Directed forgetting of emotionally toned items and mental health: a meta-analytic review

ABSTRACT

Item- and list-method directed forgetting paradigms have been used to study forgetting of emotionally toned items in clinical and control group populations for several decades. Meta-analysis of item-method studies found that clinical populations retained more remember- than forget-cued items of each valence. These effects were comparable to that shown by control populations for positive and negative items, but less than that shown by controls on neutral items. Encoding deficits may underlie clinical populations’ item-method directed forgetting since those populations retained fewer remember-cued items of each valence compared to control populations. Moderator analysis indicated larger effect size variability for some clinical populations (e.g., anxiety disorders) than other populations (e.g., PTSD, schizophrenia). Meta-analysis of list-method directed forgetting among clinical populations revealed only List 1 forgetting or costs for neutral items; i.e., better memory for to-be-remembered than forgotten List 1 neutral items, but no List 2 enhancements or benefits; i.e., better memory for List 2 items among those told to forget than remember List 1 items, for any item valence. Control populations showed costs and benefits for all item valences. Results from both paradigms are discussed in terms of clinical-control population differences in executive processes. Limitations of the meta-analyses and suggestions for future research are presented.

KEYWORDS:

• Memory
• directed forgetting
• emotionally toned items
• psychopathologies
• mental health
• meta-analysis

Acknowledgements

Reprints requests and any questions about findings in this article should be sent to [email protected]. We would like to thank the reviewers for their comments and suggestions on earlier drafts of the manuscript. We would also like to thank reference librarians Rainer Schira and Marion Ramage for their assistance in conducting the literature search, and Sylvia Henry for her patience and assistance in helping us construct the tables found in this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The only exception to the inclusion criteria were list-method studies on Alzheimer’s-Dementia. These studies were used because they used full list method procedure, and as far as we are aware, there were no other list-method studies using Alzheimer’s patients.

2 Items were coded as positive or neutral if they were so labelled by study authors. For those studies that used negative items, 38% used diagnostic-relevant negative items; i.e., negative items believed to tap negative components of the diagnostic category. A majority (53%) used negative items derived from existing word lists (e.g., Francis et al., 1982), and 9% of studies presented both diagnostic-relevant and negative items at study. While there is little discussion of the effects of the two types of negative items on information processing, no significant differences in memory were found in those studies that presented both types of negative items at study. For ease presentation (except where noted), we refer item valences as positive, negative or neutral items.

3 In list-method studies, it is necessary to examine output order because the magnitude of benefits can be influenced by whether participants retrieve List 1 or List 2 first (Golding & Gottlob, 2005).

4 The correlation between remember- and forget-cued items used here is an estimate. The correlation could vary widely across studies, between clinical populations, and between item valences. It is unknown whether the correlation used here over or underestimates effect sizes. We want to thank an anonymous reviewer for bringing this issue to our attention. Future studies on clinical and control populations should report correlations between remember- and forget-cued items.

5 There are on-going concerns about whether the Q-test and I² reflect effect size heterogeneity (Borenstein et al., 2017; Borenstein, 2019; Borenstein, 2020). Borenstein et al. (2017) argue that the PI addresses the question that is intended when asking about effect size heterogeneity.

6 At the suggestion of a reviewer, the study search was redone including the term “intentional forgetting”. This search did not yield any additional clinical-control group directed forgetting studies.

7 T = 0.000 means that all studies share a common effect size.

8 Cloitre et al. (1996), Conway and Fthenaki (2003), McClure (2005), Tudorache et al. (2020), Woodward (2004) and Wingenfeld et al. (2013) studies are excluded in these models due to insufficient details with which to compute effect size estimates.

9 The nature of the task used at test may be spurious, only Noel et al. (2012) used a recognition task.

10 Meta-analyses on remember-cued items omitting studies used to assess the avoidant-encoding hypothesis replicated findings reported above. Control populations remembered more positive (k =17, d = –0.726, SE = 0.305, z = –2.38, p = 0.017, 95%CI = –1.324, 0.128, T = 1.204), negative (k =29, d = –0.202, SE = 0.092, z = –2.189, p = 0.029, 95%CI = –0.382, 0.021, T = 1.204, T = 371), and neutral items (k = 37, d = –0.645, SE = 0.140, z = –4.954, p = 0.0001, 95%CI = –0.920, 0.370, T = 0.778) than did clinical populations.

11 References marked with an asterisk are included in the meta-analysis.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.