Emotional reactions to real-world events predict shifts in longer term affective states.

Our affective experiences, including both positive and negative emotions and longer term affective states (LTASs) like moods, color everyday life. Emotional responses, momentary affective reactions arising from individual events and lasting from seconds to hours, can impact one's mood, diffuse affective states that lack a single causal event, and last much longer (e.g., days to weeks). However, the extent to which emotional responses and the events that elicited them impact mood on these longer timescales is unclear. Using a high-stakes, naturalistic paradigm, we tested whether exam grades, grade prediction errors, or the emotional reactions they elicit predict shifts in LTAS. For several semesters from 2019 through 2022, we assessed 821 college students' emotional responses to exam grades and, in the weeks surrounding exam grade feedback, assessed their LTASs. We demonstrate that emotional reactions to exam grades, grades themselves, and exam prediction errors all predict shifts in LTAS but that these effects decay across time. After completing model comparisons of several combinations of predictors, we found that the model best fit to predict LTASs includes immediate emotional response, grade, and exam prediction errors. Upon examining the model outputs, we found that emotional response and exam grade primarily predict changes in LTASs, suggesting that LTASs are not just an aggregate of time-attenuated emotions but can reflect other features like outcome value. However, there may be limits to the generalizability of the findings due to the diversity of the age, race, and gender diversity of the sample. (PsycInfo Database Record (c) 2026 APA, all rights reserved).