The fading affect bias in the context of emotion activation level, mood, and personal theories of emotion change

Abstract

The intensity of emotions associated with memory of pleasant events generally fades more slowly across time than the intensity of emotions associated with memory of unpleasant events, a phenomenon known as the fading affect bias (FAB). Four studies examined variables that might account for, or moderate, the bias. These included the activation level of the emotions, individual differences in dispositional mood, and participant expectations of emotion change across time. Results suggest that (a) although emotion activation level was related to overall fading of affect, it was unrelated to the FAB; (b) dispositional mood moderated the FAB, but could not fully account for it; and (c) although participants’ predictions of event-related emotion change across time were somewhat veridical, the FAB emerged even when these predictions were accounted for statistically. Methodological and theoretical implications for research on the affect associated with autobiographical events are discussed.

Keywords:

• Autobiographical memory
• Emotion
• Arousal
• Valence
• Fading affect bias
• Personal theories

Notes

¹There are a number of ways to approach our analyses. One is to use the initial affect rating and final affect ratings as repeated measures. A second is to use the initial affect rating as a covariate in analyses that predict the final affect rating. A third is to calculate and analyse change scores. Three considerations lead us to prefer the change score approach. First, it is descriptively appropriate: The fading affect bias is literally defined as a difference in the rate of change exhibited by positive affect associated with memories compared to the rate of change exhibited by negative affect. Second, change scores afford an efficient method of presentation, effectively conveying the nature of the effect under investigation. Third, our analyses of these data, as well as other data sets in our research programme, show that the inferences drawn from alternative analytic methods are usually identical to the inferences that are drawn from analyses of change scores.

²Procedures that change metric variables to categorical variables (e.g., median splits) are sometimes criticised because such procedures can lower analytic power or produce misleading results. Accordingly, the analyses described in this article were also performed treating affect measures as a continuous, metric variable. Because statistical power was large, and given that the inferential results were substantially similar across analyses, ease of description led us to report the quartile-split results.

³A separate analysis was conducted to ensure that the FAB was more of a function of dispositional affect pleasantness than it was a function of dispositional life satisfaction, a cognitive appraisal variable that is central to well-being (Diener, Emmons, Larsen, & Griffin, 1985). The ratings on the life satisfaction measure were used to place participants into one of four quartiles. This life satisfaction variable was then used in hierarchical regression analyses. These analyses examined the extent to which the FAB was related to the variables of event valence, satisfaction with life, and the interaction between the two. The results suggested that affect fading for pleasant and unpleasant autobiographical events was approximately uniform across each quartile of dispositional life satisfaction, F(3, 1368) = 0.87, ns. The results support a mood, rather than cognitive appraisal, interpretation of the results in Study 3.