Directed EEG microstate transitions during affective processing complement static microstate metrics in bipolar disorder

• Task-evoked microstate syntax quantifies directed EEG state switching • Directed transitions complement static microstate metrics in affective EEG • Window-specific transition deviations were strongest in the LPP interval • Early anchor outflow showed a modest, exploratory relation to response speed • Transition results were stable to smoothing but sensitive to labeling/filter choices Electroencephalographic (EEG) microstates summarize fast, recurring large-scale brain states, yet most microstate work emphasizes static properties (e.g., duration, coverage) rather than the directed organization of state-to-state switching (“microstate syntax”), particularly in task-evoked settings. We operationalized task-evoked microstate syntax by quantifying directed transition deviations from independence within canonical ERP windows of affective processing (N200, P300, LPP). We analyzed stimulus-locked EEG from healthy controls (n = 15), individuals with bipolar disorder (n = 44), and unaffected siblings of patients with bipolar disorder (n = 14), using age-adjusted models and false-discovery-rate control within each ERP family. Static microstate metrics showed limited and pipeline-sensitive group effects. In contrast, directed transition structure revealed clearer window-specific deviations, with the strongest and clearest effects concentrated in the LPP window. Trial-level mixed-effects analyses identified a modest N200 association in which greater early “anchor” outflow covaried with faster responses on valenced trials, but this effect was small and sensitivity-dependent. Robustness analyses showed that transition-based inferences were stable to stronger temporal smoothing and that the concentration of transition deviations in the LPP window was reasonably stable under modest ERP-window shifts, whereas individual-edge and trial-level effects were more sensitive to labeling, filtering, and window-boundary choices. Together, these findings support task-evoked microstate syntax as a complementary dynamic descriptor of distributed state switching during affective processing, beyond what static microstate measures captured in this dataset.