CONSCIOUSNESS: ITS PLACE IN CONTEMPORARY SCIENCE
A briefing document prepared for the Royal Society
and the Association of British Science Writers
by Wendy Barnaby
March 1995
What is the problem and how should it be addressed? Professor
Jeffrey Gray, Head of the Department of Psychology at the Institute
of Psychiatry, introduced the meeting by speculating that the
audience might not yet realize that consciousness poses an obscure
problem. Consciousness is everything we have an awareness of:
colours and scenes around us or the feeling of pain, for example. It
is unlikely that consciousness is specifically human; it probably
goes back a long way and is very basic. It is essential for our
understanding of the world, including the entire scientific
enterprise. Yet we do not at present have any clear understanding of
how the rest of that scientific evidence produces the links which
must exist between what goes on in our brain, what goes on in our
behaviour, and the conscious experiences associated with that brain
function and that behaviour. We do not know how to incorporate the
conscious part of our existence into the scientific edifice it has
created.
Philosophers have resisted the idea that there is a problem, calling
it instead a linguistic difficulty. Neuorscientists have also
resisted, saying that collecting more data is all that is needed.
They maintain that we need to study correlations between brain
function, behaviour and conscious experiences and that will be
enough. The artificial intelligence community believes the answer
will come through studying the input-output relationships in the
kind of system we have in the head and devising computational
systems that will mimic these relationships. John Searle, the
American philosopher, has attacked this point of view (on the
grounds that it doesn't solve the problem of representational
meaning or intentionality), but it is still prominent. Some people
combine the brain correlation view with the systems approach (also
called the functionalist approach). This is what Dan Dennett has
done in his book Consciousness Explained - a mistitled book, in the
opinion of others!
In spite of these various dissenting camps, there is a growing
consensus that there is an underlying problem. The American
philosopher Tom Nagel has said that we don't have a transparent
theory of consciousness: ie a theory in the usual sense. That is,
one which allows us to see how the postulates in the theory produce
the phenomena that we can measure by experiment. What we do have at
the moment in the field of consciousness is brute correlation
between events in the brain and behaviour. What we don't have is any
account of how the one is linked to the other in a transparent way.
The problem is one of consciousness, not of understanding how the
brain is linked to the mental. We know the kinds of computations the
brain can do and how, in principle, it does them. We also know that
we do a lot of computation when we are unconscious - solving a
problem during sleep, for example. It is the problem of conscious
experience that matters, and it is a problem for scientists, not
philosophers, to solve; but not simply by the collection of data. We
need a transparent theory.
In science, we usually observe others - something external to the
observer, that is, like molecules in test tubes. We construct a
hypothesis which is meant to account for the phenomena observed. If
we had followed that path, we would have set up consciousness as a
hypothesis in other people as a way of explaining what they do. But
this is not how we have consciousness in our vocabulary at all. We
have it because we observe it in ourselves. So postulating it in
other people is making an assumption which is not based upon the
usual processes of scientific reasoning. Other people's behaviour
can be explained in terms of brain function, but this doesn't reach
or account for consciousness. The study of sleep, for example, has
revealed two kinds of sleep: slow-wave and rapid eye movement. The
various physical parameters associated with each kind of sleep gave
no hint that people dream during rapid eye movement sleep: that is,
physical measurements gave no clue to consciousness. Consciousness
has no explanatory power for others, but it is an irreducible datum
for each one of us.
A more transparent theory of consciousness would need to answer
evolutionary questions: how and where consciousness evolved and what
survival value it has. It must also answer mechanistic questions:
what is it about brain physiology which makes consciousness happen,
and how does that lead to behaviour which has survival value. These
are the tests that a future theory will have to pass.
What is consciousness for? Dr Max Velmans of the Department of
Psychology at Goldsmiths' College, London, said that most of the
psychological work done in the last three or four decades has been
to do with modelling the mind. This has often been conceived in
terms of input (eg analysis, pattern recognition, selection and
attention), storage (learning, memory), information transformation
(thinking, problem solving) and output (motor skills, speech). The
question is to what extent consciousness enters into any of those
things, and to work out the function of consciousness: that bit of
the system where consciousness is really necessary to make it work.
