To meet cognitive control demands of everyday life, we must continuously adapt our behavior, balancing cognitive stability and flexibility. According to the control-dilemma framework, stable goal-shielding and flexible goal-shifting control modes entail opposing performance costs and benefits. In a within-subjects design, we aim to induce both control modes by manipulating the frequency of task-switches in a cued task-switching paradigm. A high (75%) task-switch proportion of trial transitions in the shifting condition demands flexible shifting between tasks, whereas a lower proportion (25%) permits a more stable control mode within shielding blocks. In both conditions, we will intersperse identical short “diagnostic” test trial sequences with 50% task-switches, assuming that the contextually induced control modes will persist during these test blocks. In line with the assumption of a performance cost-benefit trade-off between goal-shielding and goal-shifting, we expect higher switch costs and reduced interference costs in the shielding condition, and vice versa in the shifting condition. At a neural level, we aim to obtain fMRI indicators to decode the underlying neural patterns of both control modes, investigating whether the respective states are reflected in more or less stable neural task representations (as indicated by enhanced or reduced decoding accuracy in shielding and shifting blocks, respectively). We predict that the shielding-shifting trade-off will show in the brain signals as more stable task representations and enhanced decoding accuracy in the shielding relative to the shifting condition. Moreover, we will assess whether more stable control modes are also reflected in more stable dynamic whole-brain functional connectivity patterns.