Zeitliche Aufmerksamkeit und Erwartungshaltung im Rahmen prädiktiver Kodierung
Zusammenfassung der Projektergebnisse
The main aims of this project were to disentangle the effects of different top-down influences on perception, namely attention and expectation, in terms of the underlying neural processing. The conducted work drew upon the predicting coding framework, an influential theoretical and computational model of brain processing. In the conducted experimental and modelling work, I focused on whether attention and expectation are dissociated in time perception. In an experiment using magnetoencephalography (MEG), the effects of temporal attention and sensory expectation were disentangled in terms of their effects on evoked neural responses and sustained neural oscillations. Applying computational modelling (dynamic causal modelling; DCM) to the acquired data allowed to explain the mechanistic implementation of attention and expectation in cortical microcircuits. In further collaborative studies, DCM was used to study the dynamic mechanisms of processing predictable and unpredictable auditory sequences (using MEG), as well as to disentangle contentbased and time-based expectancy (using electrocorticography; ECoG). Finally, the inferred mechanisms of attention, expectation and other contextual influences on perception were integrated in a coherent theoretical framework. The view on perception that emerges from these studies is that it is continuously shaped by prior experiences and current goals. By combining experimental and computational neuroscience methods, the effects of prior experiences (expectation) and current goals (attention) on perception can be understood in considerable detail, both in terms of the underlying biophysical mechanisms (such as dynamic modulation of activity in specific layers of cells in sensory cortices) and mathematical principles governing the neural dynamics. The experiments included in this project were designed to address basic research questions regarding the neural and computational mechanisms of attention and expectation in perception. However, the methods used in the project – specifically dynamic causal modelling – have been proposed as useful tools in non-invasive biomarker identification in groups at risk for developing psychiatric conditions such as schizophrenia. The studies conducted in this project offer a further validation of dynamic causal modelling as a useful method of inferring biophysical mechanisms from electrophysiological data acquired in humans non-invasively. Since both attentional and expectation-related processes are typically disrupted in schizophrenia, the work conducted here also contributes to a better understanding of the crucial mechanisms at play which can be used to improve the existing methods of diagnosing and prognosing schizophrenia.
