Measuring Memory and Metamemory: Theoretical and Statistical Problems with Assessing Learning (in General) and Using Gamma (in Particular) to Do So

In what has become a classic metamemory paper, Dunlosky and Nelson (1992) had participants study paired associates such as ocean–tree. Later, the experimenters represented the same items and asked participants to judge how likely they would be to remember the second word if shown the first word 10 minutes later (i.e., they were asked to make JOLs in the form of predicting their future recall performance). There were two independent variables of interest. The first was delay: JOLs were made either immediately (i.e., the next trial after the words were presented) or after some number of intervening (presentation or JOL) trials. The second was type of presentation at the time of making the JOL: Participants saw either the intact cue–target pair (i.e., ocean–tree) or the cue alone (i.e., ocean–?). As measured by γ, JOLs were far more accurate in the delayed cue-only condition than any other condition. The superiority of the delayed cue-only condition is an important effect (e.g., for evaluating whether one has studied enough) and has been replicated many times (see Narens, Nelson, & Scheck; Weaver, Terrell, Krug, & Kelemen, this volume, for a review). In a similar study, Nelson and Dunlosky (1991) noted that most of their participants reported trying to silently recall the target word when given a delayed cue-only JOL; that is, they made “covert retrieval attempts.” In our comment on that article, we (Spellman & Bjork, 1992) argued that some of the superiority of the delayed cue-only condition might be due to a self-fulfilling prophecy — because covert retrieval attempts could have two important, if unintended, consequences (see Figure 1). The first consequence is strategic: Participants use the outcome of the covert retrieval as a basis to predict future recall on the final test. That is, if they fail at covert retrieval on the JOL trial, they are likely to assume that they will fail again on the