Brain and behavioral indices of retrieval mode J.E. Herron and E.L. Wilding * School of Psychology, Cardiff University, Cardiff, CF10 3AT, Wales, UK Received 25 January 2006; revised 16 March 2006; accepted 28 March 2006 Available online 15 May 2006 In recent event-related potential (ERP) studies of episodic retrieval, ERPs have been acquired in tasks where participants have been cued trial-by-trial to prepare either to make episodic or semantic retrieval judgments. ERPs elicited during this preparatory cue period and separated according to retrieval task have diverged at right frontal scalp electrodes, with a relatively greater positivity associated with preparation for episodic rather than for semantic retrieval. Impor- tantly, this pattern of differences has been observed only on Fstay_ trials: those trials where the participant was cued to prepare for the same retrieval task on the previous trial. These findings have provided the basis for the proposal that the ERP modulations index processes that support the adoption or configuration of retrieval mode—a tonic process that can be sustained while recovery of episodic information is required and which facilitates the retrieval process. In these studies, however, the preparatory period on each trial was no more than 2000 ms, raising the possibility that, with more time available, neural correlates of these preparatory processes would not be restricted to stay trials. In this experiment, participants were cued trial-by-trial to complete either an episodic or a semantic retrieval task, and the preparatory period was greater than 4000 ms on the majority of trials. In keeping with previous findings, the ERPs elicited by these two cue types diverged principally on stay trials at right frontal electrode locations, suggesting that time to prepare is not the primary determinant of the onset of task-specific preparatory retrieval processing. In an important addition to previous findings, moreover, the accuracy of episodic memory judgments increased with the number of successive trials of the same task that participants completed, a finding consistent with the view that adopting a retrieval mode successfully can influence the accuracy of episodic memory judgments. D 2006 Elsevier Inc. All rights reserved. Introduction In recent years, event-related imaging methods have been employed in the study of mnemonic processes that facilitate episodic retrieval. One such class of process – retrieval mode – was introduced by Tulving (1983), who proposed that, in order to retrieve information from episodic memory successfully, it was necessary to adopt a cognitive set which ensured that cues (generated either externally or internally) would be processed primarily as probes of episodic memory (Tulving, 1983; Wheeler et al., 1997). Retrieval mode is assumed to be maintained tonically for as long as episodic retrieval is required (Buckner and Tulving, 1995; Rugg and Allan, 2000). It is also thought to remain invariant across different types of episodic memory task. It is difficult to make inferences about retrieval mode via behavioral data alone because mode is assumed to affect all items in episodic memory tasks equivalently (Tulving, 1983; Wilding, 2001). Neuroimaging methods can be employed to circumvent this difficulty, however, as neural activity elicited during retrieval tasks can be measured while people are preparing to retrieve different kinds of information from memory. Event-related potentials (ERPs) are particularly useful in this regard as the properties of the technique mean that it is possible to separate neural activity associated with preparing to retrieve from that associated with episodic retrieval itself. For example, in one study, participants completed two tasks, requiring either an old/new recognition memory judgment or a semantic retrieval judgment (Duzel et al., 1999, 2001). Participants completed short blocks of each task and were cued at the start of each block as to which task to complete. Neural activity elicited by the preparatory cues differed according to whether the cue directed participants to make recognition memory or semantic memory judgments. These differences were maintained for the duration of each test block and were focused over right frontal scalp sites, with ERPs associated with the recognition memory preparatory cue being more positive-going than those associated with the semantic cue. This divergence was interpreted as an electrophysiological correlate of retrieval mode, and these findings are consistent with functional neuroimaging evidence that retrieval mode is associated with activity within right prefrontal cortex (Nyberg et al., 1995; Lepage et al., 2000; Duzel et al., 2001; Grady et al., 2001). In other studies in which ERPs have been acquired time-locked to preparatory retrieval cues, participants have been cued trial-by- trial as to which task to complete. Morcom and Rugg (2002) employed the same tasks used by Duzel et al. (1999) and again observed more positive-going ERPs at anterior scalp locations in response to the preparatory cue signaling episodic retrieval rather than the cue signaling semantic retrieval. Importantly, however, Morcom and Rugg (2002) observed cue-related differences only 1053-8119/$ - see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2006.03.046 * Corresponding author. Fax: +44 (0) 29 20874858. E-mail address: wildinge@cardiff.ac.uk (E.L. Wilding). Available online on ScienceDirect (www.sciencedirect.com). www.elsevier.com/locate/ynimg NeuroImage 32 (2006) 863 – 870