Switch-independent task representations in frontal and parietal cortex [thesis]

Lasse Steffen Loose, Universitätsbibliothek Der FU Berlin
2020
In order to adapt to the surrounding dynamic environment humans must be able to frequently switch between tasks. This switching requires cognitive control. It is effortful and impairs task performance which is reflected in behavioral switch costs. Task switching research has investigated the contributions of top-down and bottom-up processes on switch costs and there is an ongoing debate about the composition of the behavioral performance decrease present in switch trials. fMRI studies that
more » ... tigated the underlying neural processes of task switching have mostly focused on regional activation differences. Frontal and parietal regions show increased activity for task switches and task repetitions, but investigation of activation differences yielded heterogeneous results. Recent multivoxel pattern analysis enables the investigation of neural task representations which are located in frontal and parietal brain regions. The question arises how these task representations might be subject to task switching processes: they might be strengthened by increased control demands during switching or be degraded by interference of the previous task set. Alternatively, task representations might not be affected by task switching processes and task representation might generalize across switched-to and repeated tasks. The present study cued participants to perform one of two tasks (with equally frequent task repetitions and switches). It compares the accuracy with which a linear classifier is able to decode the currently performed task from patterns of participants' brain activity. This reflects the strength of a task representation. As expected, tasks were represented in frontal and parietal cortex. However, there was no difference of decoding accuracy between switch and repeat trials. Moreover, tasks are represented in a switching independent spatial pattern in these regions. There was no evidence for the decodable strength of task representations to account for the performance cost associated with task switching.
doi:10.17169/refubium-27858 fatcat:6q3bkupf7zbljf6z6xqv6ukutu