From: SMTP%"PSYCHE-D@rfmh.org" 5-SEP-1996 00:37:01.66 To: STAPP CC: Subj: QM in Stapp&Sarfatti vs Penrose and Hameroff Approved-By: PATRICKW@CS.MONASH.EDU.AU Approved-By: STAPP@THEORM.LBL.GOV Message-Id: <960904232551.28c0069e@theorm.lbl.gov> Date: Wed, 4 Sep 1996 23:25:51 -0700 Reply-To: PSYCHE Discussion Forum Sender: PSYCHE Discussion Forum From: Henry Stapp Subject: QM in Stapp&Sarfatti vs Penrose and Hameroff To: Multiple recipients of list PSYCHE-D This is a reply to two recent [AUG30,SEPT 3] postings from Pat Hayes. The key difference between classical mechanics and quantum mechanics, at least in the "orthodox" view of Niels Bohr, is tied to the difference within these two theories of the relationship between the observer and the observed. In classical mechanics the observed system is characterized exactly by what an idealized disembodied observer could know about the system without actually interacting with it, or disturbing it. Thus in classical mechanics the physical system is specified by what could be known by an observer that is conceived to stand apart from the observed system. But according to Bohr science is based on what we can learn from experiment: "The argument is simply that by the word `experiment' we refer to a situation where we can tell others what we have done and what we have learned... This crucial point, which was to become a main theme of the discussions... implies the IMPOSSIBILITY OF ANY SHARP SEPARATION BETWEEN THE BEHAVIOUR OF ATOMIC OBJECTS AND THE INTERACTION WITH THE MEASURING INSTRUMENTS THAT SERVE TO DEFINE THE CONDITIONS UNDER WHICH THE PHENOMENA APPEAR.[Bohr emphasis (italicized)]^1" Von Neumann's theory of measurements pinned down the fact that the chain of measuring instruments extends, unbroken, to the brain of the observer, where `what we have learned' presumably somehow resides. This inseparability of the physical world and our experiences is in general accord with the views of `physicalists'. But it holds in the actual world, which is a quantum world. This inseparability must be contrasted with the idealizations that define classical mechanics. The point is that the description of the physical system provided by classical mechanics is essentially the knowledge available to an infinitely sharp-eyed observer that stands apart from the observed system, as a kind of disembodied, non-interacting viewer-from-afar. It is important, in considering what can be known about a system within the classical-mechanics idealization of it, to distinguish three possibly different kinds of knowings: 1. That which a distant idealized observer could know about a physical system, namely the values of all the local physical variables that according to the principles of classical mechanics completely specify that physical system; 2. That which this distant observer could think in relation to this physical system, namely various concepts about this physical system that rest upon what this distant observer could know about the physical system; and 3. What an observer that IS a physical system could directly know about itself, on the basis of actually being (or being intimately connected to) that system. The phenomenal realities (pains, joys, sorrows, and other felt experiences) are knowings of the third kind: the knowings, feelings, and experiencings that we call a person's phenomenal life are not the knowings-from-afar of the disembodied and non-interacting abstract observer/knower of classical mechanics; they are the knowings-from-within of the embodied self-interacting physical system. Quantum theory supports knowledge of this third kind, because there is no clean separation between observed system and observing system: they are integral and inseparable parts of the single whole quantum observer/observed system. On the other hand, classical mechanics is formulated in terms of variables that correspond to the first two kinds of knowledge alone: the dynamical rules are specified without ever mentioning pains and sorrows, etc. This fact, that classical dynamics is formulated in terms of quantities corresponding to the first two kinds of knowledge alone, does not mean that knowledge/feelings/experiences of the third kind are necessarily absent from a world governed by the principles of classical mechanics. My claim is merely that, for any classically describable system S, the presence in association with S of knowings of the third kind IS NOT ENTAILED BY the principles of classical mechanics plus the boundary conditions that specify the form of S. If, as I claim, the presence in association with S of knowings this third kind is never *strictly entailed* by the principles of classical mechanics then these knowings COULD BE absent without disrupting the dynamically complete classical description of the system. Hence those quantities are, in this sense, epiphenomenal: they can be one thing or another, or absent altogether, without affecting anything physical. But then it not possible to give a naturalistic explanation of how human consciousness evolved, together with the physical organism, from simpler forms: if it could be absent altogether, or be this or that, without affecting the physical behavior then the principles of natural selection would have no impact upon it. To prepare for my rebuttals let me give an example that illustrates the three kinds of knowledge. Consider a classically described hurricane. The first kind of description of it would be equivalent to a computer description in which the trajectorty of each particle (and value of each field at each spacetime point) is represented in the memory of the computer either by enumeration (a tough job) or by some formulas, but absolutely