X-Message-Number: 13009 Date: Mon, 27 Dec 1999 14:27:55 -0500 From: Daniel Crevier <> Subject: The downloading thought experiment References: <> Subject : the downloading thought experiment. My revival of the downloading thought experiment two weeks ago has elicited many comments. Among them was a long discussion involving mostly Robert Ettinger and Mike Perry as to whether a mere simulation can be "as good as" the real thing. Needless to say, I am on Mike's side and my answer is yes. In fact I think the downloading thought experiment, if we look at it a little more closely, provides a convincing argument to this effect. First, let us review our premisses carefully. They are : - consciousness can be explained, understood and accounted for in terms of purely physical phenomena. "Self circuits", "standing waves" or any other phenomena deemed necessary to generate consciousness must be physical in nature, and therefore describable in quantitative terms. This implies that they can be simulated by a digital computer, to any desired degree of accuracy. If they can't, then we are back to mysticism, and I won't follow anyone on these grounds. - This being the case, then it is possible to carry out a simulation of all or part of the brain that would have the same input-output properties as the original organ. Input and output do not have to be electrical: for example, the simulation could be connected to a biological neuron through an artificial half of a synapse, the other, original half residing on the biological neuron. The artificial half would emit or absorb neurotransmitters in just the right way to interface correctly with the biological half. If other kinds of signals (such as electromagnetic) are eventually discovered to play a role in communication between different parts of the brain, the simulation will also generate them just as the biological tissue did. Suppose now that we are partway throught the downloading of a human subject. The part of his brain that processes images from his left eye has been turned into a simulation performed by a digital computer (either sequential or parallel, it doesn't matter as long as it is fast and accurate enough). This simulation has the same input-output properties as the original tissue, and is connected to the rest of the brain and to the eye as described above. The rest of the brain, which includes the higher-level decision centers and the speech centers, is entirely unaffected. In particular, images coming from the right eye are still processed by the original tissue, and the subject can compare what he sees through his left eye to what he sees throught the other eye. We now show to the subject's left eye, invisible to the right eye, the image of a luscious red tomato in green foliage, and ask him what he sees. What will he answer? Let's try different possibilities. First pass: no information, no qualia. Some will say: "Since a computer simulation of a thuderstorm is not a thunderstorm and has never made anyone wet, the artificial brain part is of absolutely no use to the subject, who will answer that he is blind in his left eye and cannot tell us what he sees." But that's difficult to swallow because the rest of his brain is still getting exactly the same signals as before from the artificial part. These signals encode a description of the contents of the image, and the subject should at least be able to inform us of these contents. So: Second pass: information but no qualia. "All right", says the opposition, "the subject will be able to describe the image to us, but I'll be damned if mere bit-flipping in a digital computer can generate qualia. Describing an seeing an image are two very different things. Our subject will therefore behave as so-called blind sighted patients, who have suffered damage to the visual centers of their brains. They 'feel' blind but, due to some residual visual processing ability, perform better than chance when asked to describe what they should see. Our subject would therefore tell us that he still feels blind in his left eye, but that somehow, when we ask him what he sees, the answer assembles itself of its own in his mind". My answer to this is that such an outcome is impossible because it contradicts our premisses. But first, a short aside. Notice that the subject performs a lot better than blindsighted patients because, since he is getting *all* the visual signals of his former biological visual centers, he can describe the tomato in as much detail as we want. He can tell us the shape and number of the leaves surrounding the tomato, and whether it is completely ripe or still has some green in it. If he has any talent, he can even draw the tomato. If this is the case, are you seriously suggesting that he is not forming a mental image of the tomato? I claim that he is because the representation of the tomato in the computer simulation, and the signals describing it to the rest of the brain, *are* the mental image. Which leads us to my main argument. Remember (again) that the simulation has the same input-output properties as the original brain tissue it replaces. The remaining, unaffected, biological tissue, of course has the same input-output properties as before. Further, the simulation and the remaining biological part of the brain are interconnected exactly as the original biological tissues were. Therefore the overall system, made of the simulation and the remaining biological brain, has the same global input-output properties as the original, biological brain. If the input is the tomato image and our question as to what the subject sees, the output, which is the subject's verbal answer, should be the same as before. Therefore, the subject should not only describe the tomato, but report that his perception of it is unaffected, and that he sees as well throught his left eye as through his right eye. Which leads us to our Third pass: information but "false" qualia. "Well," replies the opposition, "this just shows that we can't believe what the subject is saying. He has been already partly transformed into a zombie by the experiment, and his sayings do not reflect the true contents of his consciousness." Again, this cannot be: the subject's decision and speech centers have been left unaffected, and I don't see how the simulation of the visual centers can control them. Therefore if the subject reports that he still perceives the qualia of red and green from his left eye, it is because he really and truly perceives them. Which demonstrates that "mere" bit flipping in a digital computer can generate qualia. QED, I think. Happy millenium to all, Daniel Crevier Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=13009