An important set of issues in computational neuroscience centers around bridging scales of modeling and neural system description. Some formulations at the cell and circuit levels are well-developed and experimentally based, and linkages between the two levels are beginning to form - for example, theoreticians are actively seeking derivations of the mean-field population equations for N-cell network models. In contrast, system level descriptions have traditionally been viewed more-or-less as black box formulations, unconstrained in many cases by system-level neurophysiology and without support by neural correlates. We are beginning to see now, the growth and appearance of system level models with closer links to anatomy and physiology. This is partly driven by new methods of system level data collection including imaging (e.g. fMRI) and multielectrode arrays with simultaneous recording in several brain areas. This workshop will bring together neuroscientists and modelers who are developing more neurally-motivated system level treatments. It will also include researchers with cell and circuit modeling experience who seek to step-up to the "higher-level" descriptions.
This higher level modeling addresses significant questions that are difficult to formulate and correlate with neurophysiology at lower levels - for example, questions about cognitive behavioral tasks, reward, attention, adaptive control, neural representation of the environment, learning and planning of motor control. A major goal is to develop formulations that capture the essence of such behaviors and the multi-level feedback loops without confounding one's understanding by excessively detailed models. In this way one hopes to identify some principles of operation. A major challenge is to chose good model systems so that formulations include potentially identifiable variables, say that might be related to imaging data, and parameters. By including some circuit level researchers we also hope to alert system-level modelers to some aspects of lower-level description that could have consequences for or clarify some underlying assumptions of the high-level models. Examples will be drawn from motor-control systems with possible links to Parkinsonism and bio-motivated robotics, language/vocalization, predictive or adaptive sensory/cortical processing, dynamics of interconnected brain areas during cognition such as delayed match-to-sample tasks.
The mathematical areas that are expected to be strongly involved in this workshop are the same as for the first workshop.
|Monday, November 18: Scales of Investigation|
|Tuesday, November 19: Motor/Sensorimotor Control|
|Wednesday, November 20: Cognitive Function, I|
|Thursday, November 21: Modeling Strategies for Multi-Scale Descriptions|
|Friday, November 22: Cognitive Function, II|