Retrieval orientation for memories encoded in emotional contexts: An ERP study

Highlights

• Retrieval orientation is modulated by the emotionality of the encoding context.

• Valence and arousal have different effects on retrieval orientation.

• Retrieval orientations for positive and negative memories involve nonequivalent neural mechanisms.

Abstract

Retrieval orientation, which is involved in recognition cue processing, optimizes goal-directed memory retrieval. However, whether the emotionality of encoding context affects subsequent retrieval orientation remains unclear. To clarify this, neutral objects were paired with either emotional or neutral background scenes during the study phase. During recognition test, only neutral objects were presented. The ERP analysis on the correctly rejected new items indicated that at least two processes were modulated by the emotionality of memory: 1) the arousal-modulated effect on the right-frontal scalp, and 2) the posterior-distributed effect, which was found to differentiate between memories with positive and negative valence. Furthermore, the magnitude of posterior-distributed effect was correlated with affective rating. The topographical distribution indicated that retrieval orientation for positive memories involves at least partially different neural circuitries from neutral or negative memories. Our results suggest that the emotionality of encoding context affects subsequent retrieval orientation.

Introduction

To reactivate the engram (i.e., records of previous experiences stored in the brain), the retrieval of past episodes requires ecphory—defined as the interaction between retrieval cues and memory traces (Tulving, 1983, Tulving, 1985). Retrieval success depends on how a retrieval cue is processed prior to and during the cue’s interaction with targeted memory traces. The processing of a retrieval cue can be varied in a pre-retrieval operation, referred to as retrieval orientation, that optimizes the compatibility between the cue and memory trace (Rugg & Wilding, 2000). Evidence for retrieval orientation has been furnished in studies that have manipulated various types of mnemonic variables, reflecting the flexibility of retrieval cue processing. However, few studies have examined the modulation of emotion—which has been demonstrated to considerably affect memory (for a review see Hamann, 2001)—on the processing of retrieval cues. Therefore, to fill this research gap, the present study examined how the emotionality of the encoding context affects subsequent retrieval orientation. Specifically, event-related potentials (ERPs) were analyzed in recognition memory tests, where ERPs recorded during the test phase for items that were encoded in emotionally valenced versus neutral contexts were compared.

The neural correlates of retrieval orientation can be revealed by comparing brain activities elicited by the same retrieval cues used to probe different types of memories (Rugg & Wilding, 2000). In a recognition test, corrected rejected new items are considered to serve as optimal retrieval cues for examining the retrieval-orientation effect; these items are optimal because the influence of the successful retrieval of different types of memory is minimal. Previous studies adopting this approach have reported that correct rejection (CR) trials were associated with different ERP patterns when new items were presented with studied items that were of different input modalities (Hornberger et al., 2004, Hornberger et al., 2006, Hornberger et al., 2006), material types (Robb and Rugg, 2002, Woodruff et al., 2006), and encoding tasks (Rugg et al., 2000, Johnson et al., 1997). The differential processing of retrieval cues has been reported to be associated with the enhancement of retrieval accuracy (e.g., Bridger and Mecklinger, 2012, Roberts et al., 2014) and to contribute to the strategic recollection of essential information (e.g., Herron & Rugg, 2003; Dzulkifli and Wilding, 2005).

Although retrieval orientation is useful in the study of memory retrieval, it remains unclear whether and how the processing of retrieval cues varies depending on the emotionality of memory. Emotional memories are more durable, vivid, detailed, and accurate than neutral memories are (Kensinger & Schacter, 2008). Many studies have examined encoding and consolidation processes that differentiate emotional memories from neutral memories (see Dolcos et al., 2017 for a review). However, relatively few studies have focused on the retrieval of emotional memory (Kensinger & Ford, 2020). An important concern in studying emotional memory retrieval is how emotion is manipulated. If emotion is manipulated by employing neutral and emotional materials as items to be remembered, the test items presented in subsequent memory tests will also comprise emotional values; this makes it difficult to discern the effect of emotion on memory encoding from that on retrieval (Maratos and Rugg, 2001, Smith et al., 2004). This problem can be prevented by presenting neutral items in an emotional or neutral encoding context during the study phase and without emotional contexts during the test phase (e.g., Erk et al., 2003, Erk et al., 2005, Maratos et al., 2001, Smith et al., 2004). Previous ERP studies adopting this approach have reported that memories for neutral items encoded in negative contexts were accompanied by greater recollection than items encoded in neutral contexts were (Maratos and Rugg, 2001, Smith et al., 2004, Jaeger et al., 2009 Jaeger and Rugg, 2012; Bowen & Kensinger, 2017). Functional MRI findings have also indicated the recapitulation of sensory and emotional processing during the recollection of memories encoded in emotional contexts (Fenker et al., 2005, Bowen and Kensinger, 2017). A recent ERP study further indicated that the effect of the valence of the encoding context was exerted during the very early stage of the old/new effects of ERPs, suggesting that recapitulation can occur in the early stage of the retrieval process (Bowen, Fields, & Kensinger, 2019). Although these studies have examined the recollection and recapitulation of emotional memories, they did not directly investigate retrieval orientation. Specifically, they could not do so because their findings could not differentiate between pre-retrieval cue processing and retrieval success.

