From stapp@thsrv.lbl.gov Thu Jan 29 16:44:29 1998 Date: Thu, 29 Jan 1998 16:42:34 -0800 (PST) From: Henry Stapp To: Pat Hayes Cc: hpstapp@lbl.gov, bdj10@cam.ac.uk, brucero@cats.ucsc.edu, klein@adage.berkeley.edu, phayes@nuts.coginst.uwf.edu, A.Sloman@cs.bham.ac.uk, keith@imprint.co.uk, ghrosenb@ai.uga.edu, chalmers@paradox.ucsc.edu, srh@ccit.arizona.edu Subject: Re: Solicitation On Wed, 21 Jan 1998, Pat Hayes wrote: > >[HPS] > >I am soliciting comments about how to make it better. > > >[Pat] > Well, I dont really know how to make it better, but I know what seems to be > wrong with it :-) Dear Pat, January 27,1998 Many thanks for your criticisms of my draft of Appendix B. Your comments all revolve around the basic difficulty in trying to ontologicalize orthodox quantum theory, staying within science: human experiences surely are not the only experiential-type things in the universe, yet human science is based on these alone. The rest of nature could, as far as the empirical evidence is concerned, be devoid of quantum collapses and the associated experiences: we have at the moment no (accepted) empirical evidence that would require any other collapses or experiences in nature. That problem is, in fact, just the one I am trying to cope with, in some reasonable way, in this appendix. As a first step toward trying to improve my appendix let me answer as directly as I can your criticisms of it. [ > indicates Pat's comment; >>indicates a passage from my appendix.] > > My main problem is that I dont know what a 'knowing' is supposed to be. > While it might not be reasonable to ask for a *definition* at this stage, > nevertheless one needs some kind of guide for ones intuition in order to > even follow what an author means when he uses a coined word or a technical > usage like this. And you start with what is apparently a protypical > example of a 'knowing': > > >A typical knowing of the kind that quantum theory is built upon > >is a knowing that the pointer on the measuring device appears to lie > >between the numbers 6 and 7 on the dial. > > OK, I can grok that; so a 'knowing' is something like an event of > perception, or a coming to believe something, or a realization that > something is true, or a having of a feeling about something, etc.. (I dont > believe for a second that your next sentence is in fact true, but leave > that squabble aside for now.) > But all of these kinds of examples clearly require there to be a knower who > is expereincing these knowings. If these are the examples I am to use to > understand the concept, then the idea of a pure knowing, bodiless and > impersonal, seems simply incoherent. > What my paper is about is precisely the question of the connection between experiential-type things and their bodily hosts. In quantum theory the human knowings are certainly connected to human bodies. I take a human being to consist of both his bodily and experiential aspects, and am trying to determine on the basis of the available evidence both the nature of these two aspects, and the nature of their relationship. In the light of the failure, at the basic level, of classical ideas about the nature of the physical world one cannot just take it for granted that the bodily aspect are basically in accord with ideas of classical mechanics, and that our experiences are thus either just some epiphenomenal ``nonphysical kind of stuff'', or nothing but matter itself as matter was classically conceived, or some aspect of nature that can be adequately conceived within the framework of an essentially classical-type conception of the nature of the physical world. This question is certainly a wide open question in science. I take human knowings to be the foundation of human science. This seems to me the only reasonable thing to do in the face of the challenge to the basic correctness of the ideas of classical mechanics that quantum theory has posed, and the fact that the orthodox scientific solution to the puzzles of quantum phenomena was to retreat to this more secure knowledge-based position. > Im prepared to accept that pigs and dogs, and even maybe - for the sake of > this argument - paramecia, might be said to have such knowings. But in all > these cases there is a knower somewhere. Which implies that until there was > at least life of some kind, 'knowings' like this simply couldnt have > existed. > You reject, then, the idea that machines could be conscious? I certainly tried to make it clear that in order to encompass other kinds of physical hosts our human knowings had to be regarded as a very special case of a general class of events. > Your own account seems to imply this also, even when it suddenly gets > technical: > > >... In the von Neumann/ > >Wigner type ontology adopted here there is, in association > >with this knowing, a collapse of the state vector of the universe. > >It is specified by acting on this state with a projection operator > >that acts on the degrees of freedom associated with