Cognitive Stage Detection and Interpretation:discovering the Underlying Stages in Various Tasks Using Machine Learning Applied to EEG and MEG

Every process of our life can be divided into steps or, in other words, stages that describe this process. This concerns simple daily activities such as preparing to go to work and more complex activities such writing a scientific paper or this thesis. Thus, even in completely different areas of life defining the constituent stages play a crucial role for understanding the process. With this thesis, we aim to provide a better understanding of the stages encompassing cognitive processing. To obtain this goal, we have used a combined technique from artificial intelligence and neuroscience, a novel machine learning algorithm called Hidden semi-Markov Model multivariate pattern analysis (HsMM-MVPA). This big-data method takes all data of an experiment into account simultaneously, to ultimately track cognitive processes in much more detail. In this thesis, we applied HsMM-MVPA to two imaging techniques used to measure brain activity during the performance of cognitive tasks, electroencephalography (EEG) and magnetoencephalography (MEG). These techniques have very high temporal resolution allowing to record brain activity on a millisecond timescale. To obtain robust conclusions on the cognitive stages, we combined HsMM-MVPA with experimental manipulations and other traditional modelling techniques. Additionally, we wanted our stage definitions to be task-independent; for this reason, we used three different tasks of increasing complexity: simple visual discrimination, lexical decision and vocabulary learning tasks