Conceptual priming in fragment completion.

Abstract

Some researchers have claimed that fragment completion tasks are dependent primarily on datadriven processing and are insensitive to conceptually driven processing. In this article we present four experiments demonstrating that conceptually driven processing affects fragment completion by showing that under appropriate conditions, studied words can facilitate identification of their picture and word fragments. We examine two theoretical explanations of this effect. First, we consider the possibility that subjects explicitly retrieve episodic representations in fragment completion. Analyses of correlations between priming and recall performance across items and subjects do not support this explanation. The alternative explanation is that there are two separate conceptual representations in memory. The first is assumed to mediate conceptual priming in fragment completion; the second is assumed to mediate free recall performance. A final experiment supports this view by demonstrating that even when differences between experimental conditions are made to disappear in fragment completion, they remain in free recall. Further applications of the notion of two semantic representations are discussed in the General Discussion.Discovering the relation between implicit and explicit memory tests has become a central issue in memory theory (Richardson-Klavehn & Bjork, 1988;Schacter, 1987). Operationally, explicit memory tests, such as recall and recognition, can be distinguished from implicit memory tests, such as fragment completion and perceptual identification, because instructions on an explicit memory tasks refer to an episode in the subject's past, whereas instructions on an implicit memory task do not. Explaining the relation between implicit and explicit memory has assumed a central role in memory theory because of the many demonstrations indicating that performance on the two types of memory tests is dissociated.We will begin this article by reviewing some of these dissociations. Following this we will describe Roediger's (Roediger & Blaxton, 1987;Weldon & Roediger, 1987; Weldon, Roediger, & Chalks, 1989) influential theoretical approach to the explanation of the dissociations between implicit and explicit memory. We will then present several experiments that demonstrate, contrary to the claims of Roediger and his colleagues, that conceptual or semantic information can play a role in fragment completion. The latter sections of the article are then devoted to explaining this novel result.

Repetition priming with Japanese Kana scripts in word-fragment completion

“…As a result, generating a target word in response to its definitionproduced reliableprimingeffectsin subsequent perceptual identification and, more importantly, the size of priming induced by prior generation did not differ significantly from the size of priming by prior reading, thus providing a sharp contrast with the results of previous studies in the English literature (Jacoby, 1983;Roediger & Blaxton, 1987b;Winnick & Daniel, 1970). Moreover, several studiesin the Englishmemory literature have also revealed that generating a target word gives rise to a reliable size of priming on later implicit memory tests, although experimental procedures and materials have differed from each other (Gardiner, 1988;Hirshman, Snodgrass, Mindes, & Feenan, 1990;Masson& MacLeod, in press;Schwartz, 1989;Toth & Hunt, 1990). Thus, it becomesa matter of concern whether the effects of prior generation on performance in implicit memory tests (referred to as conceptual priming) are substantial.…”

Rapid induction of false memory for pictures

“…The absence of priming for pictures of imagined objects on a perceptual identification task is in line with the Michelon and Koenig (2002) finding of no perceptual priming following imagination, the Hicks and Starns (2005) finding of no priming for critical lures in the DRM paradigm on a verbal perceptual identification task, and the studies that found no priming for critical lures on a lexical decision task (McKone, 2004;Zeelenberg & Pecher, 2002). The findings are also in line with studies which have found no cross-form priming from words to pictures (Hirshman et al, 1990;Warren & Morton, 1982;Weldon et al, 1995). However, results of Experiment 6 contradict McDermott and Roediger (1994) who 92 found priming in a perceptual task which involved the identification of pictures from fragments presented for 100/200ms following imagination of these pictures from their word labels.…”

“…If priming is a function of the match between study and test, then a change in form (in this case, from verbal to pictorial) should reduce priming on perceptual implicit tests. While most studies have found that a change from verbal stimuli in the study phase to pictorial stimuli in the test phase eliminates perceptual priming completely (Hirshman, Snodgrass, Mindes, & Feenan, 1990;Scarborough, Gerard, & Cortese, 1979;Warren & Morton, 1982;Weldon et al, 1995), some other studies (Durso & Johnson, 1979;Park & Gabrieli, 1995;Srinivas, 1993) have found significant cross-form priming, suggesting that certain ostensibly perceptual priming tasks can include a nonperceptual component. In order to test whether the imagination manipulation used in the current procedure creates false memories that are identifiable on an indirect perceptual priming test, it was first necessary to select a task which was, as far as possible, data-driven and unsusceptible to nonperceptual processes.…”

The picture superiority effect in a cross-modality recognition task

“…Still, examples of crossmodal priming have repeatedly been found (Weldon, 1991;Weldon & Roediger, 1987;Weldon et al, 1989), although they have typically been much smaller than intramodal effects. The cross-modal effects have aroused considerable interest, and further study has shown that they can be as great as the intramodal effects under some circumstances (Brown et al, 1991) and that they may affect such seemingly data-driven aspects of processing as perceptual information acquisition (Reinitz, Wright, & Loftus, 1989) and fragment completion (Hirshman, Snodgrass, Mindes, & Feenan, 1990).…”