nothing else is included. A second kind description might include all that information and also some meterological forecasts from the weather bureau, based on the first description, and perhaps on historical records and factors pertaining to other meterology structures in the neighborhood. A third kind of description would be a description of what the hurricane feels; what it is experiencing. As regards the third kind of description it may well be absolutely true that when a lightning bolt is unleashed in the hurricane there is, in the hurricane, a feeling that is akin to a pain that a human might have felt on some occasion. But the only issue here is whether it is logically possible to deduce the presence of such a pain, or of any other experience association with this system, strictly from the principles of classical mechanics alone. My claim is that it is not possible to do so, because those principles are expressed exclusively in terms of what a disembodied distant observer could know, and they therefore provide no basis upon which to give a strict derivation of what the experiences of an actual physical system must be. Hayes[30 aug 14:21:27 -0500] writes: "The issue is, can classical physics be compatible with an account of phenomenalism? Lets assume that it can, ie that there will eventually be a description within classical physics of how brains have a subjective life. Call this account P, and classical physics C. Then, trivially, it will follow from P and C that brains have a subjective life. So if we assume that P exists, Stapp's argument fails. But Stapp claims to have a *refutation* of the possibility of P. One might find the possibilty of P implausible on other grounds, but this aint one of them." HPS: The issue is not whether subjective life exists. It is whether, for some specified classically describable system S, its subjective life can be proved to exist from the principles of classical mechanics alone. If it cannot, then---within the framework of classical mechanics---the subjective life of that system (i.e., its consciousness) is epiphenomenal: what happens in the subjective life of S would make no difference at all, as far as the (assumed to be classical) physical universe is concerned. To bring Pat's example into conflict with my claim it would have to be shown that the subjective life P of S could be strictly deduced from the principles of classical mechanics alone, plus the boundary conditions that specify which of the possible physical systems the system S is. I have given the reason why this is impossible. Pat asks for an explanation of how, in my argument, carburettors differ from brains, and says: "Note, the point is not to claim that carburettors have a mental life: on the contrary, my point is that they clearly dont, yet Henry's argument seems to apply just as well to them as it does to mental things." HPS: Knowledge of carburettors and such things belong to the second kind of knowledge. And the properties of carburettors follow (let us say) from the principles of classical mechanics plus the boundary conditions that specify that the system is a carburettor. Carburettors can be defined in terms of structure or function, it doesnt matter which. But however it is defined, this definition is expressed in terms of what a disembodied observer far away from the carburettor could "see", and the properties deduced for this system from the PCM are similarly expressed in terms of what the disembodied distant observer can "see". So there is no valid analogy between properties of carburettors and consciousness: knowledge of the latter is knowledge of the third kind. It is knowledge of what some system is experiencing. Given the boundary conditions that specify the existence of a brain, it remains, as a matter of principle, impossible to deduce from the boundary conditions and the principles of classical mechanics what the brain is experiencing because the knowledge supplied by classical mechanics, plus the boundary conditions that specify the presence of the brain, consists only of what is available to a disembodied distant observer. About the chess-playing computer I said: .. the principles of classical mechanics give no toe-hold for passing from the behaviour of a system, no matter how complex this behaviour is, to what the system is feeling, or experiencing. One can postulate such a connection, or determine empirically the existence of such a connection. But even such behaviours as those that reveal that a chess-playing computer `understands' the rules of chess, and `knows' where on the board the `white king' is located, do not allow one to deduce, given only the principles of classical mechanics and the boundary conditions that specify the system, that this machine is having even the faintest glimmering of the experiences that we feel when we feel that we understand or know something. It is completely compatible with the principles of classical mechanics for the chess-playing machine to have absolutely no feelings or experiencings at all. Pat [3sept 13:26:36 -0500] responded: "Im sure it has no feelings, indeed: but it ISNT compatible with the principles of classical mechanics (PCM) (or with anything else) to claim that the program has no representation of the rules of chess, notice; yet PCM makes no reference to chess. So here we have a nice example of how a thing - a computer running a chessplaying program - which is compatible with PCM can display aspects which are not predictable from PCM, or indeed even expressible in them; which ought to be impossible, on your argument." HPS: No claim is made that the chess-playing computer has no representation of the rules of chess: I said