My PhD thesis showed how machine learning approaches (dynamical systems theory, Bayesian inference) and developmental systems theory could together explain how we develop emotional habits across the lifespan. In this framework, the moment-by-moment emotions that influence our actions and decision-making are a product of causal forces interacting across multiple timescales, informed by evolution, life history, and the rapidly-shifting exigencies of the social and material environment. A recent publication output from my thesis is available here, and other adapted articles are currently under revision or in preparation.

Dynamical approaches to emotion provide an interesting alternative to the view that emotions are ‘natural kinds’, meaning that every instance of the kind looks the same and shares a collection of features or properties that always co-occur (Barrett, 2006). Unlike natural kind approaches, dynamical approaches have the resources to explain: (i) natural variation in how people respond emotionally to identical triggers; and (ii) the situational and temporal context-sensitivity of emotional responses. As dynamical approaches to emotion advance, they are likely to spur novel work in the science of emotion.

Right: A simple dynamical representation of emotion (Rasa, Leonidas & Gintautas, 2020). Here, different emotion kinds are demarcated on the basis of covarying patterns of arousal (high or low) and valence (positive or negative). Each emotion kind has a distinctive pattern, and the system can move between different emotion kinds, as arousal and valence patterns shift.