There has been a great deal of interest in implicit and explicit memory stimulated in particular by the finding of several dissociations between these two hypothesized memory systems. Whereas in explicit memory tests subjects are instructed to retrieve intentionally previously studied items, in implicit tests the study episode is not mentioned and retention is measured indirectly as performance facilitation on a seemingly unrelated task. This facilitation, called priming effect, is an index of implicit memory.

This classification was further elaborated by Roediger et al. (1989) in the framework of their processing view of memory. The main assumptions of their approach can be summarized as follows. First, performance on memory tests is improved as long as the cognitive operations involved in the test recapitulate or overlap those engaged during learning. Second, implicit memory tests usually require different mental processes than do explicit tests and consequently they differentially benefit from these processes. Third, the most commonly used implicit memory tests draw primarily on perceptual or data-driven processes used in word recognition and object recognition. Such tests are referred to as perceptual implicit tests. One common test is the word-fragment completion, in which subjects are asked to find the missing letters of a word from a fragment such as e_ _ p_ _ n_. Generally subjects are more likely to find the solution if they have been presented with the word “elephant” in a previous stage. Fourth, most explicit memory tests that involve recall and recognition draw largely on conceptual processing (elaboration, organization, meaningful processing and the like), and can therefore be classified as conceptual explicit tests. Fifth, the perceptual/conceptual contrast is not coextensive with the implicit/explicit distinction: conceptual implicit tests and perceptual explicit tests can be implemented. In the case of conceptual implicit tests, subjects are given a cue that is meaningfully related to the target item that was studied beforehand. In the word association test, they are instructed to freely associate to the cue (e.g. tusk for elephant). In the category association test, they are asked to produce members of the relevant category (e.g., animals). In the case of perceptual explicit tests, subjects are instructed to retrieve encoded information with the help of some perceptual display similar to the target. A word-fragment completion test stressing on intentional remembering (word-fragment cued recall) may be considered as a perceptual explicit test.

In sum, the implicit/explicit and perceptual/conceptual classifications can be combined to yield four kinds of tests: implicit perceptual tests (e.g. word-fragment completion); implicit conceptual tests (e.g., category generation or free association); explicit perceptual tests (e.g., word-fragment cued recall); and explicit conceptual tests (e.g., free recall or semantic cued recall). It is interesting to note that using the same test cues, one can develop an implicit or an explicit test by manipulating the critical instructions used to implement the tests. Thus, using word-fragments as test cues, we can ask subjects to retrieve material from the study episode (perceptual explicit test) or simply to complete spaces with letters in order to produce a word (perceptual implicit test). Moreover, using associates or categories, one can instruct subjects to retrieve material from the study episode (conceptual explicit tests), to produce words meaningfully related to the target, or to produce members of a relevant category (conceptual implicit tests). Schacter et al. (1989) have referred to the retrieval intentionality criterion as the distinguishing feature between explicit and implicit tests.

The retrieval intentionality criterion was implemented by several researchers for the purpose of studying relations between explicit and implicit memory. All of them were looking for variables to dissociate implicit and explicit tests. A potentially interesting variable in this respect is imagery. Pictures are generally better remembered than their corresponding verbal representations on conceptual explicit memory tests such as free recall and recognition. Pictures are more memorable than words because they engage more of the elaborative encoding processes that are required by conceptual explicit memory tests. When the test cues consist of word-stems or word-fragments (i.e. perceptual tests), studied words produce more priming than did studied pictures. However, a reverse pattern of results is obtained with these variables under explicit test instructions when the same word-fragments or word-stems are used as cues. In the transfer-appropriate framework, it is assumed that priming depends on the degree to which the perceptual and conceptual operations performed during testing match those performed during study. Because the surface features of the words studied are similar to those of the test items, the perceptual analyses of both types of stimuli are similar, leading to stronger priming.