The workings of the human mind have intrigued philosophers and psychologists
for centuries. Conflicting theories jostle for consideration in the specialist
literatures: to biologists, ‘mind’ is defined purely in terms of neurophysiological
processes; to psychoanalysts it is a seething mass of unconscious instinctive
drives and unresolved repressed conflicts; to behaviourists it is an inexplicable
black box, the hyphen in the stimulus-response equation. Out of this jungle
of competing ideas, Margaret Boden offers a clear and compelling case for
the value of viewing the mind as a computational machine.
She is the author of seven books, including the ground-breaking and
enormously influential Artificial Intelligence and Natural Man (1977). Her
most recent book, The Creative Mind, was published last year and received
considerable praise. Bracketing her work with that of scientists such as
Richard Dawkins and Stephen Hawking, one reviewer described her as a ‘charismatic
pundit’ whose writing is ‘both more readable and more literate than the
average new novel’. Others heralded her as ‘a strong voice fighting obfuscation’
and as ‘one of the world’s best commentators’ on minds and machines. But
when I describe her as an expert in artificial intelligence, she protests.
‘I don’t give a damn about computers. I’m interested in the human mind,
not computers . . . I think I would probably refer to myself now as a ‘computational
psychologist’, by which I mean a psychologist who’s interested in the mind
and who uses computers as a way of thinking about the mind. I see the mind
as a representational system, and think of neuroscience in computational
terms, so the question becomes: ‘what jobs are these cells doing, and how
is it possible, given that these jobs are done, for a particular behaviour
to result?’ rather than asking: ‘what are the biochemical processes at work?’
I think AI, along with physiology, is the most promising way of thinking
about the human mind.’
Advertisement
Both lay people and some academics (the philosopher Hubert Dreyfus,
for example) have criticised the development of AI as fundamentally misguided.
They see AI as dehumanising and ideologically pernicious, undermining human
agency and responsibility, and presenting a travesty of human potential.
Boden gives such ideas short shrift. Far from the common view that machine
research tends to reveal us as mere clockwork, her belief is that AI helps
us to see how purpose and subjectivity are possible. Moreover, it is actually
very difficult to get computers to do the sorts of things human minds can
do. The main achievement of AI, she thinks, is precisely that it forces
us to appreciate the enormous richness and subtlety of the human mind: ‘Look
at the theories people who don’t use computational methods come up with.
They’re pretty simplistic, and it isn’t until you sit down and try to put
your theory into a program form so that you can get a computer to do what
– if your theory of language (say) is correct – it ought to be able to do,
that you realise how oversimplified your theory is.’
We talk over tea at Boden’s home in Brighton. Her sitting room is lined
with books on a broad array of subjects, ranging from Gauguin to Gandhi,
oriental carpet design to wild flower gardening, travel in India and the
South Pacific to the history of glass-making. As we speak, she radiates
an infectious enthusiasm for everything she talks about, and a sheer, heady
enjoyment of life. ‘My life has been amazingly much more exciting than I
ever imagined it could be. Getting into a new field, being one of the people
who was starting it and pushing it forward almost from the beginning, has
been enormously exciting. I’ve always wanted to travel, and because of my
work I’ve been invited all over the world. . And I still think it’s fascinating
stuff; there are lots of unanswered questions, and it’s wonderful to be
one of the people trying to think up new ways of answering them.’
Boden’s first degree is in medical sciences. She had chosen to study
medicine at Cambridge, intending to become a psychiatrist because she was
‘fascinated by what the mind is and how it works’. In 1955, when she went
up, Cambridge was a very male environment: ‘For a start, there was a quota
of women in medicine of either 8 or 10 per cent (I forget which) imposed
by the British Medical Association, and in the university generally the
proportion of men to women was about 12:1.’ But, she says with a wry smile,
the preponderance of men ‘had certain advantages’ and was less of a problem
for her than the fact that most of her fellow students, because of their
privileged class background, took their place at Cambridge for granted.
‘I’d never even expected to go to university. Neither of my parents
did. My mother left school at 14 and her father was a locksmith. My father’s
father was an ostler, but my father was very bright and he went into the
civil service, on the very lowest rung, and took a law degree in his spare
time, in evening classes. We didn’t have many books in the house, just a
few of the obvious classics. When I was growing up, as far as I was concerned
universities were for other people. Nobody ever said to me, ‘You’ll go to
university one day’. In those days going to university at all was a big
deal for a woman – let alone going to Cambridge. When I got there it was
an absolute liberation. It was like being born anew . . . Cambridge was
like coming home for the first time in my life. So many doors opened – intellectual
doors, social doors, cultural doors.’
After gaining a double first (in medicine and philosophy), she was offered
a job teaching philosophy at the University of Birmingham, and had to make
a quick decision over whether or not to pursue her medical training. ‘It
wasn’t that I was running away from medicine,’ she says, ‘but at that time
there was very little knowledge of neuroscience and psychopharmacology and
virtually nothing could be done for psychiatric patients. This was a problem
for someone who intended to become a psychiatrist. I had nursed in a mental
hospital just outside Cambridge, so I knew what a ward of patients in a
‘Largactil coma’ looked like. If neuroscience had been as advanced as it
is now, I might have stayed in medicine.’
