Å®ÉúСÊÓƵ

The Muse In the Machine: Computers and Creative Thought by David Gelernter,
Fourth Estate, pp 211, £16.99

Ask poets, artists, composers or novelists to describe how they work,
and they will reply in ways such as ‘the ideas rise up and overpower me’,
‘I am guided by my feelings’, and ‘I become absorbed in my own world’. The
reports of creative people are usually phrased in terms very different from
those of traditional cognitive science, with its emphasis on problems, goals
and plans.

For the past three decades, cognitive scientists have sought the common
ground between computers and thought, concentrating on those human-like
tasks most easily carried out by computer programs, such as proving theorems,
understanding language and solving problems. In considering whether machines
can think and whether the computer can be a model of the mind, cognitive
scientists have paid too little attention to unfocused dream-like thinking
and the interaction between emotion and cognition.

David Gelernter is concerned with just these issues. He has developed
a general theory of cognition and creativity. The essence of his theory
is that we have a spectrum of thinking, from ‘high focus’ analytic thought,
where the mind abstracts common features from memories to form mental concepts,
to ‘low focus’ thought, characteristic of daydreams, where whole episodes
from memory are blended and linked by a common flow of emotion.

According to Gelernter, emotions are inextricably tied up with body
states. A mind without a body could never dream. Nor could a computer –
and ‘a computer that never hallucinates cannot possibly aspire to artificial
thought’. Low focus, affect-linked thinking provides the solution to ‘the
most significant unsolved problem of cognition’, that of thinking using
analogies and creativity. Creativity, Gelernter claims, boils down to the
discovery of new analogies, which occur when one thought sparks off another
one related to it only by a deep bond of emotion, he says.

To substantiate his claims, Gelernter summons up quotations from the
Romantic poets, the Scriptures and Freud’s accounts of dream-thought (although,
inexplicably, he does not refer to Freud’s only direct account of the creative
process: Creative Writers and Daydreaming), in an attempt to show a connection
between creative daydreaming, childhood memories and ancient thought. We
should not try to interpret ancient texts in cognitive terms, but rather
see them as expressions of experience woven with the thread of emotion.

Gelernter’s central notion of a spectrum of thought is interesting and
provocative. He sketches possible cognitive mechanisms for high and low
focus thought and gives a clear account of the process of slipping from
controlled goal-directed thinking into a dream state.

But as a general explanation of creative thinking, it just does not
wash. Creativity does not arise only from low focus thought. The psychiatrist
Albert Rothenberg carried out numerous interviews with creative people and
found good evidence for some entirely cognitive and controllable creative
processes, such as ‘Janusian thinking’ (the ‘simultaneous conceptualisation
of opposites’ named after the double-faced Roman god of doorways) and ‘homospatial
thinking’ (actively conceiving two or more discrete entities occupying the
same space).

Nor do daydreams and low focus thoughts appear to be held together just
by emotion. A sound (remembered or external), a colour, an object or a word
can be the link between one dream episode and the next. Thoughts may often
evoke emotions, they may also be influenced by emotions, but there is no
good evidence that emotion is the indispensable force that binds the creative
imagination. At best, Gelernter’s account could be a step towards a more
complete theory of creativity which incorporates a thinker’s alternation
between illuminative reverie and the more deliberate evaluation and exploration
of ideas.

It is hard to feel much affection for this book. Its tone is belligerent
and smug. Much of it is given over to an attack on cognitive science for
attempting to build disembodied minds, yet it does not acknowledge research
in ’embodied cognition’ such as that by the Artificial Intelligence Laboratory
at the Massachusetts Institute of Technology described in ‘Birth of a human
robot’, (New Å®ÉúСÊÓƵ, 14 May), which attempts, just as Gelernter does,
to understand the cognitive unity of mind and body. And ultimately it is
dangerous, in calling for science to withdraw from the study of the mind
and ‘let the National Science Foundation empanel a committee of principled
poets to study this problem’.

The irony is that Gelernter is himself a scientist (a designer of artificial
intelligence software), and the greatest merit of the book is its well-written,
scientifically stated and testable thesis.

Mike Sharples is a senior lecturer in the School of Cognitive and Computing
Sciences, University of Sussex and an author of Computers and Thought: A
Practical Introduction to Artificial Intelligence (MIT Press, 1990).