X-Message-Number: 55 From arpa!Xerox.COM!merkle.pa Tue Jan 24 13:28:51 PST 1989 Received: from Salvador.ms by ArpaGateway.ms ; 24 JAN 89 13:29:09 PST Date: Tue, 24 Jan 89 13:28:51 PST From: Subject: CRYONICS: Matter, Consciousness, and the like To: <kqb%> Message-ID: <> Status: RO Robert Ettinger recently wrote an article titled 'The Turing Tape and Clockwork People' in 'The Immortalist' (Vol. 19, No. 7, July 1988). Ettinger's conclusion was: 'If even a few of those very bright downloaders will realize that work should come before play, maybe real immortalism will get some much-needed help.' There followed a spirited series of letters. The next paragraph is a brief plug for two books that introduce and clarify many of the philosophical issues involved. The letter that follows was originally sent to 'The Immortalist' (and also 'Cryonics'), hopefully to clarify some of the issues being debated so vigorously; I thought it might be of interest to readers of the Cryonics mailing list. There are infinitely many philosophical works discussing almost every aspect of consciousness -- two of which I have read and enjoyed. 'The Mind's I' (by Douglas R. Hofstadter and Daniel C. Dennett, Bantam Books 1981) is a very entertaining introduction to many of the puzzles and issues involved. It has been highly acclaimed by The New York Times Book Review, the Washington Post, and many others. Kirkus Review accurately described it as 'philosophical fun and games of a very high order'. The second book, 'Consciousness and Matter' (by Paul M. Churchland, MIT Press (now available in a new, 1988 edition which updates the older 1984 edition)), is an upper division undergraduate level introduction to the philosophy of the mind. It provides broad and even coverage of the many theories and ideas about how the mind and brain interact in a well written and readable format. What follows is a series of questions that will hopefully reduce the heat and increase the light in future discussions of uploading. The first question deals broadly with the relationship between the laws of physics and the human brain. It is: 1.) Are the ultimate laws of physics the same both inside and outside the human brain? That is, is there something 'special' about the human brain that makes its behavior fundamentally different from the rest of the universe? This question carefully refers to 'the ultimate laws of physics' rather than the current laws. This avoids tedious digressions about their completeness and accuracy and focuses instead on the fundamental question -- is there something unique about the human brain that makes it forever unpredictable in terms of any laws of physics? While Q.E.D. (Quantum Electro Dynamics) is a remarkably accurate theory that fully accounts for all the known behavior of matter under the conditions that hold in the human brain (and a wide variety of other circumstances) it is still possible to argue that current physical theories are incomplete (a statement that most physicists will support) and that a new unified theory might somehow shed new light on the behavior of the human brain (a remarkably tenuous claim. How the behavior of particles in a high- energy accelerator will alter our understanding of the basic biochemistry that governs the human brain is at best unclear). This question also completely avoids any reference to consciousness. Whether or not physical law does or does not explain consciousness is simply not considered. All that is addressed is whether or not physical law explains the observed behavior of the human brain. This avoids a second fertile area for misunderstanding and confusion. A 'no' answer to this question almost completely blocks further discussion based on the use of physical law. Essentially, it is a declaration that modern Western science is fundamentally inadequate in dealing with the human brain and so makes it difficult to draw any further conclusions that will be generally accepted. It is safe to say that almost all scientists studying consciousness, awareness, or neuroscience will answer 'yes' to this question. The second and more difficult question is: 2.) Is it possible to computationally model the physical behavior of the brain without any significant deviation between the computational model and the physical reality, given sufficiently large computational resources? Again, we carefully avoid questions of 'consciousness'. We also don't say how much computer power is 'sufficiently large'. Finally, we introduce the tricky idea of a 'significant deviation'. A computational model of a physical system will fail to precisely predict the behavior of that system down to the motion of the last electron for two reasons: quantum mechanics is fundamentally random in nature, and any computational model has an inherent limit to its precision. The former implies that we can at best predict the probable future course of events, not the actual future course of events. The latter is even worse -- we cannot precisely predict even the probable course of future events. A good example of this second point is the weather -- weather prediction more than a week or two into the future might well be inherently impossible given any error in the initial conditions or computations. Any error at all (rounding off to a mere million digits of accuracy) will eventually result in gross errors between the actual events and the events predicted by the computational model. The model predicts sunshine next Tueday, and we get rain. This kind of error cannot be avoided. Any computational model of the human brain will almost certainly deviate from the behavior of the original -- eventually in some gross and detectable fashion. If I decide that it doesn't matter which of two courses of action to follow and allow myself to decide on whim, then it seems plausible that some slight influence might cause a computational model of my brain to select the opposite course. But is this difference 'significant'? Given that our model is highly accurate for short periods of time and that any deviations are either random or represent the accumulation of slight errors, does it matter that the behavior of the model and of the original eventually deviate in some gross and obvious fashion? We can view this another way: the human brain, as a physical system, is already subject to a variety of outside and essentially random influences caused by (among other things): temperature fluctuations in the environment; microwaves, light, and other electromagnetic radiation; cosmic rays; neutrinos; gravitational forces; last nights dinner; the humidity of the air; thermal noise; etc. If the errors in our computational model are smaller than these influences, and if in particular they are smaller than random thermal fluctuations, do we really care about the difference? Is it 'significant'? The human brain can and does continue to function reasonably well in the presence of gross perturbations (the death of many neurons, for example) yet this does not detract from our consciousness or life -- I don't die even if tens of thousands of neurons do. In fact, I usually don't even notice the loss. The rather small errors that we are in principle required to tolerate in a computational model seem small by contrast. It would seem, in principle, that a computational model of the human brain can successfully model all the 'significant' behavior -- where we tolerate a small amount of 'insignificant' deviation between the model and the original. This 'insignificant' deviation can be made smaller than the deviation caused by random thermal noise (at least in principle -- remember we assumed 'sufficient' computational power). We continue to avoid any discussion of 'consciousness' -- we are merely arguing that a computational model of the behavior of the human brain that is as accurate as a real brain subjected to random variations in particle behavior of the same magnitude as thermal variations is possible. A 'no' answer implies some basic mechanism in the brain is so sensitive that 'computational noise' must inherently substantially disrupt it. This seems very unlikely, given the much greater physical noise that we already tolerate. Finally, we turn to a question about consciousness! 