On the conceptualisation and forecasting of emotion dynamics in healthy and psychiatric individuals

Throughout the day, individuals experience a variety of feelings, such as amusement, relaxation, and envy. By observing the regularities in the stream of affect, individuals learn to predict future emotions from current ones and develop accurate mental models of emotion transitions. This thesis aims to explore the cognitive architecture underlying these affective forecasts. Through a series of experiments involving both healthy and psychiatric individuals, we investigate i) the temporal boundaries of affective predictions and their evolution over time, ii) the influence of the conceptual knowledge about emotions on transition judgements at various timescales, iii) the impact of dysfunctional affective dynamics on the forecast of future emotions. Results indicate that people trust more their predictions in the near future, with confidence dropping after 24 hours. We identified nine prototypes in the temporal profiles of affective forecasts and mapped their trajectories in a two-dimensional space defined by transition plausibility and slope. Also, we characterise emotions as starting states (e.g., surprise) or end-points (e.g., irritation) based on transition judgments, and reveal asymmetry in forecasts for specific transitions (e.g., relief → fear). Analysis of the scaffolding of affective forecasts confirms the relevance of conceptual knowledge about emotions in shaping mental models of emotion transitions. Our findings indicate that similarities between emotions in certain dimensions (e.g., valence) inform predictions regardless of the time interval, while others (e.g., arousal) exert influence only within specific timescales. We demonstrate that psychiatric disorders such as depression and bipolar disorder significantly affect the architecture of affective forecasts, although these adjustments do not undermine the core predictive structure. Findings suggest that patients use their internal emotion dynamics as a reference to construct (or refine) their predictive models of emotion transitions.