After three years teaching philosophy, she left Birmingham for Harvard
University and was awarded a doctorate for the first non-empirical – purely
theoretical – psychology thesis that Harvard had ever allowed. Her doctoral
thesis formed the basis of her first book, Purposive Explanation in Psychology
(Harvard University Press, 1972), which explores the way in which the concept
of ‘purpose’ is used, the theoretical function it serves, and how it could
be understood in computational terms. In everyday life we usually explain
our own and other people’s behaviour in terms of purposes – goal-directed
behaviour driven by motives or intentions. Most psychologists also use this
concept. Given the importance of purpose in psychology, it seemed vital
to many writers to face the challenge of modelling it in computer terms.
‘Basically all my ideas were there, embryonically, in that first book,’
she says. ‘My second book, the one that made the big splash, was actually,
in my own mind, just an extended footnote to the first.’
Although she was ‘always’ interested in the human mind, Boden remembers
vividly how her fascination with computer models of mind first began. Newly
arrived in the US for her doctoral research, back in the summer of 1962,
she wandered into a bookshop, browsed along the shelves and picked up a
book by George Miller, Eugene Galanter and Karl Pribram, Plans and the Structure
of Behaviour: ‘Just leafing through it in that bookshop, I read enough of
it to change my life. It was the first book that tried – in a very crude
and speculative way – to apply the notion of a hierarchically structured
computer program to the whole of psychology: animal and human, normal and
abnormal, memory, perception, language, personality and even hypnosis. And
it was a milestone, not only in my own life, but in the history of cognitive
Գ.’
She writes as she speaks – with panache: ‘I think it’s part of our duty
as academics to try to communicate our ideas in a lively way.’ Her books
are lucid, direct, full of zest, and liberally scattered with quotations
from writers as diverse as Lewis Carroll and Gertrude Stein, while her arguments
are illustrated with unexpected descriptions of the theological beliefs
of the Abbe de St Pierre, or the flight path of the hoverfly.
Her second book, Artificial Intelligence and Natural Man, was written
when her two children, Jehane (now an art student) and Ruskin (an aspiring
lawyer) were both under seven. Engagingly well-written and very accessible,
it is also consciously anti-sexist: female characters people her anecdotes,
the female pronoun is used as a generic, and programming is described using
traditional feminine activities (such as knitting or baking) as parallels.
‘Women are too often invisible in academic life. The things that we do are
downgraded because they are women’s work. When I compared programming to
knitting I envisaged all these men having to run off to their wives and
girlfriends to ask what ‘cable’, ‘purl’ and ‘plain’ meant. Women are always
having to put up with books using examples drawn from car engines – so I
thought: ‘This time, let the men jolly well go and ask!”
Boden is eloquent in describing the gender-related difficulties she
has encountered in her career: ‘I remember the very first paper I gave was
to the Aristotelian Society and I was very young and terrified. It was on
explanation, and I said, among many other things, that to be an explanation,
something doesn’t have to be true. Certainly some explanations are true,
and sometimes, for example in science, it’s important that they be true;
but it isn’t a defining feature of an explanation that it be true. And I
gave various examples. When I’d finished my paper, the chairman, who was
an aged professor of philosophy – male, of course – didn’t even say a perfunctory
‘Thank you, Miss Boden’. He said: ‘Just like a woman! Truth isn’t important!’
I was so flabbergasted, I didn’t say anything.
‘Even now, I’m not sure what I could say, because if you’re too heavy
they regard you as an aggressive hysterical woman. I mean, you can’t win
as a woman. And women academics have all sorts of responsibilities that
men don’t have, and even if you’re on sabbatical, you still have to do the
shopping and buy the lavatory paper. When you have a baby, you have to force
yourself to work, and you don’t ask yourself, ‘Do I feel like it?’, because
you never feel like it, because you’re always tired. As a mother you miss
out on gossip with your colleagues because you have to rush home to relieve
the babyminder, and that means you miss out on a lot of intellectual stimulation
and academic exchange. For years I wasn’t even able to get to research seminars
at my own university because they were at 5 pm and I had to be home for
the children.’ She admits that life might be easier for young women academics
today, but hoots derisively when I mention the alleged existence of the
‘New Man’.
Her most recent book, The Creative Mind, was published in Britain in
late 1990 by Weidenfeld & Nicolson, with a paperback from Cardinal due
out this February. The book is a challenge to those who refuse to believe
that AI can teach us much about distinctive human processes like imagination
and creativity, however much it may have to offer for cognitive psychology
(memory, language, vision, problem solving). Her point is not that computers
can be creative, but rather that there are aspects of human creativity which
we can begin to understand through the (not always very successful) attempts
to build computer models of creativity. Creativity, she says, involves the
exploration and transformation of conceptual spaces. And the notion and
structure of a conceptual space – as well as its various possible transformations
– can be described using computational concepts.
