The Affective Ising Model: A computational account of human affect dynamics

Area of Science:

• Computational Psychiatry
• Affective Science
• Statistical Mechanics

Background:

• The human affect system generates positive and negative feelings, crucial for daily life.
• Dysregulation of the affect system underlies mood disorders.
• Understanding affect dynamics is key to explaining normal experiences and mood pathology.

Purpose of the Study:

• To propose a nonlinear stochastic model, the Affective Ising Model (AIM), for positive and negative affect dynamics.
• To explain empirical phenomena like skewness, multimodality, and non-linear affect relations.
• To assess AIM's ability to reproduce non-Gaussian features in real-world affect data.

Main Methods:

• Developed the Affective Ising Model (AIM) based on statistical mechanics and neurophysiological evidence.
• Applied AIM to large-scale experience sampling data from individuals in normal and mood disorder states.
• Compared AIM against continuous-time vector autoregressive (VAR) models using leave-one-out cross-validation.

Main Results:

• AIM effectively reproduced non-Gaussian features in affect data, outperforming benchmark models.
• Model performance was comparable for Gaussian features.
• AIM demonstrated superior predictive accuracy for most affect time series.

Conclusions:

• The Affective Ising Model (AIM) offers a robust computational framework for understanding human affect.
• AIM accurately captures complex, non-Gaussian dynamics of positive and negative affect.
• AIM shows promise for modeling mood regulation and dysregulation in clinical populations.