Sunday, 1 July 2012

Turing, Doing and Feeling

Alan Turing, born 100 years ago, invented the computer and computation and helped saved Europe by decoding Nazi messages during Word War II. 

Turing also proposed a simple way to test scientific explanations of how the mind works: the Turing Test: If we can design a robot that is able to do anything and everything that a real human being can do (including speaking) – and can do it so well that people cannot even tell it apart from a real person -- then we have explained how the mind works, and the robot has a mind. 

The challenge of passing the Turing test has created a new family of sciences called the cognitive sciences. 

But what does it mean to have a mind? Turing’s robot can do anything and everything we can do, but does that mean it has a mind? Could it not be a “Zombie,” that acts exactly the same way we do, but mindlessly? 

What does it mean to have no mind? A rock has no mind. A waterfall has no mind. A toaster has no mind. And surely computers have no minds. What do all these mindless things lack?

They lack consciousness. What is consciousness? It is the ability to feel – to feel anything at all, whether it is a pinch, or a puff of air, or the sound of a distant train, or the sight of a rainbow. 

If the robot that passed Turing’s test could not feel, it would not have a mind, even if it could do anything we can do, indistinguishably from us. This is the difference between doing and feeling. It is also called the mind/body problem. 

But what about the brain? Surely if we want to explain how the mind works then the thing to study is not robots but the brain! 

Well, yes, but alas the brain does not reveal the secrets of its functioning as easily as a heart or a kidney do. The brain can do what we can do, but observing and measuring brain activity only tells us where and when things happen in the brain: not how and why the brain can do what it can do. And doing is Turing’s territory.

What about feeling? The brain basis of consciousness is under intensive study by neuroscientists: How and why does the brain feel? This is the theme of the UQàM Summer Institute on the Evolution and Function of Consciousness

What function does feeling perform in the brain? What can be done with feeling that cannot be done with just doing, and why? Feeling is a biological trait. What was the evolutionary advantage of feeling to our ancestors? What made those who felt survive and reproduce better, with the result that the ability to feel became encoded in our genetic material? Do lower animals feel – invertebrates, like snails or octopus, or even plants? Where did feeling start? and why?

And what about robots, Turing's territory? What (if anything) can they not do – what will they never be able to do – if they cannot feel? and why?

More challenging still: If a robot that can pass the Turing test does indeed feel, what is the causal role of the feeling, in the robot's internal functioning? All of us have the intuitive feeling that feeling has a causal role: I do what I do because I choose to do it. 

It feels as if my will is a kind of  causal "force." But is it? Whether we study the brain or we study robots, it always turns out that everything that either of them can do is fully explained by their internal functioning (more "doings"): There seems to be no room for further causes. Is feeling just a superfluous frill?

It is this role of consciousness as a causal force that leading specialists from all over the world, and from all fields – brain science, computer science, robotics, evolutionary biology, psychology, and philosophy – are coming to Montreal to ponder for 12 days (June 29 – July 12) in a unique interaction one hundred years after the birth of the founder of both the computer and the cognitive sciences.

1 comment:


    L'hiérarchie des TTS est décrite ici:
    Harnad, S. (2001) Minds, Machines and Turing: The Indistinguishability of Indistinguishables. Journal of Logic, Language, and Information (special issue on "Alan Turing and Artificial Intelligence")

    Tu as sauté T2 (capacité verbale) et l'insuffisance du calcul seul.

    Tu as bien décrit T3.

    Pour T4 ça continue à s'agir des actions observables, mais cette fois-ci on inclut non seulement les actions extérieures du corps intégral, mais les actions à l'intérieur de sa tête. Tout ce qu'on peut observer et mesurer dans le cerveau doit être également présent dans le robot T4.

    (Mais, d'autant qu'on peut avoir les mêmes actions synthétiquement, ça pourrait être synthétique plutôt que biologique. Si on insiste sur toute propriété biologique, alors ça devient T5.)

    Concernant l'opérationnalisation: C'est un peu prématuré, étant donné qu'on est encore si loin de T3! Mais c'est sur que la méthodologie de Turing est tout ce qu'il y d'opérationnel!

    Concernant la masturbation intellectuelle: Tu ne vois pas d'application pour un mécanisme qui pourrait réussir T3? Et comment veux tu démontrer que tu as effectivement découvert le mécanisme qui génère nos capacités si tu ne peux pas démontrer qu'il est capable de générer les capacités en question?

    C'est ça la métho et le défi de Turing.