Rockwood Memorial Lecture
4:30PM, Monday, April 4, 2011
San Diego Supercomputer Center Auditorium
Host: Gert Cauwenberghs
There is today a wide spread belief that the brain can be reverse-engineered by an exhaustive reductionistic program such as connectomics: Once the neurons and their connections are fully described, knowledge of function will follow directly. Over the past two decades the Institute of Neuroinformatics has contributed to that philosophy by characterizing the connectivity of visual neocortex; by proposing how there current circuits of the superficial layers could implement rich computational primitives; and by engineering neuromorphic electronic systems that emulate those circuits. But these research programs stumble on the crucial questions of how the cortical circuits are intelligently constructed and configured. So, in recent years we have begun exploring a synthetic approach to these questions by simulating the methods of self-construction used by Biology to evoke the cortical circuits from only a few precursor cells. The notion is that the organization and function of neuronal circuits may be better and more efficiently understood in terms of the dynamics of their unfolding developmental process than by measuring their adult connection matrix or ‘connectome’. This talk will describe steps in that direction of research.