PROUDfOOT | 305
B is the operator who works the paper machine. (In order that he should be able to work it fairly
fast, it is advisable that he be both mathematician and chess player.) Two rooms are used with
some arrangement for communicating moves, and a game is played between C and either A or
the paper machine. C may find it quite difficult to tell which he is playing.
(This is a rather idealized form of an experiment I have actually done.)
Turing’s ‘little experiment’ is a trial to see whether or not C (the interrogator in this game) has
the ‘temptation to imagine intelligence’ in the paper machine.^16 His words make it clear that the
game tests the observer rather than the machine. Whether or not the machine is intelligent is
determined in part by C’s response; for example, if C can ‘predict its behaviour or if there seems
to be little underlying plan’, the machine is not intelligent.
If intelligence is a response-dependent concept, the Turing test is not strange, as on the
behaviourist interpretation. The proper goal of the test is to test an observer’s responses and
so it is not unnecessarily circuitous. That the concept of intelligence is a response-dependent
concept also explains why the computer is to ‘do quite a bit of acting’ and even use ‘tricks’. The
more the machine prompts the interrogator to anthropomorphize, the more the interrogator
will succumb to the ‘temptation to imagine intelligence’ in the machine. This makes sense if
Turing’s experiment tests the interrogator’s response to the machine.^17
Several modern scientists and philosophers claim that the Turing test can be useful only
until we possess a scientific theory of cognition. Typically they assume that cognition consists
in computation and so our real need is for a computational theory of cognition. Turing could
not have agreed—not because he was a behaviourist for whom thinking is nothing more than
‘thinking’ behaviour, but because in his view the concept of intelligence is ‘emotional’ rather
than ‘mathematical’. According to a response-dependence theorist, the concept of colour is very
different from the concept of electromagnetic radiation, even though electromagnetic radiation
is the physical basis of colour. Likewise, if intelligence is a response-dependent concept, the
concept of intelligence is very different from the concept of computation, even if brain processes
(implementing computations) form the physical basis of ‘thinking’ behaviour. Attempting to
explain intelligence in terms of computation is to confuse a response-dependent concept with
a response-independent concept.
Anything goes
Several influential artificial intelligence (AI) researchers take the view that intelligence is ‘in the
eye of the observer’.^18 This stance is in part a reaction to what many in AI see as ‘speciesism’ (or
‘exceptionalism’ or ‘origin chauvinism’). This is the attitude that human beings are distinctive,
because they have a soul or biological brain, or are ‘natural’ rather than ‘artificial’. This attitude
unfairly discriminates against machines, its critics claim. However, the view that intelligence is
in the eye of the observer is also problematic. Human beings have evolved to find intelligence in
unlikely places: we hear voices in the wind, see faces in clouds or the moon’s surface, and detect
personalities in Furby toys. In his song ‘Anything Goes’, Cole Porter wrote ‘The world has gone
mad today And good’s bad today, And black’s white today, And day’s night today’. If intelligence
is in the eye of the observer, is stupid intelligent in today’s AI?
Response-dependence theorists do not claim that anything goes, however. They say that an
object is yellow if (and only if ) it looks yellow to normal observers in normal conditions. If an
observer has a brain injury affecting her colour perception, she is not a normal observer; and