word frequency mirror effect. In two experiments, participants studied lists comprised of sets of associatively related words and sets of unrelated words from 4 levels of word frequency. Following a short distractor task, participants took an item recognition test with confidence ratings (Experiment 1) or old-new judgments followed by remember-familiar judgments (Experiment 2). In Experiment 1, the standard word frequency mirror effect for comparisons of LF words to HF words was observed, but the effect did not obtain for comparisons of MF, HF, or VHF words. In Experiment 2, a complete word frequency mirror effect was observed, and the patterns for hit rates and remember judgments for targets were almost identical. These findings run counter to predictions from Bayesian likelihood models (Glanzer, Hilford, & Maloney, 2009) but are consistent with the hypothesis that mirror effects are the result of differences in recollectability between stimulus classes (Joordens & Hockley, 2000). Attempts to develop a computational process model to account for mirror effects and boundary conditions on those effects such as those observed in Experiment 1 are also discussed.
Collections
- OU - Dissertations [9305]