just the opposite. As you say, the PCM make no reference to chess, and the rules can be represented in all sorts of computers, the latter being specified by the boundary conditions. This is a nice example of how a system compatible with the PCM can display aspects that *can be deduced* from the PMC plus boundary conditions. My claim is that if the machine were conscious, which you admit it is not, then neither that fact nor the content of the machine's consciousness could be *strictly deduced* from the BC+PMC [Boundary Conditions (that specify the system in question) plus the principles of classical mechanics] Why do you think there is a problem here, Pat? The essential point is that the properties of the carburettor, as it is described in classical mechanics, gives only properties of a certain "classical" kind (and the very same thing goes for brains) and these classical properties are of a special kind that, for very general category reasons that I have given, do not include consciousness. This says nothing at all about real brains, which are quantum systems, and it even as regards classically described systems there is no claim that they cannot be conscious: the claim is only that, as a matter of principle, one cannot *strictly deduce* the presence of consciousness from the BC+PCM. I said, in my previous post: By "behaviour" I mean here that which can be expressed in terms of the concepts of classical physics. The point is that classical physics deals with things that can be built out of the basic elements of classical physics, namely the particles and local fields. Particles have the feature that they are completely described in terms of a description that can be regarded as a view from the outside: the description is a description of what a huge collection of tiny marbles would "look like" to an outside observer that was able to comprehend where each of these particles was located, and in fact its entire spacetime trajectory. And the fields are basically of the same kind: they describe what can be imagined to be a bunch of wavelike structures of the kind that we see when we look at the ocean from an oceanside cliff: we are looking at the wave from a point outside the wave. Of course, the wave may have structures, and these structures may have substructures, but the entire mathematical structure provided by classical mechanics can be regarded as a description of a view-from-a-distance of a system of particles and waves. Pat responded: "Ah, this introduces a new theme, of third-person instead of first-person descriptions. Your argument so far hasnt mentioned this, but I will respond to it in a directly in-its-face kind of way. Consider a system ( described classically or quantum-mechanically or in any way you like) which claims to have a subjective life, and imagine we can look at how it works with a kind of brain-scope, so that when the system is, say, trying hard to ignore the pain in its left elbow, we can 'see' that certain internal operatings and structures and representations, etc. etc. are interacting in complex ways, with feedbacks going on and God knows what else. Suppose that people find (let us suppose) that its subjective reports are closely correlated with our objective view of its internal processes and indeed that they are able to predict what its subjective descriptions will be, by observing its inner goings-on. (Lets make the fable truly convincing. Suppose that this thing worked pretty well on whoever you turned it on, until a subject and the brain-scope disagreed, and this was widely touted in such publications as NYRB as evidence of the Freedom of the Will, but it was later discovered that in fact this subject had been lying about its internal impressions (having been bribed to do so by a fundamentalist Christian organic chemist) and the subjective impressions observed by the brain-scope had in fact been accurate.) "Now, according to your argument, we would never be justified in putting forward our third-person view as an explanation of its first-person view. Right? Because no matter how detailed our account of how its innards worked, it could be a zombie, with no first-person view at all. Even if our third-person account spoke of the informational content of its 'views' and 'thoughts', and of the conclusions it was able to draw, and the intensity or urgency of them, etc., : no matter how detailed an account it gave of the structure (rather than essential nature) of subjectivity, we could never get a toe-hold into the subjective world. Right? (If so, by the way, you dont need to keep on about PCM, which is just a distraction.) OK, I'll just say: wrong, for three reasons." HPS: The issue is not what one is justified putting forward as an explanation of first-person phenomena; it is only whether one can *strictly deduce* the presence of first- person phenomena from BC+PCM. From the kind of behavioural evidence you descibe it would be reasonable to make the "presumption" that consciousness is present. But such a presumption involves a logical jump or leap: is not equivalent to a strict proof from the BC+PCM. But if the consciousness logically COULD BE absent, with all behavior being unchanged, then consciousness WOULD BE epiphenomenal. Pat gave his three reasons: "First, if one wants to get *this* careful about attributing subjectivity to other things, then the only natural conclusion seems to be solipsism. We regularly attribute consciousness to other systems (notably people, but also eg other mammals) on the basis of far, far scantier evidence than the kind hypothesised in this fable. It seems simply irrational to insist that something that claimed to be conscious should be forbidden