Implicit and explicit memory and response bias

Abstract

The influence of response bias on tests of implicit and explicit memory was assessed by using word-stem completion and recognition. During acquisition, subjects were presented either with a list of words unrelated to the task in the retrieval phase or with a list of words whose stems could be completed in only two ways (e.g. " appeal," "appear"). In the word-stem-completion retrieval phase, the subjects, who were given either explicit or implicit memory instructions, were asked to complete these stems. Retrieval of items in implicit and explicit memory tasks was highly correlated (r = .593 and .625, respectively) when the acquisition list had contained the words completing the stems and similarly correlated (r = .667) when the acquisition list had contained no words completing the stems. There was no correlation between recognition and explicit stem completion. In contrast to stem completion, there was no overall correlation between " recognition" in the condition in which the acquisition list contained words to be recognized later and the guessing condition, in which the acquisition list had contained no words presented during the retrieval phase. However, when the items were divided into two groups differing in memorability, a negative correlation appeared between recognition and guessing probability with the more memorable items, and a positive correlation appeared with the less memorable items. This suggests that response bias, rather than the memory the tasks are designed to test, is responsible for the normally reported lack of correlation between recognition and word completion.In a widely used experimental paradigm for studying memory, subjects are presented with a list of words and then are tested, first using a recognition task and then using a cued-recall or word-completion task (for a review, see Nilsson , Law, & Tulving, 1988). A measure of statistical association is then computed between the scores of the two tasks . In the present experiments, a paradigm is used that measures the same essential relationships but that has been modified to enable a simple comparison between the way underlying biases operate on the two tasks between which a measure of association is calculated.It has been proposed (Tulving & Schacter, 1990) that there are two independent types of memory, implicit and explicit, whose presence can be revealed by the use of different tests . The absence of correlation between the results of such different tests has been used to argue for the independence of two memory processes, although the validity of such arguments has been challenged (Hintzman & Hartry, 1990). However, the probabilities of response on the two tests could be influenced by another factor besides the memory processes being tested (e.g., item familiarity). This nonmemory factor or bias could be the major determinant of the pattern of results on one of the tests. If the major determinant of the other test being correlated is a different bias, or a memory factor, a lack of correlation between the tests would not show that...

Violations of the independence assumption in process dissociation.

“…On the other hand, the item-based correlations between all three measures were significantly positive. The large correlation between^loid andv4 new confirms the fact that baseline item characteristics are a dominant source of variance in stem completion (see also, Deutsch, 1992;Hintzman & Hartry, 1990;Ostergaard, 1992). More important, the positive correlation of R o \ a with both/l new and^4 o id represents a violation of the assumption that R and A are independent.…”

“…Hintzman and Hartry (1990) found that priming accounted for just over 5% of the variance in completion rates, whereas a large percentage of variance was attributable to characteristics of the cues in word-fragment completion, which is similar to stem completion (Roediger, Weldon, Stadler, & Riegler, 1992). In general, performance on an indirect test such as stem or fragment completion may only slightly be mediated by priming when participants can successfully complete many stems without previous study (see also Deutsch, 1992; Ostergaard, 1992). Thus, correlations arising from inherent differences among items are particularly likely in a task such as stem completion.…”

Toward a redefinition of implicit memory: Process dissociations following elaborative processing and self-generation.

“…Of course, this line of argument extends to manipulations other than generation or LoP. The general point is that the strategy advocated in the retrieval intentionality criterion—equating cues and varying only instructions across tests—may not produce only differences in retrieval orientation (implicit or explicit) but instead produce only differences in response criterion (Deutsch, 1992; Reingold & Merikle, 1991).…”

“…Under these conditions, an indirect test is assumed to be process pure (i.e., uncontaminated) when it does not respond, or when it responds differentially, to a variable that affects performance on the direct test. Two problems for this method are (a) potential differences in sensitivity or response criteria across the two tests (Deutsch, 1992; Reingold & Merikle, 1991) and (b) the evidence that unconscious processes contribute to performance on direct tests of memory (Jacoby, Toth, & Yonelinas, 1993). These findings suggest that it may not always be appropriate to use performance on a direct test as a benchmark or ideal against which to compare performance on an indirect test.…”