The Revelation Effect for Autobiographical Memory: A Mixture-Model Analysis
Bernstein, Daniel M., Rudd, Michael E., Erdfelder, Edgar, Godfrey, Ryan, Loftus, Elizabeth F., Psychonomic Bulletin & Review
Participants provided information about their childhood by rating their confidence about whether they had experienced various events (e.g., "broke a window playing ball"). On some trials, participants unscrambled a key word from the event phrase (e.g., wdinwo-window) or an unrelated word (e.g., gnutge-nugget) before seeing the event and giving their confidence ratings. The act of unscrambling led participants to increase their confidence that the event occurred in their childhood, but only when the confidence rating immediately followed the act of unscrambling. This increase in confidence mirrors the "revelation effect" observed in word recognition experiments. In the present article, we analyzed our data using a new signal detection mixture distribution model that does not require the researcher to know the veracity of memory judgments a priori. Our analysis reveals that unscrambling a key word or an unrelated word affects response bias and discriminability in autobiographical memory tests in ways that are very similar to those that have been previously found for word recognition tasks.
How do people decide whether a particular event occurred in the past? One important factor is the manner in which they process the event they are trying to remember. Unscrambling a word (anagram) just prior to making a recognition decision on that word or on an unrelated word increases the belief that the target word was seen before- a phenomenon called the revelation effect (Watkins & Peynircioðlu, 1990; Westerman & Greene, 1998). The revelation effect, which is mostly observed with verbal stimuli, has been extended to childhood autobiographical memory (Bernstein, Whittlesea, & Loftus, 2002). Participants express more confidence that events happened in their childhood if they unscramble a word embedded within descriptions of those events (e.g., "broke a nwidwo playing ball") prior to making the confidence judgment.
Verde and Rotello (2004) have shown that revelation experiments in which the anagram is the same word as the target word (nwi dwo-window) yield different effects than do experiments in which the anagram is unrelated to the target word (eblndre [blender]-window). Using signal detection theory, they demonstrated that the revelation effect for unrelated anagrams is due to increased response bias only (i.e., a general tendency to judge items as "old"), whereas the revelation effect for target-word anagrams is due to both increased response bias and impaired ability to discriminate old and new words as measured by the discriminability parameter d'.
The present work had two goals. First, we wished to test whether there would be a revelation effect in autobiographical memory when an anagram was presented immediately prior to (rather than simultaneously with) the rated life-event item. Doing this would conceptually replicate results found for word recognition tasks. Second, we wished to show that solving anagrams that were related versus those unrelated to life event-items would produce effects on discriminability and response bias similar to those previously found in standard recognition paradigms for anagrams that were identical to rather than unrelated to target items (Verde & Rotello, 2004). However, unlike in Verde and Rotello, we could not use standard signal detection (SD) methods to achieve our second goal, because it is generally unknown which life events depicted in the test really happened to a participant ("true events") and which did not ("false events"). We therefore developed a new SD mixture distribution model that helped us answer our research questions.
Assume that an unknown proportion, p, of items in the autobiographical memory test corresponds to true events from the participants' past. By implication, a proportion (1 2 p) of the test items must then describe false events. In keeping with the tenets of SD theory (see, e.g., Macmillan & Creelman, 1991), assume also that the familiarities of true and false events are independently normally distributed with the means dt and df (dt > df), and the standard deviations σt and σf, respectively. …