the honor just because we can see how it works. (To put the point another way: when faced with a something which acts as though it were conscious, the burden of proof should be on the one who claims its a zombie.)" HPS: The issue is not honor, but causal connection: it is a logical question. PH:"Second, the standard objections to this kind of identification of subjective with objective all seem to turn on the rather simple fact that one of the observers is in a uniquely special relationship to the mental phenomena, ie is actually experiencing them. Our third-person view of it is different from its first-person view: we are looking at it from the outside, while it is feeling itself. But of course the system being viewed is in a different (and unique) position: its at the small end of the mind-scope. It is the system in which these things are happening. When observing *ourselves*, we are all in a unique position (not neccessarily uniquely privileged or authoritative, but unique), simply as a matter of logic. Tautologous objections carry little argumentative force." HPS: Yes, I completely concur that there is a view-from-within: quantum mechanics is based upon it, and can integrate it into an ontology that includes it as a core element of the physicist's representation of nature. The problem with classical mechanics as a foundation for a science of consciousness is that it does not have any such natural place for the view from within: it can be left out of the classical-mechanical ontology without disrupting the classical theory, though it is important in quantum theory, and I'll bet in nature as well. PH: "Third, the state hypothesised here is precisely the situation that *any* succcessful theory of consciousness is going to have to achieve. People have a long history of deciding that behaviors cease to be regarded as evidence of consciousness (or of having a spark of divinity, or being Blessed with the power of the Word, or of being Higher than Brute Nature, or whatever other special property we clothe ourselves in) whenever we can understand how they work. When we do understand how some mental act could be done, we demote it to the obviously non-mental (and then call the earlier way of talking a 'misuse of language'.) Until the 1940's, being able to do rapid, accurate mental arithmetic was widely regarded as one of the highest intellectual talents: now, to call someone a calculator would be a contemptuous insult. Unless we can learn to cast off this 'mysterious-mind' attitude, we will never be able to make real progress in understanding ourselves and how we work." HPS: The occurrences of views-from-within are indeed needed in any satisfactory theory of consciousness. The problem I am concerned with is how do they fit rationally and coherently into physical theory. I said: Pat is suggesting that "given an adequate account of how consciousness can be physically realized" the principles of classical mechanics could entail the existence of the experientially felt realities. But how does one arrive at such an `adequate account'? Does one: (1), postulate the sameness of certain classically describable activities and certain felt experiences, or (2), derive this connection from the principles of classical mechanics. Pat admits that one cannot derive this sameness from the principles of classical mechanics: he admits that option (2) is not open. But if option (2) is blocked and one must postulate the connection, rather than derive it, then one can, logically, deny this postulate. But then the existence of consciousness cannot actually be derived from the classical principles: consciousness *could be* absent without violating the dynamically complete classical principle, and hence consciousness *would be* epiphenomenal. Pat responds: "The answer is (1). Your option would always be open, of course: one can always deny the reality of anything of any reasonable size. Eg I might insist that there are, in fact, no trees, since Newton said nothing about trees. Faced with an oak, I can respond that all it is (from the perspective of the PCM) is a large bunch of local particles and fields all interacting, etc., so the concept of 'tree' is epiphenomenal. It would all work just as well as does even if it werent a tree. "The reasoning here is exactly parallel to yours, but one can clearly see the mistake. That collection of particles and fields IS a tree: thats what we refer to when we say "tree". So to insist that it isnt a tree would be to make a nomological error. But notice, before biology had the concept of a cell, and urea had been synthesised, and atoms were taken as obviously real, thats NOT what "tree" meant. Theres nothing in the ordinary concept of "tree" that reduces it to atoms and fields. That all follows from our understanding of how it is that trees are actually constituted; an understanding achieved very slowly and often revealing truths that seemed very surprising (Your 'boundary conditions' include almost all of science.). Why should something like this not also happen for the mind?" HPS: There are several different points here. In the first place, if you accept (1), then you accept my main claim that within classical mechanics the occurrence of consciousness is not a consequence of the basis principles, but rather an ad hoc add-on, and hence epiphonomenal. Quantum theory, adequately formulated, is altogether different: the view-from-within is an integral component of the theoretical framework, and consciousness is not epiphenomenal. Then there is the point about "trees". In classical mechanics, to the extent that it covers trees, we can take a tree to be a collection of particles and fields connected together