the brain of the > >perceiver, and that reduces the state of the body/brain of the observer, > >and consequently also the state of the whole universe, to the part of > >that state that is compatible with this knowing. > > Notice that the knowing is located in an observer; and in fact seems to be > identified with a change of that observer's physical state. So (to labor > the point) if there were no observer, there would be no state to change, > and hence no knowing. > > ... > >That ``person'' is a system consisting of a sequence of > >knowings bound together by a set of tendencies that are specified by > >the state of the universe. This state is essentially a compendium of > >prior knowings. However, these knowings are not merely human knowings, > >but more general events of which human knowings are a special case. > > But now you shuffled the cards. All the examples we have been given refer > to human or human-like experience embodied in physical bodies. How do we > generalise this intuition to these 'more general' events? What category is > this new, more general, concept, of which we have no examples other than > the proper subset with which we are all familiar? Can you give us just a > sketch of what a "knowing" might have been at the time when, say, the > universe's temperature was so high that atoms had not yet formed? Or inside > a main-sequence star? Or in deep space where the 'boiling vacuum' is busily > creating high-mass particle-antiparticle pairs and then reanihhilating them > so quickly that the violation of mass conservation doesnt escape the > Heisenberg limit? > The theory of the early universe is not in very good shape: I think all experts agree that a lot of tinkering is still needed. The question of what role the quantum collapses played then has not really been put in. It is possible that the universe evolved for billions of years with no collapses until the first primitive life form occurred, and then there was a stupendous collapse. Perhaps the ``anthropic principle'' could be explained by some such scenario. But I think we should allow, at this point in the development of an adequate understanding of the early universe, also the possibility that there were collapses at earier times, prior to the emergence of life. Perhaps there would be no ``experiential-type'' event associated with such a collapse. But there is the problem that there exists today no understanding of what it is that chooses the particular collapse event that actually occurs, insofar as one stays within the confines of the ``physical'' aspect of nature specified by the (Hilbert-space) state vector: that (Hilbert-space) aspect seems naturally suited to give only statistical aspects. Some other aspect of nature seems to be needed to fix what actually happens: i.e., to fix which one of the various possible collapses actually occurs. So it seems that a purely physical description of nature (i.e., Hilbert-space description) is not complete: some other aspect seems to be needed. This problem of the early universe is at least as hard as the mind-matter problem because of the sparsity of data. We have human brains in abundance, close at hand, and can do all sorts of experiments on them. So I do not think it makes much sense to say that we have to produce a detailed theory of the early universe as a part of a proposed approach to the mind-matter problem at the human level: things could go in many diiferent ways in the extrapolation to the early universe. But in answer to your challenge to what a ``knowing'' during the early stages of the universe ``might have been'' I refer you to section 14, ``What is Consciousness?''. The purpose of that section was to provide an answer to your question. If the collapses are basically to states defined by vibrations in certain degrees of freedom of the universe, then I can at least dimly imagine that the raw basic experience could be some basic prototype ``sound'' of those vibrations. Of course, the ``sounds'' that we experience are surely tremendously enriched by the input elicited from our memories, and it is perhaps impossible was us to have a clear idea of the raw pure experience of the sound of a vibration outside of our human situation. Still, if one wants to hold onto the idea that there are collapses outside animals, and indeed in the infant universe, and that SOME sort of generalization of human experience should be associated with such collapses, then I think that this idea that `primordial experience is like the experience of the sounds of the vibrations that are being actualized' is an intuitively accessible notion. Since the experience of a ``chord'' is more complex than just the sum of the experiences of the individual notes there seems to be the possibility that all experiences could be basically like the experiencing of the sounds of those chords vibrations that are being actualized. These vibrations will be vibrations in certain degrees of freedom of the state of