ABSTRACT
Aversive memories have the potential to impair one’s psychological well-being. It is desirable to reduce the anguish over such memories, as well as the chance that they will be retrieved. In two experiments, we investigated whether retrieval stopping reduces the distress elicited by negative memories retrieved from cues and how the effects of retrieval stopping are modulated by mental disorders, such as depression and anxiety. Participants engaged in retrieval stopping of aversive scene memories without any diversionary thoughts (direct suppression, Experiment 1) or with diversionary positive thoughts (thought substitution, Experiment 2). Direct suppression reduced arousal elicited by the retrieval of aversive memories, while thought substitution did not only reduce arousal but also increased positive valence. Self-reported anxious/depressive symptoms negatively modulated the effects of direct suppression. For no or mild anxious/depressed individuals, direct suppression alleviated negative valence and high arousal when retrieving aversive memories. The negative relationship was not observed between the severity of the symptoms and the effect of thought substitution. These findings suggest that both retrieval stopping strategies can reduce distress from aversive memories.
Acknowledgements
This research was supported by Grant-in-Aid for JSPS Fellows (18J20948) to S.N.; and the Global Education Office, Kyoto University. We thank Miho Nishiguchi, Chifumi Sakata, Tomoya Watanabe, Takumi Aoki, Naru Shimazu, Renya Mizuno, Yume Kato, Mikihiro Shinohara, Katsuaki Fujii, and Rina Miyahara for their assistance in data collection.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Author contributions
S.N. and S.S. contributed to the study design. S.N. performed the experiments and analyzed the data. S.N. drafted the manuscript and both authors revised the manuscript. Both authors approved the final manuscript for submission.
Open practices statements
The data and R scripts for all experiments are available at Open Science Framework (https://osf.io/nep93/) and neither experiment was preregistered.
Notes
1 It should be noted that “retrieval stopping” is not always used as an umbrella term for direct suppression and thought substitution. Some recent studies seem to refer to retrieval stopping and direct suppression as equivalent (e.g., Anderson & Hulbert, Citation2021). Furthermore, there is no term covering the two strategies that TNT researchers have agreed upon. Nevertheless, we use “retrieval stopping” here as an umbrella term, following Bergström et al. (Citation2009), who were the first to compare the two strategies and referred to them as “strategies for retrieval stopping” (p. 728).
2 The magnitude of effect sizes that would likely be obtained for the retrieval stopping effect on changes in affective ratings was unclear because few previous TNT studies have examined this effect. However, we expected a conventional medium effect size, based on the assumption that retrieval stopping would produce an effect size on changes in affective ratings similar to the size of forgetting in the TNT paradigm (dz > 0.7, in our previous studies; Nishiyama & Saito, Citation2021). dz represents the standardized mean difference effect size for within-subjects designs (Lakens, Citation2013).
3 Some readers might be conservative in using one-sided t tests. The reason why we employed one-sided t tests was that we had clear predictions for the directions of the effect of retrieval stopping as in the main text, and that one-sided t tests can reflect more clearly the prediction as “the mean difference is greater/less than zero” compared to two-sided t tests in which the alternative hypothesis is that “the difference is not zero”. According to the decision, the power analysis for the sample size calculation was also performed assuming a one-sided t test. To keep consistency between the power analysis and actual hypothesis testing, the analysis should be one-sided t test in the current study.
4 This trend was numerically observed in Experiment 1, but not in Experiment 2. As the number of non-recognized items was small, we did not conduct statistical tests.
5 According to a reviewer’s suggestion, we conducted repeated measures ANCOVAs on Δvalence and Δarousal with each questionnaire score as a covariate, using JASP (JASP team, Citation2021). However, most of the results of the ANCOVAs for the data of Experiment 1 showed interactions between the main independent variable (No-Think vs. Baseline) and a covariate, indicating that homogeneity of regression slopes, an assumption of ANCOVA, was violated. Therefore, although such violations were not found in ANCOVAs on the data of Experiment 2, we decided to report correlation and regression analyses in the two experiments to maintain consistency of analytic approaches. The ANCOVA results may be found in JASP files (ancova_e1.jasp and ancova_e2.jasp) on the OSF repository.
6 It should be noted that this speculation about more intrusions in anxious/depressed individuals was not supported by an additional correlation analysis between intrusion proportion and questionnaire scores, in which weak correlations were found (rs around -0.1 to 0.1, see Figure S4).