Förster and Dannenberg‘s (2010) GLOMOsys theory claims that people process perceived events and internal information in a more global or more local processing mode and that adopted modes should transfer to other, unrelated tasks. If so, global/local processing modes would qualify as metacontrol states that are assumed to regulate processing dilemmas, like persistence/flexibility, exploitation/exploration, or speed/accuracy (Goschke, 2000). Given increasing rates of nonreplications of previously demonstrated far transfer from prime tasks that are likely to induce a particular global or local processing bias to logically and temporally unrelated probe tasks, we tested whether near and far transfer can be demonstrated under conditions that should be optimal for such transfer. We reduced the temporal distance between prime and probe trials by integrating them into a dual-task paradigm and used probe tasks that were either almost identical to the prime task or at least shared the relevant modality and attentional demands. We obtained significant transfer effects between almost identical visual global/local tasks, irrespective of the degree of cognitive conflict that these tasks generated, but did not find any evidence for somewhat farer transfer to other visual tasks, like a flanker task and an attentional blink task. That is, any substantial change in the probe task’s characteristics compared with the prime task eliminated almost any signs of transfer. Altogether, we conclude that either global/local processing modes as envisioned by GLOMOsys do not exist or they normally do not transfer from global/local tasks to other, unrelated tasks.