In manipulations of stimulus strength between lists, a more lenient signal detection criterion is more frequently applied to a weak than to a strong stimulus class. However, with randomly intermixed weak and strong test probes, such a criterion shift often does not result. A procedure that has yielded delay-based within-list criterion shifts was applied to strength manipulations in recognition memory for categorized word lists. When participants made semantic ratings about each stimulus word, strength-based criterion shifts emerged regardless of whether words from pairs of categories were studied in separate blocks (Experiment 1) or in intermixed blocks (Experiment 2). In Experiment 3, the criterion shift persisted under the semantic-rating study task, but not under rote memorization. These findings suggest that continually adjusting the recognition decision criterion is cognitively feasible. They provide a technique for manipulating the criterion shift, and they identify competing theoretical accounts of these effects.
The ability to recognize previously encountered stimuli is a fundamental index of proper memory function. For the past half century, signal detection theory (SDT; Green & Swets, 1966) has been widely embraced as contributing to a model of recognition (Banks, 1970; Parks, 1966). According to this perspective, people evaluate studied stimuli (targets) and unstudied ones (lures) on a scale of strength. Probes are accepted if they exceed a specific degree of strength, which is treated as the decision criterion. Criteria that demand only modest probe strength may be thought of as lenient, and those that demand great strength as strict.
There is considerable interest in the variables that affect the placement of the SDT criterion. People readily comply with instructions to adopt lenient or strict criteria (Estes & Maddox, 1995; Verde & Rotello, 2007). The composition of the test set is important: Recognition lures that are closely related in meaning to the targets demand a more strict criterion than do unrelated lures (Benjamin & Bawa, 2004; S. Brown & Steyvers, 2005; Reder, 1987). Criterion placement becomes systematically more lenient as test delay increases (Kintsch, Welsch, Schmalhofer, & Zimny, 1990; Singer, Gagnon, & Richards, 2002; Singer & Wixted, 2006). People can use feedback about response accuracy in a recognition test to adjust their decision criterion (Rhodes & Jacoby, 2007; Verde & Rotello, 2007). The application of different criteria as a function of such variables is called a criterion shift (e.g., J. Brown, Lewis, & Monk, 1977; Hirshman, 1995).
Also implicated in criterion shifts is the strength of stimulus encoding. The relationship, however, is a complex one. When strong and weak items are studied and tested in separate lists (between-list designs), criterion shifts are regularly detected (e.g., Hirshman, 1995; Stretch & Wixted, 1998b, Experiment 1). In contrast, when strongly and weakly encoded items, whether studied together or not, are intermixed in the test lists (within-list designs), a uniform criterion is often applied to all items (e.g., Stretch & Wixted, 1998b, Experiments 3-5; Verde & Rotello, 2007).
This study scrutinized a discrepancy in which withinlist criterion shifts have been detected under delay manipulations (Singer et al., 2002; Singer & Wixted, 2006) but not under strength manipulations (Morrell, Gaitan, & Wixted, 2002; Stretch & Wixted, 1998b). The discrepancy is puzzling: Because immediately tested items ought to be represented more strongly than those tested at a delay, the impact of delay ought to parallel that of strength manipulations. Accordingly, the present strategy was to determine whether the materials and study procedures of Singer and Wixted would yield criterion shifts in the context of repetition-based strengthening. As was noted earlier, repetition has frequently not promoted within-list criterion shifts (Stretch & Wixted, 1998b). …