As far as I know, there are only two arguments that attempt to show that no computer-controlled system could possibly have genuine thoughts or feelings. One appeals to Godel's Incompleteness theorem. What the argument overlooks is that, in order for a system to be capable of talking, proving theorems and doing other intelligent things, the microprocesses controlling it do not themselves also have to be capable of those intelligent things. Our being able to prove theorems doesn't require tiny theorem-proving processes inside us. So there is no reason to think that microprocesses insude us that underlie our theorem-proving are themselves capable of even elementary arithmetic. Since Godel's theorem applies only to systems that are capable of elementary arithmetic, there is no reason to think it applies to the microsystems that make us up.

The other argument is Searle's famous Chinese Room argument. There is an excellent refutation of it in Daniel Dennett's Consciousness Explained, which doesn't actually explain consciousness itself, but does a lot of good all the same.

Robert Kirk. Department of philosophy, University of Nottingham, England.

My short reply is: Yes, they will.

But there is more to consciousness than being able to recreate images of sheep and count them. We are presently trying to bridge the very difficult 'explanatory gap' between matter and mind, and determine what a material system needs for it to have to have consciousness. When this hap has been bridged, then the extra additions to machines that would give them consciousness will become clear.

I would like to suggest that there are two aspects needed by a system for it to be conscious. Firstly, it must have a competitive process going on between its inputs, so as to allow for the consciousness of new inputs to occur and for disambiguation of scenes, etc. This needs time for continued activity so that the competition can be properly carried out. Then secondly, the new inputs which have been attended to need to be 'filled in' with content arising from relations to past memories, so leading to a 'Relational Mind.' These two features – competition and relationality – seem to lead to properties of processing similar to those we experience initially as 'raw feels' an then, shortly thereafter, as more complete experiences. In addition, there would further be required a sense of self and of the value of inputs. These two extras can be included by relational memories based on specific memories and of rewards received. So this (very schematic) approach can, I claim, lead to experience in the machine possessing it, which would (after a certain amount of learning had occurred, as in a child) begin to appear as if there was consciousness. I expect the present research along these lines to lead to conscious machines some time in the middle of the next century. However, there are enormous problems of packaging – of providing enough neurons, sensors, and effectors for the machine to be able to get around effectively. So, to begin with, there will be 'niche-conscious' machines, developed only in a very special niche and having value in being able to function in that niche as conscious (and intelligent) systems. Only somewhat later will there be machines beginning to approach our own complexity and flexibility 0 and I do not expect that for centuries, due to the technological problems being faced. However I may be too pessimistic in this regard – it may be sooner than I think!

John Taylor. Centre for neural networks, department of mathematics, King's College, London, England.

The concept of an android, a most human-like machine, raises questions about the difference between machines and organisms, particularly conscious organisms. There are some who will argue that, fundamentally, there is no difference, that anything an organism can do can be completely explained in mechanistic terms and could be feasibly replicated by a machine such as a computer or robot. Others would hold that organisms are quite different from machines, and not because of any mysterious addition, or 'vital force,' but due to the highly context-dependent nature of the parts and whole.

The traditional mechanistic concept is based on the idea that the universe is made up entirely of little independent and inert chunks of matter in motion. A 'billiard ball' universe, if you like. We can understand fully the functioning of the individual parts of a machine by themselves, and as unrelated to any other part. This can then lead to a full understanding of the functioning of the entire machine. This reductionistic approach is fine for machines but ultimately fails as a full description of everything in reality. For instance, modern physics has shown that matter at the most basic level does not exist in isolation and so is not without influence, to and from its environment.

A cell in the body of an organism, and a neuron in a brain, also do not exist in isolation and are influenced by, and have an influence on, other cells, and the organism as a whole. We can, of course, examine the behavior of a single cell out on the laboratory bench, separated from its usual organic environment, and while this may be a useful source of information, it does not give us a complete understanding of how that cell functions within the organism itself. The parts of an organism are fundamentally relational from the start. This means that an atom will behave as it does in some way because of its part in a molecular system. A molecule in a cell will behave as it does in some way because of its part in that cellular system, and similarly for the behavior of a cell as part of a larger organism such as an animal. Further, an animal behaves as it does in some way because of its part in its ecosystem.

The mechanists have a great faith in their approach, as so much of the world can be explained and modelled this way. Critics, equally, have great faith in their view that, while robots may move about in clever ways, we never feel that they are in anyway alive, and no matter how sophisticated computer hardware and software systems can be, we are hard put to claim that they have even a modicum of conscious experience. No doubt it will be many years before the debate is resolved, if at all.