the universe, and hence in some physical system: I do not envisage any experience that is not tied to certain physical degrees of freedom, and to an actualization of some fact that is completely characterized by a collapse of the state vector of the universe. This collapse involves placing a restriction on those degrees of freedom. > >In strict Copenhagen quantum theory this theory is regarded as merely a > >set of rules for making predictions about human knowledge on the basis of > >human knowledge: horses and pigs do not make theoretical calculations using > >these ideas about operators in Hilbert space, and their ``knowings'' > >are not included in ``our knowldge. > > > >But in a science-based {\it ontology} it would be unreasonable to posit > >that human knowledge plays a singular role: human knowings must be > >assumed to be particular examples of a general kind of ``knowings'' that > >would include ``horse knowings'' and ``pig knowings''. These could be > >degraded in many ways compared to human knowings, and perhaps richer > >in some other dimensions, but they should still be of the same general > >ontological type. And there should have been some sort of things of this > >general ontological kind even before the emergence of life. > > Another huge shuffle! OK, we can all accept horse knowings and pig > knowings, but what about vacuum knowings or neon-laser knowings or > gamma-ray knowings, etc..? > My answer here is essentially the same as above; I do not think the boundary between life and nonlife should be that sharp or critical. It probably true that ``knowings'' could be defined so that they are emergent: so that that they just fade into nothingness as the system associated with the collapse loses various features that are present in normal human brains. One could define ``knowings'' narrowly in this way, and that might indeed be less abusive of language. So that is certainly a justified position. But IF human thoughts are somehow connected with the choice or selection of WHICH (from among the various possible) collapses are made in human brains---and it is this connection what would give our thoughts some nonepiphenomenal role of the physical world---then it would seem that some ANALOG of our thoughts would be needed for the other collapse: if our thoughts are not epiphenomenal, but do play an important role in the determination of which state will be actualized, then it would seem that, for ANY collapse, SOME analog of a human thought would be needed to play this causal role. That is my assumption, and the basis of my broad way of defining ``knowing''. > (BTW, for the record; very, very few human beings make theoretical > calculations about operators in Hilbert space, so even if we exclude horses > and pigs, there is still a lot of knowing going on that has very little to > do with pointers and dials.) > The strict Copenhagen interpretation is not ontological: it does not seek to say what exists. It says that the quantum formalism is merely a tool that scientists use to make predictions about what they will learn by doing certain kinds of experiments, which they describe to each other in the language of classical physics. The strict Copenhagen interpretation is about ``our knowledge'', but the "our" is never sharply defined. It certainly could be just a communicating group of scientists and technicians. Indeed, insofar as a measuring device is not understood dynamically, or by empirical calibration, but by its communicating its observations to us by language we will probably want to treat it as one of "us" > >Science is an ongoing endeavor that is expected to develop ever more > >adequate (for human needs) ideas about the nature of ourselves and of > >the world in which we find ourselves. Newton himself seemed to > >understand this, although some of his successors did not. But the > >present stage of theoretical physics makes it clear that we certainly > >do not now know all the answers to even the most basic questions: physics > >is still very much in a groping stage when it comes to the details of > >the basic underlying structure. So it would be folly, from a scientific > >perspective, to say that we must give specific answers now to all > >questions, in the way that classical physics once presumed to do. > > Now what is this paragraph for? It seems to me to be an attempt to build a > dyke against the flood of obvious objections and questions; but if so, it > has gaps in it. The questions Ive raised arent scientific questions: they > are questions about what it is that you *mean*. > Your questions are about how the human knowings that we know become generalized when we try to expand the pragmatic/epistemological strict Copenhagen approach to an ontological one. The problems are indeed enormous, and that is why the Copenhagen people declined to get involved at any detailed level with the ontological issues. We have no data about the ``knowings'' of non-humans, and no data about collapses