This thesis investigates the neural basis of the decision threshold. The neural basis of decision making is one of the most central topics in cognitive and systems neuroscience. The ultimate goal of this research is to provide a complete account of the mechanisms and neural substrates underlying our ability to make choices in complex and ever-changing environments. Perceptual decision making has previously been described as the continuous accumulation of noisy sensory evidence in the brain until a decision criterion or threshold is reached. A decision threshold ultimately determines the decision process by balancing evidence accumulation and deliberation time in order to appropriately select an action. Let us consider a simplified scenario in which foragers sample the environment for available patches of food. When plentiful resources are available, foragers should thoroughly examine and thus deliberately choose rich patches over empty patches to maximize reward. Compare this to a situation where resources have been nearly exploited. Now, decisions about which patch to choose have to happen quickly as foragers compete for the remaining resources. While these two situations differ in their incentive structure, they do share a common component. The forager adjusts her decision criterion in each situation that would appropriately address the change in circumstances. Whereas accumulation of sensory evidence has been extensively researched having acquired a substantial body of knowledge, the mechanisms of the flexible decision threshold remain poorly understood. Hence, motivated by compelling theoretical frameworks - such as the sequential sampling framework of decision making - I have investigated mechanisms of the decision threshold during perceptual decision making. In the first project, I investigated the neural mechanism of decision threshold modulation for reward maximization in a direction of motion discrimination task. By combining functional MRI (fMRI) with computational models of perceptual decision making, i.e. the drift diffusion model (DDM), it can be shown that modulation of the decision threshold is achieved by adjusting the effective connectivity within cortico-striatal and cerebellar-striatal brain systems. This change-in-connectivity mechanism, which has the effect of balancing different decision relevant sources of information, reflects a central aspect of decision making. Most importantly, a modulation of the decision threshold influences performance: a lower threshold leads to faster but also less accurate decision and vice versa. It has been theoretically shown that sequential sampling models, especially the drift-diffusion model can adjust the decision threshold in an optimal way for simple forms of decision making. Optimality under these conditions is defined as the optimal trade-off between decision time and performance accuracy. Notably, in these decision situations, time for deliberation and error rate are negatively related. The model can minimize decision time for a specified performance level. As a consequence, this notion of optimality has potential implications for formulations of information processes explaining perceptual decision making. In the second project, I investigated neurocomputational models of optimal I investigated neurocomputational models of optimal decision making via cortico-basal ganglia circuitry to investigate this feature in detail. These models have ascribed a crucial computational role to the subthalamic nucleus (STN) and are (mathematically) equivalent to the models used in the first study. In this project, participants suffering from Parkinson’s disease (PD) were also asked to perform a motion discrimination task. Here we used a combination of psychophysics, neurostimulation and computational modeling and probed perceptual decision making under a) different states of deep brain stimulation (DBS) of the subthalamic nucleus (STN) and b) changing response instructions. The results indicate that DBS of the STN significantly influences a) performance for perceptual judgments under high decision conflict, and b) the magnitude of decision criterion adjustment. These findings are of particular interest, as they demonstrate on an individual basis that STN computation is crucial for an optimal balance between competing decision demands. Taken together, I investigated mechanisms of the decision threshold and showed that the modulation of a threshold for perceptual decision making is instantiated by a change in connectivity between cortio- striatal and cerebellarstriatal brain systems. In addition, I demonstrated that the optimal adjustment of the decision threshold is likely to be based on computations by the STN. Together these findings advance the idea of the decision threshold as a general mechanism for decision making and shed light on the neural mechanisms implementing it.
In dieser Arbeit habe ich Mechanismen des menschlichen Gehirns untersucht, die zur Anpassung des Entscheidungskriteriums führen. Bei Wahrnehmungsentscheidungen, wie z.B. dem Erkennen der Bewegungsrichtung eines Objekts (z.B. nach links oder rechts), wird aufgrund theoretischer und empirischer Befunde angenommen, dass sensorische Information kontinuierlich bis zur Überschreitung eines Entscheidungskriteriums gesammelt wird. Dieser Prozess lässt sich durch theoretische Modelle, sogenannte Akkumulatormodelle, mathematisch beschreiben. Aufgrund umfangreicher Verhaltensdaten, neurophysiologischer und bildgebender Befunde, wird angenommen, dass diese Modelle den zugrundeliegenden Informationsverarbeitungsprozess im Gehirn sehr gut abbilden. Ein zentraler Parameter dieser Modelle ist dabei das Entscheidungskriterium, weil es den Entscheidungsprozess determiniert. In dieser Dissertation werden zwei Projekte diskutiert, in denen neuronale Mechanismen des Entscheidungskriteriums von mir untersucht wurden. Zentral im ersten Projekt ist die Frage nach dem neuronalen Mechanismus der Anpassung des Entscheidungskriteriums (engl. decision threshold modulation). Ausgehend von einem biophysikalischen Modell das den erwähnten Entscheidungsprozess in einem Neuronalen Netz implementiert, untersuchte ich die Veränderbarkeit des Entscheidungskriteriums. In dem Neuronalen Netz wird das Entscheidungskriterium durch die Anpassung von Interaktionen zwischen Kortikalen Akkumulator- (engl. integrator) und Striatalen Neuronen moduliert. Unter Anwendung bildgebender Verfahren und komputationaler Modelle konnte ich zeigen, dass das Entscheidungskriterium durch die Modulation von Interaktionen, zwischen Hirnregionen, die für eine Entscheidung relevant sind, angepasst wird. Akkumulatormodelle lassen sich theoretisch aus statistischen Tests über optimales Entscheiden herleiten. Ausgehend von einem komputationalen Modell über optimales Entscheiden in kortiko- basalganglionischen Netzwerken, habe ich im zweiten Projekt untersucht, wie der Nucleus subthalamicus (engl. STN, Subthalamic Nucleus), der in diesem Modell eine zentrale Rolle bei Entscheidungen spielt, bei einfachen, perzeptuellen Entscheidungen das Entscheidungskriterium beeinflusst. Dazu nutzte ich die Möglichkeit der Tiefenhirnstimulation (engl. DBS, Deep Brain Stimulation) des STN bei Parkinson Patienten. In dieser Studie konnte ich die Modellannahmen über die Funktion des STN bestätigen. Ebenso konnte ich zeigen, dass die DBS des STN die Modulation des Entscheidungskriteriums einschränkt. Zusammengefasst zeigt meine Dissertation erstens, dass das Entscheidungskriterium durch eine Veränderung in der Kopplung zwischen Kortikalen und Subkortikalen Hirnsystemen moduliert wird und zweitens, das diese Anpassung durch Signale des STN beeinflusst wird.