The expectation-updating mechanism in gratitude: A predictive coding perspective

The fluctuations in emotions during constant help are unexplained by traditional emotion theories but may align with the predictive coding theory. This theory suggests that individuals tend to form expectations of others' help during social interactions. When outcomes exceed expectations, positive prediction errors are generated, potentially increasing gratitude. Conversely, constant help may build up expectations that surpass outcomes, resulting in negative prediction errors and reduced gratitude. Nevertheless, there is a lack of studies to examine the relationship between prediction errors and gratitude and its underlying mechanism. Here, we conducted two studies. Study 1 consistently found that higher expectations were associated with lower gratitude, when benefactors refused to help, in both reward-gaining and punishment-avoiding tasks. Moreover, prediction errors were positively and reliably linked to gratitude. Study 2 further identified that gratitude dynamically changed through an expectation-updating mechanism. A computational model incorporating predictive coding outperformed traditional theories in predicting the dynamics of gratitude. The findings support predictive coding theory, providing a temporal perspective and a mechanistic understanding of the fluctuations in gratitude, thus having implications for new interventions to improve mental health and well-being. (PsycInfo Database Record (c) 2025 APA, all rights reserved).