PFC and Episodic Retrieval

The contributions of cognitive control to episodic memory are not restricted to the building of memories, but also extend to episodic retrieval––that is, the ability to recollect contextual details surrounding a prior event or to detect stimulus novelty or familiarity. Our efforts to understand the interaction between cognitive control and declarative memory have included a focus on identifying and characterizing the multi-component nature of PFC contributions to retrieval with and without recollection.

In one line of investigation, we explored the neural correlates of source memory, a form of remembering that depends on recollecting a specific contextual detail about a past event. Relative to memory without recollection, source retrieval is thought to entail the recapitulation of neocortical representations present during the encoding of an event, and to differentially depend on strategic engagement of cognitive control. To test these hypotheses, we used fMRI to index neural activation during retrieval with and without source recollection. Our results revealed that (a) source recollection is associated with the recapitulation of representations in perceptual and conceptual processing regions that were engaged during initial encoding of events, (b) such recapitulation effects are observed not only for accurate remembrances, but also are present when we falsely recollect having encountered a novel stimulus, and (c) although PFC processes are differentially engaged during attempts to recollect, strategic engagement of these processes is gated by an initial assessment of whether the memory probe is familiar or novel [Kahn et al., 2004]. Results from a recently completed MEG study suggest that this familiarity-based gating of PFC function may emerge through an interaction with lateral parietal regions [Kahn et al., in prep; for a review of parietal cortex and episodic retrieval, see Wagner et al., 2005]. Collectively, these findings inform neural models of recollection, and address core debates surrounding cognitive theories of remembering.

In a second line of investigation, we have begun to more precisely specify the nature of the PFC control mechanisms engaged during retrieval, focusing in on those that support “pre-retrieval processes” (e.g., representing and elaborating on retrieval cues in working memory) and those that operate “post-retrieval” (e.g., monitoring the products of retrieval as a basis for action) [Wagner, Desmond et al., 1998; Dobbins et al., 2002; Dobbins et al., 2003; Dobbins & Wagner, in press]. To this end, we first identified and characterized separable PFC structures that are recruited depending on whether a retrieval decision requires recollection of contextual details or an assessment of stimulus familiarity [Dobbins et al., 2003]. Our results revealed distinct left lateral PFC and parietal structures that distinguished attempted source recollection from judgments of relative stimulus familiarity, with these retrieval orientation effects being independent of retrieval outcome [for related strategic retrieval effects, see Wagner, Desmond et al., 1998]. Subsequently, we tested the nature of the left PFC processes recruited during source recollection [Dobbins et al., 2002; Dobbins & Wagner, in press]. Using conjunction analyses, we isolated multiple PFC retrieval responses, some that appear correlated with “pre-retrieval” mechanisms and some with “post-retrieval” mechanisms. In the former class, activation in distinct PFC regions was related to semantic analysis/cue specification (left anterior VLPFC), perceptual analysis/cue specification (right VLPFC), and cue maintenance (left posterior VLPFC). In the latter class, effects compatible with post-retrieval monitoring were observed in multiple structures (dorsolateral and frontopolar PFC).

Though many questions remain, and an integrated model delineating how these putative mechanisms interact awaits specification, our investigations constitute initial progress towards specifying how PFC-mediated cognitive control processes support attempts to recollect the past. Our long-term goal is to develop an integrated model of the cascade of neural and cognitive events that give rise to episodic remembering.