AS a job description, 鈥渂rain builder鈥� would seem pretty hard to beat. Yet
Australian researcher Hugo de Garis has quit his position as head of the Brain
Builder Group at Japan鈥檚 Advanced Telecommunications Research Institute (ATR) at
a critical point in his work. He has just finished constructing a machine that
attempts to model the function of the human brain.
His machine, completed last month, is based on neural network technology.
Over the next few months, about 75 million artificial neurons will 鈥渆volve鈥�
inside the flexible computer chips known as field-programmable gate arrays into
about 64 000 functional modules. These will then be assembled into a (human
defined) brain. To show off its capabilities, later next year the artificial
brain will be used to control the behaviour of a purpose-built robot kitten.
Lots of purring and whirring.
Whether it will work or not is anybody鈥檚 guess. Plenty of scientists dismiss
de Garis鈥� predictions as unrealistic. But whatever happens, de Garis won鈥檛 be
around to see it. Ostensibly he is leaving because his division, Evolutionary
Systems, was axed due to the poor performance of the Japanese economy. In fact,
de Garis says, he had had enough of working in Japan.
Advertisement
His story is a cautionary tale for foreign researchers. It begins in the late
鈥�80s when de Garis, with many others, thought Japan would overtake the US to
become the world鈥檚 number one economy by the Year 2000. He also reasoned that
because of its wealth, Japan would be able to attract the best researchers on
the planet.
So, in 1992 De Garis took up a post-doctoral position at the Japanese
government鈥檚 Electrotechnical Laboratory in Tsukuba Science City. The following
year, he was hired at an attractive salary by ATR, on the strength of a
presentation at an artificial life conference of his vision of building a
brain.
ATR was set up with funds from the privatisation of Nippon Telegraph &
Telephone. (Can you imagine the Howard government proposing to use some of the
proceeds of selling off Telstra for such a purpose?) It is a focal point of
Kansai Science City, at the centre of a triangle defined by the cities of Osaka,
Kyoto and Nara. Unfortunately, just as the project was getting underway, the
bubble that was Japan鈥檚 economy burst. Many of the proposed labs were never
built, leaving ATR in splendid isolation.
De Garis went to ATR with a dream. He hoped Japan eventually would invest in
a major national brain building project that would be equivalent in size and
energy to NASA, and would rank with putting a man on the moon.
But the national project never materialised and de Garis found ATR a bitter
disappointment. 鈥淭he intellectual atmosphere was sterile,鈥� he says, 鈥渢here were
no seminars, no groups, no-one to really brainstorm with鈥攐ne person per
project鈥攁nd no synergy.鈥�
Over the years De Garis became very unhappy with Japan in general. 鈥淚t鈥檚 a
culture where criticism is taboo, hence there is no real grass-roots democracy
nor intellectual life,鈥� he complains, 鈥淚鈥檓 a passionate intellectual, so I
withered there.鈥� In 1996, in an attempt to stimulate debate, de Garis posted
some critical essays on his Japanese website. His efforts were not appreciated,
and he was nearly fired as a result.
Such complaints of isolation and frustration have been common since Japanese
laboratories began hiring Western researchers in the 鈥�80s.
At least for de Garis, the story has a happy ending, or a new beginning. He
has moved to Starlab, a privately owned institute based in Brussels. There he
will work on the next-generation, billion-neuron brain. At Starlab, other
鈥渃reative crazies鈥� are working on topics such as time travel, faster than light
travel, nanotechnology, quantum computing and DNA computing. De Garis should fit
right in.