It is difficult to convey the tone of The Creative Mind: erudite yet
accessible, clear without condescension, peppered with quizzes to engage
the reader but neither gimmicky nor patronising. She’s pleased when I mention
her use of humour in the book – the occasional wry comment, the ironic turn
of phrase. One of the book’s strengths lies in the detail, specificity and
range of the examples of creativity upon which she draws: from Mozart’s
symphonies to jazz improvisation, from the sources of Coleridge’s inspiration
to Kekule’s discovery of the structure of the benzene molecule. ‘And dressmaking,’
she adds firmly (she is a keen dressmaker herself). ‘Dressmaking is an underrated
creative art.’
For those determined to view creativity in terms of divine inspiration,
intuitive insight, or inexplicable genius, computer modelling of creativity
is deeply suspect. Boden is concerned to address these sceptics: she discusses
their arguments, and points out that computer art can gain admiration, and
computer music applause – particularly if no one knows they were composed
by machine. A picture hangs over the desk in her study: a pleasing line
drawing of three acrobats, one pirouetting, the others delicately poised.
‘Lots of people admire that, and ask innocently who drew it. It’s computer
generated by Aaron, a program consisting of a few hundred rules on artistic
style. Aaron has generated thousands of different drawings (some of them
have been displayed in the Tate and other art galleries, and not just for
their curiosity value). Although a human artist, Harold Cohen, wrote the
program, he couldn’t possibly anticipate all the permutations in which the
code could be applied.’
But is Aaron creative? ‘Aaron isn’t creative, in the sense that all
its drawings could have been done before, using the same program. It’s like
a human artist who has found a style and is sticking to it. A truly creative
drawing program should be able to switch to a new content or generate a
new style. It should be able to think (like Picasso, perhaps) ‘I’m bored
with acrobats – I’ll draw Minotaurs instead,’ or ‘I’ll explore a new style
– I’ll try drawing limb parts as straight-sided geometrical figures, and
see what happens.’ To do that the program would need a way of reflecting
on its own knowledge, and would have to be able to construct, inspect and
change various maps of its mind. But precisely by its failings Aaron has
taught us something about human creativity.’
To be genuinely creative, something has to break, or transform, the
frame of what has gone before. Kekule’s sudden revelation of the molecular
structure of benzene is a case in point. What came to him, as he sat dozing
by the fire, was the image of a snake biting its own tail, which gave him
the idea that the structure could be ring-like instead of string-like. In
that leap of imagination he created the possibility of a whole new science:
aromatic chemistry. It is taken for granted in many of the accounts of Kekule’s
fireside vision that he could see the analogy between a long row and a snake
biting its tail; but to psychologists, the perception of that connection,
and the subsequent creative breakthrough, require explanation.
Over the past decade there has been a dramatic explosion of work in
AI, particularly within the study known as ‘connectionism’. This involves
parallel processing models, broadly inspired by ideas about the brain and
made up of a large number of very simple interconnected units, where the
activity of one unit can excite or inhibit the activities of its neighbours.
Boden is sceptical of some of the claims made for connectionist models:
‘People say they’re more ‘humane’ and less ‘mechanistic’ (than those of
traditional AI), which is just silly because although they don’t work through
step-by-step sequences in the way a traditional computer program does, they
work according to differential equations, and I don’t think there is anything
particularly warm or cuddly about a differential equation.’ Hybrid systems
which combine the complementary insights of both traditional AI and connectionism
are, as she sees it, the way forward. She draws on connectionism – which
makes it easier than in traditional AI to address questions of analogical
thinking and generalisation – to conjecture how Coleridge came up with the
imagery for his epic poem The Ancient Mariner.
‘It’s quite clear that Coleridge was reading an extraordinarily wide
range of sources and that they somehow came together in his mind as new
images. But the psychological question is how, in theoretical terms, it
was possible for these different sources to become associated with one another
and to give rise to similar but nonetheless novel images. Connectionist
systems give us the significant beginnings of an understanding about how
this can happen. When people say that this type of creative analogical thinking
is typical of people and you can’t possibly do that in a computer, they’re
thinking about a traditional sort of computer. So it’s not surprising that
people are excited about connectionism – but it’s just a part of AI.’
The book on creativity is one she’s been wanting to write for years.
Now that it is finished, she’s reluctant to commit herself too soon to another
project. She hopes to have more opportunity to travel – the Fijian war clubs,
the statue of the Hindu elephant god Ganesha, and the Hawaiian statuette
in her study bear witness to some of her many trips abroad. And she will
continue to be busy in her work at the School of Cognitive and Computing
Sciences at the University of Sussex, England – a school of which she was
founding dean in 1987 – ‘one of my proudest achievements’. Besides providing
degrees in ‘straight’ computer science, it offers interdisciplinary courses
demanding of students an integration of AI, psychology, philosophy and linguistics.
It is her ability to transcend disciplinary boundaries that makes Boden’s
work so important. Her capacity to speak to people from a wide variety of
specialisms on their own terms, and to offer them insights from other fields
of knowledge, make her books highly accessible, original and stimulating
reading.
Celia Kitzinger is a lecturer in psychology at the Department of Psychology,
University of Surrey.
