A hierarchical latent stochastic differential equation model for affective dynamics

Area of Science:

• Psychology
• Computational Neuroscience
• Affective Science

Background:

• Core affect, a fundamental 2-dimensional concept, underlies all human affective experiences.
• Understanding the temporal dynamics and individual variability of core affect is crucial.
• Existing models may not fully capture the continuous and individual-specific nature of affective changes.

Purpose of the Study:

• To present a continuous-time stochastic model, the Ornstein-Uhlenbeck process, for modeling core affect.
• To account for temporal changes in core affect at a latent level.
• To investigate individual differences in affective dynamics by allowing model parameters to vary across individuals.

Main Methods:

• Utilized the Ornstein-Uhlenbeck process, a continuous-time stochastic model.
• Extended the model hierarchically to accommodate individual differences in key parameters (mean, variance, regulatory mechanisms).
• Developed a continuous-time state-space model for analyzing repeated, potentially irregular, longitudinal data, incorporating covariates.

Main Results:

• The proposed model successfully captures latent temporal changes in core affect.
• The hierarchical extension allows for the estimation of individual-specific parameters, revealing unique affective dynamics.
• The model demonstrated flexibility in analyzing intensive longitudinal data from a diary study.

Conclusions:

• The Ornstein-Uhlenbeck process provides a robust framework for modeling core affect dynamics.
• The hierarchical extension effectively models individual differences in affective processes.
• This approach advances the understanding of both general affective mechanisms and unique individual emotional trajectories.