Neural Correlates of Direct and Indirect Suppression of Autobiographical Memories

Abstract

Research indicates that there are two possible mechanisms by which particular target memories can be intentionally forgotten. Direct suppression, which involves the suppression of the unwanted memory directly, and is dependent on a fronto-hippocampal modulatory process, and, memory substitution, which includes directing one's attention to an alternative memory in order to prevent the unwanted memory from coming to mind, and involves engaging the caudal prefrontal cortex (cPFC) and the mid-ventrolateral prefrontal cortex (VLPFC) regions. Research to date, however, has investigated the neural basis of memory suppression of relatively simple information. The aim of the current study was to use fMRI to identify the neural mechanisms associated with the suppression of autobiographical memories. In the present study, 22 participants generated memories in response to a series of cue words. In a second session, participants learnt these cue-memory pairings, and were subsequently presented with a cue word and asked either to recall (think) or to suppress (no-think) the associated memory, or to think of an alternative memory in order to suppress the original memory (memory-substitution). Our findings demonstrated successful forgetting effects in the no-think and memory substitution conditions. Although we found no activation in the dorsolateral prefrontal cortex, there was reduced hippocampal activation during direct suppression. In the memory substitution condition, however, we failed to find increased activation in the cPFC and VLPFC regions. Our findings suggest that the suppression of autobiographical memories may rely on different neural mechanisms to those established for other types of material in memory.

Keywords: autobiographical memories; direct suppression; fMRI; memory retrieval; think/no-think.

FIGURE 1

(A) Experimental procedure for session 1. (B) Experimental procedure for the ATNT task in session 2.

FIGURE 2

Mean percentage of memories recalled as a function of instruction type. Error bars represent ±1 standard errors of the mean (SEM).

FIGURE 3

Regions of activation in the think/memory substitution vs. no-think1 and 2 contrasts. (A) Regions demonstrating significant activation in the think vs. no-think1 contrast. (B) Regions demonstrating significant activation in the memory substitution vs. no-think2 contrast. Images in both panels are thresholded at p < 0.001 with clusters comprising a minimum of five contiguous voxels. In both panels, the sagittal slice is marked with an L indicating the side of the left hemisphere and is rendered where x = -5. The coronal slice is rendered at y = -28.

FIGURE 4

Four ROIs and their beta amplitude responses. Illustrations of the ROIs in red (left) and their beta amplitude responses to think, memory substitution, no-think1 and no-think2 (right) are shown for (A) left hippocampus, derived from a cluster within the think vs. threshold contrast thresholded at p < 0.0001, and 5 mm radius spheres centered on (B) right DLPFC [32, 38, 26], (C) left caudal PFC (cPFC) [-52, 9, 24] and (D) left VLPFC [-50, 25, 14] (BA 45). Error bars represent ±1 SEM.

FIGURE 5

Regions which show a no-think-couple/think-decouple relationship with left posterior hippocampus. Results of a whole-brain PPI analysis using a 4 mm sphere centered on MNI coordinates [-21, -28, -8]. Regions highlighted couple with the seed region during the no-think condition and decouple during the think condition. Images are thresholded at p < 0.001 with clusters comprising a minimum of five contiguous voxels.