Zum Inhalt springenZur Suche springen


Bei Interesse an einer Abschlussarbeit in unserer Arbeitsgruppe kontaktieren Sie bitte Gerhard Jocham.

Bitte beachten Sie Folgendes:
Es ist nicht erforderlich, dass Sie bereits über Programmierkenntnisse verfügen, wenn Sie Ihre Abschlussarbeit in unserer Abteilung anfertigen wollen. Allerdings werden Sie im Rahmen Ihrer Tätigkeit fast zwangsläufig mit Analyse-Skripten in Kontakt kommen, die beispielsweise in Python, Matlab, oder R geschrieben sind. Sie sollten daher bereit sein und Interesse daran haben, mit diesen Werkzeugen umzugehen und gegebenenfalls auch Grundkenntnisse im Programmieren zu erwerben.

Cortical mechanisms of decision making
Computational models of choice are based on competition between pools of neurons in recurrent cortical networks. A key element in these circuits is the balance between slow recurrent excitation at NMDA glutamate receptors and GABAergic feedback inhibition. Further, reward feedback modifies synaptic strength via Hebbian plasticity rules. In this project, we investigate the role of transmission at NMDA glutamate, GABA-A, and M1 muscarinic acetylcholine receptors in (i) perceptual and reward-based decision making and (ii) reinforcement learning in stable and volatile environments.

Note: Acquisition of these data sets (MEG + behaviour) has already been completed. This project would be particularly suited for students who have a keen interest in coding and in formal models of behaviour and neural activity (drift diffusion models, algorithmic models of choice and learning).

Patch-leaving decisions
Patch-leaving decisions are essentially a kind of stay-or-leave problem - should an organism keep exploiting resources in its current habitat or leave for a potentially richer area? This requires considering the reward rate in your current habitat, the overall reward rate in the environment, and the cost (energy and foregone reward) of leaving. This kind of choice is thus somewhat different from 'standard' reward-based choice, where one typically selects the option with the highest value out of two or more alternatives. Yet, there is good reason to assume that most choices are indeed building on the very same mechanims that have evolved to solve patch-leaving decisions. Indeed, even high-level multi-alternative reward-based choice can be cast as a series of accept-reject decisions, much like in patch foraging.

Here, we investigate how various environmental parameters guide patch-leaving. Moreover, there is evidence to assume that patch-leaving is driven by bursts of noradrenergic activity emanating from the locus coeruleus. We therefore use taVNS (transcutaneous auricular vagus nerve stimulation) to indirectly manipulate noradrenergic activity. We combine this with recording of pupil size, as changes in pupil size closely track (amongst other parameters) neural activity in the locus coeruleus. For this project, students should have an interest in collecting and analyzing behavioural and psychophysiological (pupil recordings) data and in applying taVNS.