Page 125 - Contribution To Phenomenology

118 TOMNENON

of reasoning, and reasoning itself in mathematical and formal logical
terms, it seemed quite natural to think of machines as duplicating what
made mental processes a form of intelligence. Hence the name "artificial
intelligence," usually shortened to "AI" and applied not only to the man-
made systems of machines and programs, but to a whole approach
governed by the model of reasoning as symbol manipulation according to
formulae or algorithms both in man and machines."^ Cognitive science
emerged as a distinct field by gathering together researchers from various
disciplines around this model of intelligence or cognition as symbol
manipulation according to pre-set, programmed rules.
Within about the last six or seven years, however, there has emerged
a new way of thinking about these issues, one that can be seen both as
a competitor within the field of cognitive science or "classic AI," and in
another sense as a possible successor to the project of cognitive science
as a whole. It is a competitor in the sense that it offers a competing
model of information processing that could change the way systems and
machines are constructed to perform cognitive tasks, especially those that
were difficult or impossible to accomplish with classic computers. If
successful, it would be a successor in that it could be seen as displacing
the classic project of explaining all cognition in terms of rule-governed
symbol manipulation. Even so, most versions of Connectionism at this
point still share with classic AI some of the same basic assumptions
about the possibility of drawing important conclusions about human
intelligence from non-human information processing systems, although they
tend to emphasize the reverse direction more strongly, i.e., they think
that one can derive at least a few clues from human cognition about how
to construct expert systems that would be better able to accomplish the
kinds of things that human beings do well, but classic machines do not.
These include such tasks as pattern recognition, especially where all of
the relevant parameters are not spelled out in advance; dealing with
vagueness; and handling multiple-constraint problems with indefinite
outcomes.
Connectionism is a new way of thinking about things that is at least
indirectly and partially inspired by the model of the human brain as a

philosopher) like the formulae of formal logic." (1-2)

* A phrase chosen here simply for the sake of alliteration. It is not intended to
imply any particular stance on the recently popular question as to whether such a model
reflects a specifically gendered, i.e., male way of thinking or not.