ABSTRACT
Emotional stimuli (e.g. words, images) are often remembered better than neutral stimuli. However, little is known about how memory is affected by an environmentally induced emotional state (without any overtly emotional occurrences) – the focus of this study. Participants were randomly assigned to discovery (n = 305) and replication (n = 306) subsamples and viewed a desktop virtual environment before rating their emotions and completing objective (i.e. item, temporal-order, duration) and subjective (e.g. vividness, sensory detail, coherence) memory measures. In both samples, a Partial Least Squares Correlation analysis showed that an emotional state characterised by high negative emotion (i.e. threat, fear, anxiety) and arousal was reliably associated with better memory in both objective (i.e. item) and subjective (i.e. vividness and sensory detail) domains. No reliable associations were observed for any temporal memory measures (objective or subjective). Thus, an environmentally induced state of negative emotion corresponds with enhanced memory for indices of episodic memory pertaining to “what” happened, but not necessarily “when” it happened.
Acknowledgments
We thank Dr. Rebecca Todd for feedback on this work, Jared Whitney for coding the temporal-order memory task, and Young Ji Tuen, Tyler Lin, Isabel Wilson, and Sarah Lacusta for checking reported statistics. We thank Dr. Lawrence Ward for suggesting running the PLS analysis using random data splits.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability
The data and code for this study are publicly accessible at https://osf.io/f3gtn/.
Notes
1 Since the time of our submission, additional papers have been published on this topic. Those papers are cited in the discussion.
2 Missing age data from one participant due to a reporting error.
3 Confidence data was not analysed.
4 Prior to calculating d’, a standard correction was applied to cases in which the hit rate was 1 or the false alarm rate was 0. In these cases, hit rate was computed as and false alarm rate as (Macmillan & Kaplan, Citation1985). In the discovery sample, we applied this methodology to adjust for a hit rate of 1 and a false alarm rate of 0 in 4 and 40 participants, respectively. In the replication sample, we adjusted for a hit rate of 1 and a false alarm rate of 0 in 7 and 59 participants, respectively.
5 The temporal error scores were calculated by averaging the absolute value of the difference between the position indicated by the participant and the correct order for each item.
6 The relative order scores were calculated by counting the number of items appropriately judged as following each ranked item, divided by the total number of possible correct relative judgments.
7 Correlation coefficients between temporal-order memory scores are as follows: Spearman and relative score, r = .98; Spearman and temporal error score, r = –.95; relative score and temporal error score, r = –.97.
8 Originally, this was planned as an in-person study, but was conducted online due to the COVID-19 pandemic, which is why we opted to use a video instead of full-blown navigational virtual reality.
9 Questionnaires were completed before and after the virtual world experience to assess health history, anxiety, depression, trauma exposure, sleep patterns, etc. As most of these measures were ancillary to the goals of the present study, they are described only in the Supplemental Materials. Only questionnaires that were specific to the present paper are described.