Interactions among Forms of Memory

Extensive neuropsychological, electrophysiological, behavioral, and neuroimaging data indicate that memory is not a unitary system, but rather consists of multiple forms of learning and remembering that differ in their functional characteristics and neuroanatomic substrate. Memory research over that past two decades has placed an emphasis on identifying and dissociating memory systems, with less attention directed towards understanding when and how these different systems interact. One emphasis of our research is on this latter issue. That is, what is the nature of the “cross talk” between different forms of memory?

Broadly, we have begun exploring possible memory systems interactions along three dimensions.

(1) We tested dual-process theories that hypothesize that one point of contact between declarative and nondeclarative memory is during recognition performance. From this perspective, familiarity-based recognition is thought to reflect the operation of perceptual priming (a form of nondeclarative memory). Contrary to this hypothesis, we demonstrated that perceptual priming and familiarity-based recognition are functionally and anatomically dissociable, indicating that this form of nondeclarative memory does not contribute appreciably to explicit recognition [Wagner, Gabrieli, & Verfaellie, 1997; Wagner, Stebbins et al., 1998; for a discussion see Wagner & Gabrieli, 1998].

(2) Although priming does not support explicit recognition, we have become interested in whether priming negatively impacts episodic encoding under some circumstances. This hypothesis stems from the synthesis of three findings. First, the magnitude of left PFC activity during word encoding correlates with subsequent recognition memory [e.g., Wagner, Schacter et al., 1998; Kirchhoff, Wagner et al., 2000; Davachi, Maril, & Wagner, 2001 ; Clark & Wagner, 2003; for reviews see Wagner, Koutstaal, & Schacter, 1999; Paller & Wagner, 2002]. Second, prior encounter with a word results in decreased activation in these same left PFC regions during reencounter of the word [e.g., Wagner, Desmond et al., 1997; Wagner, Koutstaal et al., 2000; Kirchhoff, Wagner et al., 2000]. Finally, studies of amnesic patients indicate that this repetition-induced decrease in left PFC activation reflects the operation of nondeclarative memory processes (i.e., priming). Taken together, these three observations suggest that priming may impair episodic encoding. We conducted an initial test of this hypothesis, observing a negative correlation between neural priming during re-encoding and the impact of this second encoding trial on later recognition memory [Wagner, Maril, & Schacter, 2000]. We are presently exploring the generality of this putative memory systems interaction, including a direct test of our hypothesis that priming hinders episodic encoding under some circumstances by reducing encoding variability.

(3) Our focus on the role of cognitive control in episodic encoding also emphasizes the relation between working memory mechanisms and episodic learning. The behavioral literature has been interpreted as indicating that the extent to which a stimulus is maintained in working memory is unrelated to whether the stimulus will be later remembered or forgotten, though some behavioral evidence suggests that rote rehearsal may impact later item recognition. Directly testing whether rehearsal impacts episodic encoding, we obtained fMRI evidence linking working memory maintenance and episodic encoding [Davachi, Maril, & Wagner, 2001; see also, Clark & Wagner, 2003]. Specifically, we observed that subsequent item recognition is superior when rote rehearsal is accompanied by marked activation in brain regions––PFC, parietal, cerebellar, and supplementary motor cortices––known to subserve verbal working memory.