Abstract
Functional MRI (fMRI) research has traditionally investigated task processing using static blocked or event-related designs. Consequently, our understanding of threat processing remains limited to findings from paradigms with restricted dynamics. In this paper, we applied switching linear dynamical systems (SLDSs) to uncover the dynamics of threat processing during a continuous threat-of-shock paradigm. Unlike typical systems neuroscience studies that assume systems are decoupled from external inputs, we characterized both endogenous and exogenous contributions to the dynamics. We first demonstrated that the SLDS model learned the regularities of the experimental paradigm; states and state transitions estimated from fMRI data across 85 regions of interest reflected both threat proximity and direction (approach vs. retreat). After establishing that the model captured key properties of threat-related processing, we characterized the dynamics of states and their transitions. The results reveal how threat processing can be viewed as dynamic multivariate patterns whose trajectories are determined by intrinsic and extrinsic factors that jointly drive how the brain temporally evolves. Furthermore, we developed a measure of region importance to quantify individual brain region contributions to system dynamics, complementing the system-level SLDS formalism. Finally, we demonstrated that an SLDS model trained on one paradigm successfully generalizes to a separate experiment, capturing fMRI dynamics across distinct threat-processing tasks. We propose that viewing threat processing through the lens of dynamical systems offers vital avenues to uncover properties of threat dynamics not unveiled by standard experimental designs.