A multistable slow-fast model of affective state switching under circadian drive

Rhythmic mood fluctuations have long been linked to circadian ([~]24 h) timing, but how this physiological rhythmicity relates to pathological affective cycling such as bipolar episode recurrence remains unclear. Here we introduce a slow-fast dynamical model with multistability, motivated by hypothalamic-pituitary-adrenal (HPA) axis feedback. We build on a reduced slow-fast formulation with a slow endocrine variable and a fast affective variable, and reshape the fast nullcline to yield four stable fixed points and allow the model to distinguish between normal diurnal mood variation and pathological, depression-like/mania-like extremes. A sinusoidal circadian drive promotes regular alternation within the physiological pair, while temporally correlated fluctuations modeled as an Ornstein-Uhlenbeck process trigger probabilistic escapes. Simulations show that weakened circadian amplitude increases the probability of transitions into pathological attractors and produces prolonged dwell times in affective extremes. Small geometric biases in the nullcline can generate predominant polarity toward depressive or manic episodes. The model provides a conceptual framework linking circadian forcing, persistent stochastic perturbations, and multistability, and generates testable predictions for how circadian disruption destabilizes affective trajectories in mood disorders.