Suppressing Unwanted Memories Reduces Their Unintended Influences

Abstract

The ability to control unwanted memories is critical for maintaining cognitive function and mental health. Prior research has shown that suppressing the retrieval of unwanted memories impairs their retention, as measured using intentional (direct) memory tests. Here, we review emerging evidence revealing that retrieval suppression can also reduce the unintended influence of suppressed traces. In particular, retrieval suppression (a) gradually diminishes the tendency for memories to intrude into awareness and (b) reduces memories' unintended expressions on indirect memory tests. We present a neural account in which, during suppression, retrieval cues elicit hippocampally triggered neocortical activity that briefly reinstates features of the original event, which, in turn, are suppressed by targeted neocortical and hippocampal inhibition. This reactivation-dependent reinstatement principle could provide a broad mechanism by which suppressing retrieval of intrusive memories limits their indirect influences.

Keywords: direct/indirect memory tests; explicit/implicit memory; retrieval suppression; suppression-induced forgetting.

Fig. 1.

Procedural overview and results for a think/no-think task (TNT; a) and assessment of involuntary TNT intrusions (b). In the TNT paradigm (a), participants first learn cue-target pairs during the encoding session. During the TNT session, participants are repeatedly presented with the original cue words in either green (“think”) or red (“no-think”) font colors and are asked to think or to not think about the associated target memories, respectively. In a subsequent cued-recall session, participants are prompted to recall each target that was paired with the original cue word. Results show that repeatedly suppressing the no-think items (approximately 10–16 times) reduces the likelihood these memories can be recalled (Anderson & Green, 2001). This basic paradigm has been extended to investigate the suppression of different types of materials, and the consequences of suppression have been assessed with a variety of tests. On each trial of one study assessing involuntary intrusions during TNT sessions (Levy & Anderson, 2012), participants were asked to report how often they thought of the associated targets upon seeing think and no-think reminders (b). Involuntary intrusions on no-think trials, indicated by ratings of 2 or 3, declined with repeated suppression.

Fig. 2.

Results from Hu, Bergström, Bodenhausen, and Rosenfeld (2015) revealing the effects of suppressing unwanted autobiographical memories. “Guilty” participants enacted a lab crime in which they took a ring from a professor’s mailbox, whereas “innocent” participants wrote their initials on a board (a). Event-related potential (ERP) difference waves (ERP for crime-relevant stimulus—“ring”—minus the average ERP for crime-irrelevant stimuli—e.g., “wallet,” “bracelet”) revealed effects of retrieval suppression on autobiographical memory (b). A classic guilty-knowledge effect was evident among guilty participants without suppression instructions (guilty-standard group), as shown by enhanced retrieval-related ERP positivity during the 300- to 800-ms poststimulus window (for a recent review, see Rosenfeld, Hu, Labkovsky, Meixner, & Winograd, 2013). However, retrieval suppression largely attenuated this ERP positivity while enhancing the subsequent late posterior negativity (800–1,300 ms). Thus, individual guilty-suppression participants could be accurately detected when both ERP components were combined in a peak-to-peak manner. In an autobiographical Implicit Association Test (aIAT), compared to guilty-standard participants, guilty-suppression participants showed a significantly weaker implicit expression of their autobiographical memory (c). D scores reflected the strength of automatic activation of criminal memories and its unintentional influence on participants’ behavior (for rationales behind the aIAT and D scores, see Sartori, Agosta, Zogmaister, Ferrara, & Castiello, 2008). Error bars indicate 95% confidence intervals.

Fig. 3.

A schematic illustrating parallel, targeted inhibition of hippocampal, amygdala, and neocortical traces exerted by the prefrontal cortex during retrieval suppression. During retrieval suppression, sensory inputs from no-think reminders feed into the hippocampus (blue region), where they elicit pattern completion through reentrant connections to the amygdala (pink region), anterior and posterior parahippocampal gyrus (dark and light green regions, respectively), and visual cortex (black and white voxel grid). Completed patterns, symbolized here by red reactivated voxels in the visual cortex, reinstate neural activity that contributes to episodic experience (i.e., involuntary yet conscious intrusion) and interfere with goal-directed suppression. Such intrusions may trigger inhibitory control mediated by the right middle frontal gyrus (rMFG) to target both hippocampal and reactivated sites, gradually disrupting the corresponding neural/memory representations and impairing both intentional retrieval and unintentional memory expressions.

Fig. 4.

Results from Gagnepain, Henson, and Anderson (2014) showing that suppressing perceptual memories reduced subsequent perceptual priming on both behavioral and neural measures. Suppression recruited the right middle frontal gyrus (a) to down-regulate the left fusiform gyrus (b), as established via effective connectivity analyses. On a perceptual-identification test conducted after the think/no-think phase, reaction times revealed impaired behavioral priming effects for no-think trials compared to think and baseline trials (c). Results from fMRI scans during the final perceptual-identification task revealed impaired neural repetition-priming effects for no-think items (d; left), particularly when the right middle frontal gyrus had effectively down-regulated the left fusiform gyrus during the earlier think/no-think phase (d; right). Reprinted from “Neural Mechanisms of Motivated Forgetting,” by M. C. Anderson and S. Hanslmayr, 2014, Trends in Cognitive Science, 18, p. 288. Copyright 2014 by Elsevier.