ABSTRACT
Most contemporary models frame emotion as an intrapersonal dynamic system comprised of subcomponents such as appraisals, experience, expressive behaviors, and autonomic physiology that interact over time to give rise to emotional states. Predicting and intervening in these temporal interpersonal emotion systems (TIES) requires statistical models that represent complex, dynamic interdependencies across emotional subcomponents over time, both within individuals and between partners in an emotional transaction. Cross-disciplinary collaboration research team includes social and computational scientists working together to represent TIES theory in the form of testable mathematical models, a process we call "mindful modeling". The social scientists provide a theory about some aspect of TIES, along with relevant data, and the computational scientists provide a Bayesian generative model that embodies the key aspects of that theory, along with the ability to estimate the parameters of the model. A Bayesian generative model is a specification for the combine probability distribution of: observed data, parameters of a mathematical model, and latent variables.