It is typically thought that consciousness is necessary for every
stage of functioning. But to adopt a third-person perspective
typical of science - looking at bodies from the outside -
consciousness is not seen to do anything. Furthermore, the models we
generate to explain what's going on don't need consciousness to make
them work. A neurophysiological textbook for example doesn't talk
about any gap between the synapses which only consciousness could
fill to make neural processes work. Cognitive psychology is more
concerned with the functional subdivisions of the mind and will
generate an informational processing model which describes a system
that works to enable humans to do what they do. But again, we don't
need consciousness to make those systems work. This produces a
causal paradox. From one point of view, it's obvious from our own
personal (so-called first-person) perspective that consciousness is
necessary for a whole range of psychological functions; but if we
look at things from the outside, we don't have to postulate
consciousness in other people to explain how their bodies and brains
work. Much of psychology side-steps this issue.
One way of trying to find out what consciousness is for is to
contrast processes which are conscious with processes which are not,
and look for functional differences between them. One area which has
been explored in this way is selective attention: the fact that we
can be conscious of certain things that are going on but not others
also happening simultaneously. At a cocktail party, our attention is
confined to the conversation we're having - unless someone mentions
our name on the opposite side of the room! There is therefore a
sense in which we're monitoring everything that's going on, but
we're only conscious of some of it. One typical way of studying this
is to have different messages coming simultaneously into a subject's
earphones. The subject pays attention to one message and repeats it,
but doesn't have much idea about what the other message was. What
the subject pays attention to enters consciousness; what is not
given attention, does not. Clearly then, consciousness is closely
linked to attention.
But this is by no means the whole story. It was initially thought
that consciousness was necessary for the analysis of meaning, but in
the 70s this was shown not to be true - either for simple or
possibly even for complex meanings. The fact that we analyse
individual word meanings before they enter consciousness can be
shown by reading, silently, the following sentence: "The forest
ranger did not permit us to enter the park without a permit." By the
time we have consciously grasped the meaning of this sentence, we
have already allocated different stresses to permit and permit. That
is, we have already analysed not only potential meanings of the
individual words but their particular meaning in the sentence.
Consciousness doesn't take part in the analysis - it simply receives
the result of it. In speech analysis, which is one of the most
complicated areas of human processing, we are not conscious of any
of the processing itself: we are only aware of whether we are
understanding speech, or not, which is the result of the processing.
Given this, it might be that consciousness relates most closely to
some late-arising stage of focal attentive processing. But what
might this be?
The phenomenon of blindsight may give us a clue about the relation
of brain function to the point at which consciousness arises.
Blindsight occurs where, due to cortical damage, one half of the
retinal field may be literally blind, but when people with this
damage are asked to guess, they can tell a great deal about stimuli
that have been flashed to that "blind" side (they can distinguish
horizontal from vertical lines, "Os" from "Xs" and so on). The
subject knows what the stimulus is, but doesn't know that he knows.
The best way of explaining this so far is a model of the brain in
which much of the initial information processing is modular ie
specialised for particular tasks. In order for us to function in an
integrated way, the information has to be distributed throughout the
system as a whole. Consciousness then relates to the dissemination
of information: we are conscious of something if all of us knows it.
Even with this idea, however, we know of no reason why information
dissemination can't take place in a system which is not conscious.
So there's a moral: certain questions cannot be investigated by this
kind of modelling. We can theorise about the processes that might
relate most closely to consciousness, but we can't capture
consciousness as such from a third-person perspective, ie one that
stands outside the phenomena being investigated. The conclusion is
that we need a restoration in science of the first-person
perspective. Part of the problem is philosophical.
We've convinced ourselves that the third-person perspective is the
only one to adopt for scientific work, and in doing this we have
ignored the fact that the reason we do scientific work is to explore
our first-person perspectives: our own individual experiences of the
world.
Why and how we are not zombies According to Stevan Harnad, Professor
of Psychology and Director of the Cognitive Sciences Centre at
Southampton University, nobody will ever have the faintest idea of
why and how we are not zombies. It is silly to deny that we have
conscious states: that we are aware of pain, or colour, or meaning.
To people who insist that there's only one kind of thing in the
world and it's made up of matter/energy, the further recognition of
conscious states is an error called "dualism" (assuming the
existence of too many kinds of stuff under the sun). Professor
Harnad insisted, however, that "epiphenomenalism" - a form of
dualism - is right. An epiphenomenalist accepts that consciousness
is real but cannot be explained in the usual scientific way because
it has no causal power independent from the matter and energy it
somehow piggy-backs upon.
