Workshop 2: Cognitive Neuroscience (December 10-14, 2012)
Organizers: Peter Dayan, Michael Shadlen, Eric Shea-Brown, and Murray Sherman
Cognitive neuroscience presents superb opportunities for mathematical contributions, especially in connecting different theoretical and experimental frameworks. On the experimental side, methods ranging from single-neuron recording to human behavioral tests are flourishing, and mathematical models are beginning to suggest how one leads to the other. Rigorous theoretical treatments from microeconomics are often applied, including Bayesian estimation and optimization, but details of how they might be implemented in stochastic, dynamic neural circuits have only recently been proposed. By bringing together experimentalists and theorists working on different levels, a workshop will move the field closer to a long-held goal of understanding and predicting behavior in increasingly rich cognitive tasks.
Each day will feature a different theme, as described in more detail below, and will emphasize both work at the level of algorithms and phenomena, and at the level of implementation by circuits of spiking cells.
The organizing committee suggests that this workshop include a select group of students and post-docs as participants. Time could be made available each day for brief student/post-doc presentations (short talks and/or poster sessions), and the day capped by evening sessions that encourage interactions among the students and post-docs. In addition, a student or postdoc will be chosen to be a "reporter" for each day to summarize the day's events and highlights. These will then be assembled into an overall workshop report that could be published by pre-arrangement with an appropriate journal.
Attention. Where are attentional effects "generated," how are they coordinated across multiple brain areas, how is attention fed back to earlier levels of sensory processing, and what are the underlying mechanisms at the level of circuits of spiking cells?
Decision making. How are diverse sensory and task cues integrated over time and combined into a "single" decision signal, how are decision rules applied to this signal, and what is the role of dopamine and other modulators in this process?
Coordination of neural circuits. Under different behavioral constraints, different brain areas form cooperative units. What is the role of thalamocortical and basal ganglia circuitry here? Nascent physiological work is in need of a theoretical counterpart, both to reveal how signals are gated and amplified, and to compare the performance and efficiency of different possible mechanisms and network architectures.
Reinforcement learning. Complex tasks require learning and updating of rules that relate reward to action in changing environments. What algorithms can perform this updating optimally, in the face of uncertainty about rewards and sensory cues? What neural circuitry can implement these algorithms?
Accepted Speakers
- Rafal Bogacz, Computer Science, University of Bristol
- Matthew Botvinick, Neuroscience Institute and Psychology, Princeton University
- Anne Churchland, Cold Spring Harbor Laboratory
- Sophie Deneve, Group for Neural Theory, LNC, DEC, ENS
- Michael Frank, Cognitive, Linguistic & Psychological Sciences, Brown University
- Stefano Fusi, Center for Theoretical Neuroscience, Columbia University
- Ray Guillery, Anatomy, University of Wisconsin
- Okihide Hikosaka, National Eye Institute, NIH
- Phil Holmes, Program in Applied & Computational Mathematics, Princeton University
- Sabine Kastner, Psychology, Princeton University
- Yael Niv, Psychology, Princeton University
- Alex Pouget, Brain and Cognitive Sciences, University of Geneva
- Antonio Rangel, Humanities and Social Sciences, Caltech
- Misha Tsodyks, Department of Neurobiology, Weizmann Institute of Science
- Marty Usrey, Center for Neuroscience, UC Davis
- Xiao-Jing Wang, Department of Neurobiology and Kavli Institute for Neuroscience, Yale University School of Medicine
