Decision-Making in Systems Neuroscience: Neural Dynamics and Algorithms

This lecture delves into the complexities of decision-making within the framework of systems neuroscience. It begins with an overview of the course's objectives, emphasizing the importance of understanding the relationship between neural dynamics and behavior. The instructor revisits David Marr's three levels of analysis—computational, algorithmic, and implementation—highlighting their relevance in studying decision-making processes. The lecture discusses the challenges of decision-making, including the credit assignment problem, where individuals struggle to link actions to outcomes. The instructor introduces perceptual and value-based decision-making, illustrating how sensory evidence and subjective values influence choices. Key experiments, such as the random dot motion task, are presented to demonstrate how neural circuits accumulate evidence for decisions. The role of dopamine neurons in encoding prediction errors and their implications for learning are also explored. The lecture concludes with a discussion on reinforcement learning algorithms and their potential neural implementations, providing a comprehensive understanding of the neural basis of decision-making.