Retrospective confidence judgements usually track participants’ performance in memory tasks, so that experimental conditions which yield high performance also yield high mean confidence. However, Tulving (1981) documented two important exceptions using a 2-alternative forced-choice recognition test: 1) increasing choice similarity (i.e., the similarity between the two alternatives presented together on a given recognition trial) increased performance while reducing confidence, and 2) increasing memory similarity (i.e., the similarity of a given alternative to the memory trace) of an incorrect alternative – from dissimilar to highly similar – reduced recognition performance while having no effect on confidence. We perform a replication of Tulving’s (1981) findings concerning choice similarity in a multi-block design to investigate potential changes in metacognitive monitoring due to experience with performing recognition tests. We further extend the findings concerning memory similarity of foils, manipulating memory similarity continuously by using only similar foils and varying the number of presentations of their parent targets.

Tulving, E. (1981). Similarity relations in recognition. Journal of Verbal Learning and Verbal Behavior, 20, 479-496.