in a certain way. There is no argument to the effect that because a tree is a certain conglomeration of particles and fields that it is not a tree. That is what trees are. And as such they, like hurricanes, are causally efficacious. And one can, in principle, *deduce* their properties from the principles of classical mechanics and the boundary conditions that specify the initial configuration of the particles and fields. Then you ask, given the prospects of huge future developments in science, why a "tree" is basically different from a "mind" The point is the by considering the classical mechanics idealization we are dealing with a conceptual system that is vastly simpler than the real world, whatever it turns out to be: it does not include things that depend upon detailed properties of materials, chemistry, or biology. It is conceptually very simple. It is this tremendous conceptual simplicity and clarity that allows one to say some general things about classically described idealizations, as logical systems, that will remain true forever. Pat gives a quotation from my earlier posting: It is certainly true that the principles of classical physics do not specify which systems actually exist in nature: some extra boundary-value-type information is needed to fix that. But all of that is beside the point. It is only after the particular physical system that is supposedly implementing some property is specified that the question arises as to the relationship of that physical system, as it is represented in the physical theory (i.e., classical mechanics), to various other things arises. If the classical-mechanics model of the physical system is adequate then it must account for the functional behaviour that is observed "from the outside". Let us suppose that it does so. Then we arrive at the question of what the system is experiencing or feeling. This brings us back to the logical problem: the fact that boundary conditions are needed to specify what the physical system is does not allow this logical problem to be evaded. Pat responded: "The logical problem is yours. Heres the logic: the hypothetical 'boundary conditions', as you call them, are a theory of how phenomenal experiences could exist in physical nature. I know you find this implausible, but lets take it as a reductio assumption and try to derive a contradiction. So if physics is P, and these boundary conditions are B, then (ex hypothesii) (B&P) entails E, the existence of experiences. (P alone does not, of course.) NOw, where is there any contradiction? You have told us that that P has a resolutely localist view, but so what? We agree that B could explain how larger and more complicated things might be implemented in littler things. You have told us that P has a third-party view of things. Again, so what? B is hypothesised to give an account of how to translate first-person reports into third-person reports (via the brainscope and various mental-physical identity principles.) So again there is no contradiction here that I can see." HPS: The boundary conditions that I was talking about are something that classical physics needs in order to reduce the vast set of conceivably possible systems covered by the theory to some particular system of interest: the boundary conditions are an integral and essential part of the classical-physics framework. These boundary conditions are expressed within the set of concepts that classical mechanics deals with. Now I do not find it at all implausible that, if classical mechanics were adequate as a foundation for a theory of consciousness then there would be needed an extra theory, your B, of how phenomenal experiences could exist in physical nature: B would be a theory of ``inner views" added onto P=PCM. And then B+P would entail E=experience, as you say. That is perfectly fine! There is no contradiction at all. I have repeatedly said that there is no contradiction between classical mechanics and the existence of consciousness. But the question is whether the existence of consciousness is deducible from BC+PCM, where BC represents a boundary condition that specifies, for example, whether, within the classical framework, the system we are dealing with is a hurricane, or a carburettor, or a chess-playing machine: it is formulated in terms of the sort of variables that classical mechanics deals with, which are in principle expressible in terms of the basic variables of the classical theory. It is a *restriction* on the set of classically allowed possibilities, not a theory of how something not mentioned in classical physics is to be added on. By adding to the PCM not merely the conditions BC that specify which of the infinite collection of classical systems we are considering, but also a logically independent postulate B that goes beyond this, and specifies how views-from-within enter into the theory you are conforming exactly to what I say you must do, if you wish to incorporate conscious into classical mechanics. The answer to all of the questions was essentially the same answer. That is tedious, but it means that I do not have to scurry around to find some way to meet each new challenge: one argument says it all. But seeing how it work in the various contexts may serve to flesh out this one important point, which I think must be taken as the secure basic starting point of a satisfactory theory of consciousness: although hurricanes, and carburettors, and chess-playing computers are causally efficacious within the classical physics idealization, thoughts are epiphenomal. But in the actual (quantum) world they needn't be. Henry P. Stapp 1. Niel Bohr, in Albert Einstein: Philosopher-Scientist, Tudor, New York, 1951 p.210.