Title:
At the Limits of Computation: How Nature Deals with the Physical
Constraints on Computation


William B Levy, Ph.D.
Professor of Neurosurgery & of Psychology
University of Virginia
Charlottesville, VA

Abstract:
   This talk will address, from a biological perspective, constraints on
computation including time, energy, and space (size).  Because evolution
has had a long time to work out computation and communication within the
brain, there is reason to suspect that it is microscopically optimized.
It is certainly true that, at the micron and submicron level, certain
basic principles and parameterizations (e.g., the voltages, capacitance,
the signal shape) are found throughout the animal kingdom.  Sometimes the
optimizations make sense - short distance communication is analog, long
distance communication is binary, and sometimes they may be surprising.
E.g., more energy is spent on communication than on computation, but
little energy used for communication goes to improving signal-to-noise
ratio.  From the computational perspective there are, perhaps, other
counterintuitive constructions.  E.g., a special kind of randomness
actually improves computation and the energy efficiency of computation.
   The required neuroscience is minimal and will be defined in the
lecture.