Michael Arbib
(1940-)
Michael Arbib is a theoretical neuroscientist and computer scientist who studies the brain from a computational perspective.
Arbib earned his Ph.D. in Mathematics under
Norbert Wiener in 1963.
Arbib founded the Department of Computer and Information Science at UMass Amherst in 1970. With his colleague David Kaplan, Arbib wrote a lengthy defense of "cooperative computation" as a basis for understanding neurolinguistics, which led to their 1982 book with John C. Marshall
Neural Models of Language Processes.
Cognitive scientists like Arbib see the mind as containing parts that can be analyzed as logical systems much like the parts of digital computers. This is a very powerful metaphor, but
man is not a machine in the Newtonian sense, as the
reductionist behaviorists assumed, and the
mind is not a computer, although like a computer, it is an
information processing system which acquires, creates, stores, and manages the information needed to guide the actions of its body.
Arbib's 1983 Gifford Lectures (published with Mary Hesse as
The Construction of Reality, 1986) considered the question of
free will in the context of whether computers could be designed to have free will.
Chapter 5 of the book, whose primary author was Hesse, argues that quantum
indeterminacy in the brain is only a necessary and not a sufficient condition for free will.
A factual question arises here as to whether quantum indeterminacy in the brain is at the right level to generate probabilities at points at which one would wish to say "choices" are made.
(The Construction of Reality, p.95)
But Hesse/Arbib then says this is no help for free will. "It is just as difficult to ascribe responsibility to quantum jumps as to a sequence of predetermined states."
This is the
standard argument against free will.
Hesse/Arbib considers
Daniel Dennett's
two-stage "mechanist" model, as they call it. "Of all recent writers on this tangled subject," they say, "Dennett perhaps comes nearest to a sympathetic reconstruction of the [libertarian's] case." (p.96)
But they say three things are wrong with Dennett's model (p.97).
- It begs the question of whether determinism or flipping a coin are the alternatives at the point of decision or at any other point.
- The model falls into an infinite regress about how to account for early "decisions" about what considerations to allow into deliberation and how to weigh them.
- Free will is not well simulated by a last-minute flip of the coin, which seems to abrogate rational deliberation.
Hesse/Arbib do not fully appreciate Dennett's idea that randomness can be limited to the generation of
alternative possibilities. Dennett's problem, as he himself admits, is the
location of the randomness at a place and time (relative to the decision) that does not compromise
responsibility.
And in the second edition of his book
Brainstorms,
Dennett came to accept the two-stage model of free will first suggested by
William James and reinvented by
dozens of other philosophers and scientists in recent years.
Computational Neuroscience
Arbib earned his Ph.D. in 1963 from MIT, where he worked under
Norbert Wiener. He met
Warren McCulloch and
Walter Pitts, who in 1943 had imagined that neurons in the brain were solving problems in propositional logic. It was a few years before digital computers were actually built (the COLOSSUS in 1945 and ENIAC in 1946), but the idea of neurons as logical processors led to the idea that the brain itself is a digital computer, setting in motion work on cybernetics, cellular automata, and artificial intelligence, even to the extreme notion that the whole universe is a computer (see
Seth Lloyd).
For Teachers
To hide this material, click on the Normal link.
|
