- seCtIon FoUR: eVoLUtIon
representations, what could it be doing? This is a challenge for embodied cogni-
tion, and for enactive and sensorimotor theories, but there is growing evidence
that embodied, enactive, and extended processes contribute to all these activi-
ties: that performing congruent actions helps us understand action-based words
and concepts (even in ‘dead’ metaphors like grasp an idea), that we move our eyes
in similar ways when we imagine and when we see, that we incorporate bodily
stimulation into our dreams, and that we use dreaming as a chance to rehearse
and optimise the interactive hypothesis-testing of our probabilistic minds.
Interestingly, however, some of the very people who were most attracted to
nonrepresentational robotics have discovered that rather than giving robots rep-
resentations from the outset, they can be built to construct their own internal
models. In one example, a wall-following robot builds concepts about itself and
the walls it follows to construct a map of its environment, as well as a model of
itself, both of which it uses to make decisions about its behaviours by estimating
their outcomes (Holland and Goodman, 2003). But robotics researchers Owen
Holland and Rod Goodman also point out that as robots become more and more
complicated, the challenge of knowing whether an internal model is present,
what it corresponds to, and how it is being used gets greater and greater.
There have been many other important developments in AI, but the few cov-
ered here at least provide a sketchy outline to guide us when we ask whether
a machine can be conscious. This is bound to be a tricky question. How can we
know? How can we tell whether we’ve succeeded? We may get a little help from
Turing’s famous test for whether a machine can think.
‘We are not cognitive
couch potatoes idly
awaiting the next
“input”, so much as
proactive predictavores’
(Clark, 2015, p. 52)
PRACTICE 12.1
AM I A MACHINE?As many times as you can, every day, ask yourself: ‘Am I a machine?’
The idea is to watch your own actions and consider them in light of the
ideas presented here. Are you like a simple autonomous robot? Could a
machine created by humans ever do what you are doing now? If so, would
the machine feel like you do? You may discover that asking these questions
while going about your ordinary life makes you feel more machine-like.
What is going on here?
If you find an inner voice protesting ‘But I am not a machine!’,
investigate who or what is rebelling.THE TURING TEST
Turing’s classic paper of 1950 begins, ‘I propose to consider the question “Can
machines think?” ’ (p. 433). He dismissed the idea of answering it by analysing the
terms ‘machine’ and ‘think’ as no better than collecting a Gallup poll of opinions,
and proposed instead to base a test on what machines can do.