Computer scientists and brain specialists have been
discussing how they could work together to speed progress towards
computers that show human intelligence and improve understanding of
the human brain. Both goals are some way off but areas where the
two fields might help each other have been identified.
Forty computer scientists and brain scientists met at a BCS
debate, organised as part of the national Foresight Programme, last
month. The meeting was held under a rule of anonymity.
One issue is that much of the computer science work on cognitive
systems has little to do with biology. Models are devised by
computer scientists but there is no reason to believe the brain
works in the same way: the fact that a system gets the same result
as a human is not proof.
The debate asked whether it was possible to build a computer
model of the brain, and what use it would be. If a machine that
matched a brain was built, mapping 10,000 billion neurons, each
with about 1,000 connections, and assumed that the function of each
connection could be represented by one bit, it would need 10 square
metres of silicon at today's electronics density rates.
Much existing work on artificial neural networks is on a small
scale and it would be hard for humans to get to grips with such a
huge machine, the debate heard. Even with a big enough machine, it
would still not reveal how the brain works. It would, however,
enable theories and models to be tested.
Considering how neuroscience could help computer science, the
debate heard that one area already seeing progress is that of
technology interfaces with humans. Interfaces between a machine and
the brain or nervous system involve implanted devices controlling
artificial limbs or stimulating muscles affected by damage such as
spinal lesions.
Another interesting area is systems that understand speech, or
intuitively search databases, working much like human memory.
Humans currently have to adapt to systems to use them; life would
be easier if machines had to adapt to human foibles instead, the
debate heard. Understanding the brain could also help in the design
of database systems handling large numbers of queries.
Computer scientists at the debate said they had been
disappointed to find that neuroscientists were unable to explain
how human memory works - they had hoped to apply it to computer
memory design. Such comments, along with the need for common
terminology, highlighted the problems of bringing the two
disciplines together.
However, there is middle ground where practical work is being
conducted on robots and other devices that help elderly and
housebound people. These show some human characteristics but do not
involve an understanding of how the brain works.
The debate concluded that computer cognition could take a
decade, a century or a millennium, but the potential rewards for
success were great.