Abstract
The constructivist nature of autobiographical memory leaves its retrieval susceptible to biases based on the current context. The present study addressed the hypothesis that the same memory could be recalled differently depending on a person's current regulation goals. In Experiment 1, a total of 58 participants recalled three events at a baseline session and again 2 weeks later when expecting to meet with an emotional individual. Individuals expecting to meet with a happy individual decreased the negative—and more specifically sad—words they used compared to the baseline session, whereas those expecting to meet with a sad individual showed the reverse effect. These findings were constrained by individual differences in extraversion. Significant effects were confined to events recalled second in order, suggesting the changes in emotional details might be due to controlled, regulatory processes. Using a false memory list-learning paradigm, Experiment 2 ruled out an alternative interpretation of the findings and confirmed that individuals can bias their memory in accord with regulation goals.
Keywords:
Acknowledgements
Thank you to Jenny Wong, Alicia Kinton, Jillian Burdziak, and Amy Peters for help with data collection and coding, and to Hiram Brownell for helpful discussions regarding statistical analysis. This research was supported by a Dana Foundation grant to EAK and an NDSEG fellowship to ACH.
Notes
1A series of one-way ANOVAs was run to compare the baseline emotional language according to the LIWC categories of interest (overall affect, positive and negative emotion, and sad words) in narratives about graduation vs spring break vs first week of classes. There was only a significant effect of event on the number of positive emotion words used at the baseline session, F(2, 173) = 4.89, p=.009. Post-hoc tests revealed that participants used more positive emotion words to describe graduation (M=3.96, SD=1.45) than the event from first week of classes (M=3.03, SD=1.67), p =.007, but neither was significantly different from the percentage of positive emotion words used in the spring break events (M =3.35, SD=1.88). Given that the emotional language used in each event was roughly equivalent, it does not seem likely that always recalling graduation first would have influenced subsequently recalled events any more than recalling events from spring break or the first week of class first in order would have.
2A total of 30 Boston College freshmen were randomly assigned to hear either the same happy, sad, or neutral stories about a second experimenter as in Experiments 1 and 2. An additional 10 participants were not told about any meeting with a second experimenter. Immediately before and after hearing the stories, participants rated the extent to which they were currently experiencing several negative (e.g., sad, depressed, down, in a bad mood) and positive (e.g., happy, elated, pleased, enthusiastic) emotions. The mean negative and positive emotion ratings from before and after the second experimenter stories were compared using two 4 (Group: happy, sad, neutral, or no story)×2 (Time: before and after hearing the story) within-participants ANOVAs. Extraversion was used as a covariate in the ANOVAs to maintain consistency with Experiments 1 and 2. There was no significant Group×Time interaction for positive emotions (happy story M 1 =2.98, SD 1 =.68, M 2 =2.93, SD 2 =.64; sad story M 1 =2.68, SD 1 =.68, M 2 =2.7, SD 2 =.48; neutral story M 1 =3.06, SD 1 =.52, M 2 =2.52,, SD 2 =.4; no story M 1 =2.82, SD 1 =.72, M 2 =2.47, SD 2 =.45), F(1, 35) = 1.72, p=.18, partial η2=.13. There was also no significant Group×Time interaction for negative emotions, (happy story M 1 =1.14, SD 1 =.28, M 2 =1.64, SD 2 =.42; sad story M 1 =1.2, SD 1 =.37, M 2 =1.93, SD 2 =.39; neutral story M 1 =1.23, SD 1 =.28, M 2 =2, SD 2 =.49; no story M 1 =1.23, SD 1 =.28, M 2 =1.88, SD 2 =.56), F(1, 35) = 0.59, p=.62, partial η2=.05.
3A total of 17 participants heard the same happy (n=8) or sad (n=9) stories about a second experimenter as in Experiments 1 and 2. Critically, they were not told that they would have to interact with the second experimenter, so as to remove any possibility that they might try to regulate their emotions in advance of the meeting. After hearing the stories, individuals performed a lexical decision task meant to measure semantic priming to emotional concepts (adapted from Innes-Ker & Niedenthal, Citation2002). The lexical decision task included 16 happiness-related words, 16 sadness-related words, 32 neutral words, and 64 pseudowords. A 2 (Group: happy story, sad story)×2 (Emotion word type: happy, sad) within-participants ANOVA revealed no significant difference in reaction times to the happy and sad words on the lexical decision task depending on whether individuals heard a happy or sad story (happy story: positive word M=.79, SD=.15, negative word M=.81, SD=.12; sad story: positive word M=.77, SD=.16, negative word M=.84, SD=.25), F(1, 15) = 0.54, p=.48, partial η2=.04.