The problem is not that we can't observe consciousness (except in a
first-person way). Quarks, too, cannot be observed, yet the theory
that postulates their existence explains observations. This is not
true about consciousness. There is a "peek-a-boo",
take-it-or-leave-it, quality about consciousness: some observations
may seem to bring us close to it, but we never capture it. This is
illustrated by the Turing test.
Alan Turing was a mathematician who drew up the abstract prototype
for the computer. His test was to see whether a mathematician could
be distinguished in any way from a person with a mind. Turing was
concerned with intelligence, not consciousness, but his test could
apply equally to consciousness. Three people take part in the test:
an interrogator in one room, and a man and a woman in another. The
interrogator's task is initially to determine which of the others is
the man and which is the woman; each can be interrogated in writing,
about anything at all, including their mental experiences, to find
out, which one is which. At some point, either the man or the woman
is replaced by a machine, but the test goes on, for a whole
lifetime, in principle.
Now Turing's question is: Would the interrogator have any rational
basis for rejecting the machine as a mindless trick, upon being
informed it was a machine, when just a second before he or she
thought it was another mind understanding the questions? Turing said
no. If a lifelong penpal is indistinguishable in every respect from
a real penpal, we do not have any basis for concluding that it's not
conscious merely because we're told it's a machine. To generalise
this test, the machine could be a robot, totally indistinguishable
from us in all of its functional interactions with the world:
optically, acoustically, and mechanically. To go even further, the
machine could be structurally and biochemically indistinguishable
from us. However, even at this level, Turing's account works equally
well whether we factor consciousness in or leave it out. The same
thing applies to evolution: the random mutations and evolutionary
selection which designed us through trial and error could not have
favoured the conscious beings and excluded the zombies because the
two are functionally equivalent and functionally indistinguishable,
and survival and reproduction are purely functional matters. We are
not mind-readers and neither is anything else in nature. Turing
indistinguishability is functional indistinguishability, and that is
scientific indistinguishability.
We accordingly have to conclude that consciousness cannot be
investigated directly. It can, however, be investigated indirectly,
and perhaps eventually cornered by a series of approximations. The
way forward is reverse bio-engineering: second-guessing what gives
creatures like us our functional capacities. We should give up on
trying to figure out why and how we are not zombies. We should
instead try to reverse-engineer a zombie which is functionally
indistinguishable from us. Although the theory which produces this
zombie will be entirely mute about consciousness, we can hope that
consciousness will somehow piggy-back on it. Later, during the
discussion, Professor Gray said that what we need is a theoretical
account of consciousness which would allow Professor Harnad to
predict at what point in the complexity of his reverse
bio-engineering consciousness would become a property of his zombie
(ie at what point the zombie ceases to be a zombie). But, of course,
if that point does exist, there's no way for us to know it (without
being the zombie), Harnad would reply.
Blindsight and memory without remembering Professor Larry Weiskrantz
of the Department of Experimental Psychology in the University of
Oxford explained that he studies patients who have lost various
aspects of consciousness. In the realm of neuropsychology, there are
patients who have very good capacities without any acknowledgement
that they have them.
Patients who have blindsight are able to process information even
though they don't "see" the information they are able to process.
The retina projects to nine or ten different centres in the brain.
The only one usually talked about is the major one, which goes to
the striate or visual cortex. If the striate cortex is damaged in
animals or humans, the half field of vision opposite the hemisphere
that's damaged is altered. Animals in this condition can still
discriminate stimuli presented to their damaged cortex, but people
in the same circumstances say they don't see anything. When we talk
about the half-field being blind, it's not literally blind in terms
of processing in the brain, because there are still nine pathways
that can carry information even after the primary one is damaged.