The present study investigated whether different cognitive operations were engaged in the processing of retrieval cues when the encoding contexts of emotionally neutral retrieval targets had different valences. We generated emotional memories by presenting neutral object pictures in encoding contexts (specifically, scenes) that were neutral, positive, or negative in different blocks. The Positive Affect Negative Affect Scale (PANAS, Watson, Clark, & Tellegen, 1988) was administered after each of the three encoding sessions to confirm that the valenced pictures triggered the target emotions during encoding. In the subsequent recognition tasks, only pictures of neutral objects were presented. Typically, the modulation of the encoding context valence on retrieval-orientation is examined by comparing ERPs associated with CR trials in three test phases. However, the ERPs associated with CR trials could also comprise brain activities related to the retrieval mode, a tonic state maintained during retrieval to ensure the stimuli are processed as retrieval cues (Rugg and Wilding, 2000, Tulving, 1983). Moreover, viewing a series of valenced stimuli has been demonstrated to affect subsequent cognitive operations (e.g., Bradley et al., 1996, Smith et al., 2005, Sutton et al., 1997). The ERP differences between CR trials in three test phases could also be due to brain activities associated with processes irrelevant to memory retrieval orientation, such as retrieval mode or emotional states after viewing a series of positive or negative pictures. To exclude this potential confounding factor, we presented blank trials (see Methods section for details) during the recognition test to record brain activities associated with processes irrelevant to retrieval orientation. ERPs associated with the blank trials were subtracted from ERPs recorded in the CR trials of the same test phase to examine the retrieval orientation effect.

An advantage of our design was the ability to differentiate retrieval-orientation effects specific to valence from those related to arousal. A study argued that emotional experience can be evaluated using the quality of subjective feelings (valence) and the intensity of physical arousal (Russell, 1980). Previous studies have shown that emotional valence and arousal processes can be dissociated at the neural level (Gerdes, 2011, Lewis et al., 2006) and may contribute differently to memory performance (Kensinger & Corkin, 2004). However, these studies have focused on the encoding rather than the retrieval stage, and no study has examined the valence and arousal effect on retrieval orientation. Thus, in our study, we matched the arousal levels of positive and negative context pictures (both have a higher arousal level than the neutral context pictures) to make them differ only in terms of the valence. Consequently, any retrieval-orientation effect exhibiting a difference between positive and negative encoding contexts can be considered an indicator of the influence of valence. However, any effect indicating differences between neutral and emotional contexts but not between positive and negative encoding contexts can be considered to indicate the effect of arousal.

Several studies have suggested the possible modulation of emotion on retrieval orientation with ERPs. Maratos & Rugg (2000) reported that the ERP old/new effects were smaller for negative words than for neutral words mainly because more positive-going waveforms were associated with the CRs to new negative items than to new neutral items. Bowen et al. (2019) similarly reported that the valence of the encoding context exerted an effect on the very early part of the ERP old/new effects over the parietal scalp region. In a study examining the effect of emotion on strategic retrieval, Herron (2017) reported that emotional context exerted an effect on an ERP positivity over the right frontal scalp region. Another ERP experiment that used emotionally neutral, pleasant, and unpleasant images (Lavoie & O’Conner, 2013) also demonstrated that in later time windows (1000–1400 ms) unpleasant new images elicited more positive-going waveforms relative to old images. Based on these findings, we speculated that the effect of an emotional encoding context manifests over the parietal and right frontal regions in the early and later parts of the recording epoch, respectively. In addition, memories for emotional materials have been suggested to be modulated by trait anxiety (e.g., Marzi et al., 2014, Nugent and Mineka, 1994). The Trait Anxiety Inventory (STAI-T, Spielberger, Sydeman, Owen, & Marsh, 1999) was therefore administered to investigate whether the modulation of emotion on retrieval orientation is affected by trait anxiety.