By using the task-shifting paradigm we plan to examine how controlled and automatic processes interact to enable efficient performance under ambiguous stimulus conditions. Our previous findings suggest that under these conditions controlled sequential selection stages are needed to prevent errors caused by automatic processing. In the first part of the project we will examine to what extend these selection stages can be flexibly arranged in order to cope with different task requirements. This will be investigated by varying the temporal order of cue and stimulus. It is expected that this order determines a corresponding mental order of judgment selection and response selection. The second part is concerned with the role of controlled processes for error detection. We assume that internal error feedback modulates associative learning, which, in turn, is responsible for the residual shift costs. More specifically, we hypothesize that learning is prevented if an error is detected, and that this reduces the costs of task shifting. Specific predictions concerning the effects of error type, different error correction mechanisms, and conscious error detection on the shift costs will be tested. Both parts of our project will contribute to a better understanding of the interaction between executive control and automatic processes.