Beyond the trade off: testing a two-dimentional framework of adaptive control in a ERP study

This thesis aims to investigate how adaptive control modulates cognitive flexibility and cognitive stability in response to environmental demands. It compares two competing theoretical accounts: the trade-off view, which conceptualizes flexibility and stability as antagonistic processes, and the more recent two-dimensional framework, which characterizes them as orthogonal and independently regulable dimensions of cognitive control. The study introduces four main innovations. First, flexibility and stability are measured simultaneously by embedding a Stroop paradigm within a task-switching design, allowing both control demands to be manipulated and assessed within the same experimental context. Second, trial-by-trial expectancies are modeled using a Bayesian observer (Hierarchical Gaussian Filter, HGF) to derive continuous expectancy variables indexing predicted incongruency and predicted switching. Third, the paradigm employs a visuospatial Stroop variant that closely follows Kornblum’s dimensional overlap taxonomy, corresponding to a Type-8 ensemble, thereby strengthening the operationalization and interpretability of stability-related conflict effects. Fourth, EEG recordings are collected to obtain converging behavioral and ERP evidence (mainly P3 and Switch Positivity ERP) relevant to the (in)dependence of the two control modes. Overall, the findings provide no evidence for a strong unidimensional stability–flexibility trade-off. Expectancy updates are control mode specific: stability-related expectancy mainly affects incongruency, whereas flexibility-related expectancy mainly affects switching. Their influence did not propagate to the other control mode, as indicated by the absence of higher-order interactions. This pattern was mirrored at the neural level, where a clear double dissociation emerged: incongruency expectancy reliably modulated the Stroop-related ERP (P3) but not the switch-related ERP (Switch Positivity), whereas switching expectancy reliably modulated the Switch Positivity but not the P3.