Danger! Negative memories ahead: the effect of warnings on reactions to and recall of negative memories

ABSTRACT

A trigger warning is an alert that upcoming material containing distressing themes might “trigger” the details and emotion associated with a negative memory to come to mind. Warnings supposedly prevent or minimise this distress. But, do warnings really have this effect? To simulate the experience described above, here, we examined whether warning participants—by telling them that recalling a negative event would be distressing—would change characteristics associated with the immediate and delayed recall of a negative event (such as phenomenology e.g., vividness, sense of reliving), compared to participants who we did not warn. Generally, we found that time helps to heal the “emotional wounds” associated with negative memories: negative characteristics—such as emotion, vividness etc.—faded over time. However, the event’s emotional impact (the frequency of experiences related to the event such as “I had trouble staying asleep”), subsided less over a two-week delay for participants who were warned in the first session. Our findings suggest that warning messages may prolong the negative characteristics associated with memories over time, rather than prepare people to recall a negative experience.

KEYWORDS:

• Trigger warnings
• warnings
• nocebo effects
• autobiographical memory
• misinformation

Disclosure statement

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

Notes

1 These conditions were collapsed for our main analyses, but we report key findings here (below) and full results can be found at: https: https://osf.io/x6t7v/

2 We originally planned to analyse our dependenat variables using a 2 (Session 1 warning condition: warned, unwarned) × 2 (Session 2 warning condition: warned, unwarned) × 2 (Session Time: Session 1, Session 2) mixed design. We conducted an a priori power analysis for a 4 (between) × 2 (within) repeated measures ANOVA with the smallest effect size we would be interested in (f = .15), power of .95, and r = .48, based on a prior correlation between repeated measures of emotion about a recent negative event (Takarangi & Strange, 2010). The recommended sample size was 204. We calculated this power analysis because G*Power does not have the capability to calculate power for mixed designs beyond a single between subjects’ level. However, a previous reviewer rightly pointed out that we were likely therefore underpowered for a 2 (between) × 2 (between) × 2 (within) subjects design. While we could have analysed our variables using 4 (between) × 2 (within) subjects’ analyses, we do not believe this analysis reflects the true nature of our design, because participants are only in two groups (warned or unwarned) in Session 1. Additionally, the repeated warning in Session 2 was a secondary interest. Therefore, we reframed our analyses to focus on the effects of the Session 1 warning condition and analysed our variables using a 2 (Session 1 warning condition: warned, unwarned) × 2 (Session Time: Session 1, Session 2) design, collapsing the Session 2 warning condition. This change allowed us to reach suitable power. We report the analyses of the full 2 × 2 × 2 design here: https://osf.io/x6t7v/, and report any notable findings related to our secondary aim regarding the accumulative effects of the warning message for participants warned in Session 1 and Session 2 in our results section below.

3 When reconstructing events from autobiographical memory, a person’s belief in the memory actually occurring (rather than being imaginary) is enhanced if the event is recalled from a visual perspective that matches how the event-related information is retained in memory (Marsh et al., 2014). Recent memories are more likely to be recalled from a first-person rather than a third-person perspective—therefore when recalling a memory from the past-two weeks, someone would be more likely to believe that it has really occurred if it is recalled from an observer (first person) versus field (third-person) perspective (Marsh et al., 2014).