One blindsight subject, called G, had a severe head injury at the
age of eight and is blind in his right half-field under many
conditions. He is asked to imitate a moving spot presented to him in
his so-called blind half-field. Unless the spot is moving very
slowly, he is able to indicate with his arm the direction in which
the spot moved. He is aware that there is an object moving, but is
unable to describe it and says he cannot see anything. (This
condition is not actually blindsight because he is aware of the
object moving.) In a further experiment, he has been asked not only
to guess the direction of the movement but to say whether he is
aware or not aware that something is moving. This is so-called
first-person discourse: information about the subjective state of
awareness of the patient. What emerges is that even under these
conditions when he has no awareness, he can still guess the
direction of movement correctly 90% of the time. The parameters that
define the awareness mode have been defined but those defining the
unawareness mode are different. It is known that they extend way
outside the range of the aware system and will probably turn out to
have quite a different set of parameters. The next experiment is to
define these - which will probably take another couple of years -
and then see exactly how they compare with the awareness mode.
How does this relate to brain function? An MRI (magnetic resonance
imaging) scan on G shows that he has no striate cortex left. He
could not therefore be using undamaged remnants to see the stimuli.
At the same time, a PET (radioactive blood flow image) scan taken
while he's doing the movement tasks shows that no areas of striate
cortex light up, but other areas in the brain do. So we have a scan
of what the subject can do when he has some awareness. We need to
take a scan of when he's not aware, and then to compare the two to
see whether and how they're different. This provides a way into the
subject which tells us what structures might be involved.
A next logical stage would be intervention. If, for example, we knew
the exact neural circuit that lit up when someone reported a
conscious experience, would intervening to light up that circuit in
another, unaware, subject produce awareness?
Speakers
Jeffrey Gray
Professor of Psychology
Institute of Psychiatry
De Crespigny Park tel: 0171 703 5411
London SE5 8AF fax: 0171 703 5796
Dr Max Velmans
Department of Psychology
Goldsmiths' College
Lewisham Way tel: 0171 919 7171
London SE14 6NW fax: 0171 919 7873
Stevan Harnad
Professor of Psychology
University of Southampton
University Road
Highfield tel: 01703 595000
Southampton SO17 1BJ fax: 01703 594597
Larry Weiskrantz F.R.S.
Emeritus Professor of Psychology
University of Oxford
South Parks Road tel: 01865 270000
Oxford OX1 3UD fax: 01865 310447
Some helpful contacts
Sir Roger Penrose F.R.S.
Professor of Mathematics
Mathematical Institute
24-29 St Giles tel: 01865 270000
Oxford OX1 3LB fax: 01865 273583
Argues there is a problem but it can be solved by mathematical
logic, physics and biology. Author of Shadows of the Mind.
Dr Nick Humphrey
Darwin College tel: 0223 315151
Cambridge CB3 9EU fax: 0223 335667
Author of A History of the Mind. Argues that the crucial component
of consciousness is feeling.
Dr Michael Lockwood
Department of Continuing Education
University of Oxford
1 Wellington Square tel: 01865 270309
Oxford OX1 2JA fax: 01865 270370
Colin Blakemore
Professor of Physiology
University of Oxford
Parks Road tel: 01865 202500
Oxford OX1 3PT fax: 01865 272469
From the artificial intelligence community
Igor Aleksander
Professor of Neural Systems Engineering
Imperial College of Science, Technology
and Medicine
Exhibition Road tel: 0171 594 6189
London SW7 2BT fax: 0171 823 8125
John Taylor
Professor of Mathematics and
Director of the Centre for Neural Networks
King's College London
The Strand tel: 0171 836 5454
London WC2R 2LS fax: 0171 836 1799
Aron Sloman
Professor of Computer Science
University of Birmingham
Edgbaston tel: 0121 414 4775
Birmingham B15 2TT fax: 0121 414 4281
Ref: PR 13 (95) INFORMATION NOTE
Enquiries to:
Miss Anna Link
Science Promotion Section
The Royal Society
6 Carlton House Terrace
London SW1Y 5AG
Tel: 0171 839 5561 ext 315 22 March 1995
CONSCIOUSNESS: ITS PLACE IN CONTEMPORARY SCIENCE
On 7 February 1995 a press briefing on consciousness, organized
jointly by the Royal Society and the Association of British Science
Writers, was held at the Royal Society as part of a programme to
further the public understanding of science. The enclosed document
was prepared after the briefing to summarize key issues raised by
the speakers at the time and also to provide a list of relevant
contacts for future reference. This document does not necessarily
constitute the views of the Royal Society, and views expressed in it
should not be attributed to the Society.
The document is free of copyright and may be used without reference
to source.