Spontaneous thought dynamics as a signature of positive and negative affectivity

Spontaneous thought unfolds as a dynamic stream that reflects continuously changing internal states and is closely tied to individual differences in mental health. However, quantifying the temporal structure of spontaneous thought remains challenging due to limited behavioral paradigms and analytic tools that preserve its continuous, multidimensional nature. Here, we introduced Density Map-Based Predictive Modeling, a framework that characterizes spontaneous thought dynamics elicited by the Free Association Semantic Task within valence–self-relevance–time space. Density Map-Based Predictive Modeling summarizes each individual’s thought stream using rating and vector density maps that capture both where thoughts concentrate and how they transition over time. Across multiple independent datasets (total N = 392), Density Map-Based Predictive Modeling yields robust, generalizable predictions of positive affectivity and negative affectivity using principal component regression. Feature analysis indicated that affectivity is reflected not only in the content of thoughts, but also in transition dynamics, with self-relevance as a key organizing dimension. Furthermore, higher predicted positive affectivity was associated with lower salivary C-reactive protein, suggesting a link between thought dynamics and inflammation. Together, these findings establish a scalable, interpretable approach for quantifying spontaneous thought dynamics and relating them to transdiagnostic affective traits and potential physiological correlates. Density Map-Based Predictive Modeling (DMPM) quantifies spontaneous thought dynamics in valence–selfrelevance–time space to predict positive and negative affectivity across independent datasets, linking thought-dynamic signatures to salivary Creactive protein.