Ai Dreams Forum
Robotics => Robotics News => Topic started by: Tyler on February 17, 2008, 07:07:44 am
-
Challenging Connectionism
Connectionist theory posits that there are no centralized controllers in
the brain. This fundamental aspect of connectionism leads connectionists
to reject cybernetics or control theory as an underlying principle of
brain function. Or, so says Asim
Roy, who has published a new paper challenging this position. In Connectionism,
Controllers, and a Brain Theory (PDF format) Asim argues that
connectionist systems actually are based on standard control theory.
Whether it's a simple back-propagation neural network or the human
brain, Asim argues that traditional controllers are at work.
Connectionists are not amused. Some say Roy is missing the point, that
connectionists are merely arguing against the idea of a homunculus-like
central controller in much the same way Dennett and others have - but not
arguing against the idea that some portions of the brain could control
other portions of the brain. Roy thinks his theory works
around limitations in current connectionist theories, pointing the way
to better autonomous learning systems. For more see the recent
story from the Arizona State University.
http://robots.net/article/2468.html