3.) Given that the answer to both the first and second questions is 'yes', is such a computational model conscious? The question is largely unanswerable because we have no adequate definition of 'consciousness'. Even worse, many view consciousness as being inherently subjective and therefore any 'objective' definition (varifiable by others) is impossible. We illustrate the quandry in the following paragraphs. First, we imagine that a flesh-and-blood person and their computational model are both before us -- and that the computational model has been provided with a sufficiently realistic body that neither we nor the model know which is which. We do not ask 'can we distinguish between the model and the original' for we already know the answer: no. Given that we have answered 'yes' to both the first and second questions, then it is possible in principle to build a computational model that we cannot distinguish from the original by any test (assuming we cannot predict thermal noise). Therefore, it is necessarily completely futile to conduct any test, ask any question, or try in any fashion to 'trick' the computational model into revealing its 'true' nature -- we know in advance this can't be done. What, then, can we do? The subjective experience of the model is, by definition, not available for our examination. The objective data shows no significant behavioral deviation between the model and the original. Any definition of 'consciousness' that rests on behavioral considerations will necessarily conclude that both the model and the original are conscious to the same degree. Any definition that depends solely on subjective experience has already postulated that the needed information is unavailable, and therefore that the subjective state of both the model and the original is unknowable by anyone else. We must know the definition of consciousness before we can answer the question -- and once we define it, the answer is either obviously 'yes' or forever unknowable. I have a very powerful subjective feeling that I'm 'conscious' -- would a computer model feel the same? Would anyone (other than the model) know (or care) if it didn't? If it didn't have the same feeling of consciousness it wouldn't be able to tell anyone about this - - because it was programmed to faithfully imitate an original which did think it was conscious, and so the model would tell anyone who asked that it was conscious. By subjective standards I have no real reason to believe anyone else is conscious -- for I have no first-hand experience of your consciousness. Although you claim to be conscious, such claims cannot be accepted as evidence of actual consciousness (unless we are then willing to accept the claims of our computational model). Yet I believe that other humans are conscious -- is this merely blind faith? This topic is considered much more extensively in Matter and Consciousness , particularly in chapter 4, 'The Epistemological Problem', which considers both 'The Problem of Other Minds' and 'The Problem of Self-Consciousness'. Finally, we ask a question whose answer might actually affect the real world! 4) Given that the answer to the first, second and third questions is 'yes', is it possible to construct such a computational model in practice? Modeling the behavior of every single electron in the human brain will take LOTS of computer power. It might even be impossible to build a big enough computer to do this. This, however, is not an answer but simply a statement that a particular method of modeling the brain might not work. An obvious question to ask is whether some other method would work -- for example, a computational model based on the behavior of individual neurons and synapses might prove both satisfactory and feasible. There are roughly 10**11 neurons, and even more roughly 10**15 synapses. These are large numbers. However, when we consider that a single cubic centimeter can hold well over 10**18 molecular size gates, then a computational model based on the behavior of neurons seems plausible. Before using such a 'simplified' model we must return to the question of what is a 'significant difference'. Clearly, such a model ignores a great deal of the chemistry and biology of the human brain -- can it still capture those elusive things we call 'consciousness' and 'self'? If such a model walked up to us and struck up a conversation, what criteria would we use for deciding if it was conscious? Even if we decide the model is conscious, is it the 'same' person as the original? If we use behavioral criteria, could we distinguish between the model's behavior and that of the original? Our model is now based on a host of assumptions about the behavior of individual neurons -- how they work, how they interact, how they change. Are these assumptions all correct? If we've made an error, would we be able to tell? If we could tell, would we care? Would the model care? And even if the answers to all these questions were acceptable, many more questions would remain. Do these computer models break down a lot? Does society at large regard them as real people with real rights, or as funny computer programs that can be turned off when they start acting oddly? Has everyone else bought 'Advanced Mark XXIII Quantum Brains', now available at discount prices? Or were the last three people who attempted uploading shot and killed for 'crimes against nature?' Fortunately, the utility of cryonic suspension does not depend on the answers to these questions. It seems highly probable that at least one method for reversing cryonic suspension will prove feasible and generally acceptable (an excellent candidate is molecular repair via nanomachines). It also seems clear that we have inadequate information at the present time to determine the 'best' method, taking into account the broad range of technical, philosophical, and societal possibilities that confront us. At the moment, it seems prudent to delegate our choice to the best judgement of those dedicated individuals who we sincerely hope will still be tending our dewars when restoration becomes feasible. Once we are again able to make our own decisions we will face a wide range of choices -- and we will hopefully have both the means and the wisdom to make them successfully. At the very least, we will know very much more than we do today. Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=55