Inducing amnesia through systemic suppression

Abstract

Hippocampal damage profoundly disrupts the ability to store new memories of life events. Amnesic windows might also occur in healthy people due to disturbed hippocampal function arising during mental processes that systemically reduce hippocampal activity. Intentionally suppressing memory retrieval (retrieval stopping) reduces hippocampal activity via control mechanisms mediated by the lateral prefrontal cortex. Here we show that when people suppress retrieval given a reminder of an unwanted memory, they are considerably more likely to forget unrelated experiences from periods surrounding suppression. This amnesic shadow follows a dose-response function, becomes more pronounced after practice suppressing retrieval, exhibits characteristics indicating disturbed hippocampal function, and is predicted by reduced hippocampal activity. These findings indicate that stopping retrieval engages a suppression mechanism that broadly compromises hippocampal processes and that hippocampal stabilization processes can be interrupted strategically. Cognitively triggered amnesia constitutes an unrecognized forgetting process that may account for otherwise unexplained memory lapses following trauma.

Figure 1. Hippocampal modulation (HM) paradigm and predictions.

After studying word pairs (phase 1), participants perform trials requiring that they either retrieve (green) or suppress (red) retrieval of the second word in each pair, given the first word. Inserted between these trials are novel ‘bystander' pictures that participants encode by silently answering the question, ‘Why is the pictured object in this location?' Even/odd buffer judgments performed before and after bystanders match the immediate task context across retrieve and suppress trials. The line graph illustrates the predicted efficacy of hippocampus-dependent memory processes (‘mnemonic processing efficiency') during the above events. Of interest is whether surrounding bystanders with suppression trials affects later memory for bystander pictures, causing an ‘amnesic shadow.' This is assessed in phase 3, in which the participant must recall the associated object for each context scene.

Figure 2. Amnesic shadow in cued recall.

(a) Immediate cued-recall accuracy for bystanders by number of adjacent suppression epochs. Difference between the left- (peach colour) and right-most (dark orange) bars reveals an amnesic shadow (F-test). (b) Shadow observed after 24 h delay. (c) Experiment 3's two No-Think strategies. (d) Direct suppression, not thought substitution, caused a shadow. (e) Experiment 4 replaced No-Think trials with a difficult ‘Think-Harder' task. (f) No shadow was observed in experiment 4 (left subpanel), despite a significant difficulty disparity across types of surrounding retrieval epochs (right subpanel, paired t-test). Error bars reflect within-participant s.e.m. *P<0.05; ***P<0.001.

Figure 3. Nature and growth of the amnesic shadow.

(a) Cued-recall accuracy in experiment 5 for the zero- and two-dose conditions compared with baseline (purple) with F-tests. Whereas the above-baseline benefit was not significant, below-baseline disruption was. (b) The amnesic shadow (negative deflection of the red line away from the x axis) grew across quartiles, consistent with a practice effect. (c) Unlike the growing shadow apparent across immediate cued-recall studies (red line), thought substitution (experiment 3), Think-Harder (experiment 4) and baseline (experiment 5) studies showed no such growth with practice (purple line). Error bars reflect s.e.m. *P<0.05; **P<0.01.

Figure 4. Amnesic shadow in recognition.

(a) Memory for context, not items, is impaired for words (F-test). (b) Replication with pictures. (c) Context amnesia in experiment 7. (d) Right-most inset bar illustrates mean (error bar, s.e.m.) bilateral hippocampal modulation (blue anatomical region of interest seen in adjoining glass brain) during suppression, relative to retrieval in arbitrary units (a.u.). Modulation (two-tailed t-tests) in left and right hippocampi also is plotted separately for exploratory purposes (with Bonferroni correction for multiple corrections across the two hemispheres). Across participants, modulation predicted context amnesia (outlier-skipped Pearson bootstrap 95% confidence interval (CI): (0.13, 0.79)). Q (e) Suppression-related frontal engagement (yellow No-Think>Think functional mask) predicted hippocampal modulation (bootstrap CI: (0.03, 0.87)). Robust correlation removed bivariate outliers from relevant analysis/plot. Error bars for behaviour (a–c) reflect within-participant s.e.m.; red bands, 95% CI; ‡significant correlation by bootstrap test; *P<0.05; **P<0.01; +P≤0.10.