that are not associated with accretions in human knowledge. So within the bounds of contemporary science one is pretty much confined to the human mind-brain, plus the reasonable general principle that human mind-brain process should not be some unique process in nature: it MUST (I hold) be a special case of a more general process. So all I am trying to say here is: let us admit that human mind-brain process is just a special case of a general natural process, but nonetheless focus FOR NOW on the mind-brain human problem, where we have a kind of knowledge that we do not have in the general case. Science is to some extent opportunistic: it seizes upon opportunities as they open up. It focusses on issues where there is relevant data. > ... This lack of data about > >nonhuman knowledge would make it presumptuous, in a science-based approach, > >to try to spell out at this time details of the nature of nonhuman knowings, > >beyond the reasonable presumption that animals with bodies structurally > >similar to the bodies of human beings ought, to the extent they also > >behave like human beings, to have similar experiences. But knowings cannot > >be assumed to be always exactly the kinds of experiences that we human beings > >have, and they could be quite different. > > So we are now left not having the slightest idea what it is that you are > talking about. A 'knowing' can be anything, it seems, and you even consider > it to be presumptious to try to say what it is. This hardly seems like a > good basis on which to try to build a new physics. > The idea is to start where we do have data, and can gather more, and try to build out from there on the basis of the constraints imposed by such data, by using a parsimonious theoretical structure that we have invented that is in accord with all known data, and that is supported by the validity of a huge array of highly nontrivial predictions. A ``knowing'' cannot be just anything, within that framework: it must be the coming into being of a selected new fact that is completely represented physically by a collapse of the state of the universe to a new form. This collapse is associated with certain degrees of freedom of that state, and hence with a physical system. > > > >The knowings that I mentioned at the outset were percepts: knowings > >that appear to be knowings about things lying outside the person's body. > >But, according to the von Neummann/ Wigner interpretation, each such knowing > >is actually connected directly to the state of the person's > >body/brain, after that event has occurred. This state of the body/brain > >will, in the case of percepts of the external world, normally be correlated > >to aspects of the state of the universe that are not part of the > >body/brain. But experienced feelings, such as the feelings of warmth, joy, > >depression, devotion, patriotism, mathematical understandings, etc. are not > >essentially different from percepts: all are experiences that are > >associated with collapse events that reduce the state of the > >body/brain to the part of it that is compatible with the experience. > > > >I have spoken here of a body/brain, and its connection to an experience. > >But what is this body/brain? It seems to be something different > >from the knowing that it is connected to. And what is the nature of this > >connection? > > > >The body/brain is an aspect of the state of the universe. > > EVERYTHING is an aspect of the state of the universe. So this doesnt help > us very much in trying to pin down what you are talking about. > By ``state of the universe'' I mean the Hilbert-Space state. This is not what the ``state'' is conceived to be in classical mechanics. And in Copenhagen/vonNeumann/Wigner quantum mechanics there is the question of ``what is it that selects [or fixes or determines] which collapse actually occurs''. That aspect of nature is not normally conceived of as being an aspect of the Hilbert-Space state of the universe. So the Hilbert-space state of the universe does not necessarily represent everything in nature. In my terminology it represents the ``physical'' part of the universe, because the causal/temporal structure of that Hilbert-space part is governed by equations that are analogous to the equations that govern the physical universe of classical physics. Yet this ``physical'' part only specifies, as far as we now know, only the statistical feature of nature: it specifies only the probabilities for the occurrences of knowings. > .... > > >... a > >probability is more like an idea or a thought than a material > >reality. > > Thats a remarkable claim. I find it quite odd, in fact. After all, > probabilities can often be measured objectively with arbitrary accuracy, > given large enough populations. They seem like very objective, external, > things; aspects of the universe rather than subjective, internal things > like ideas or thouights. > This is, of course, a very big issue for philosophers of science: Is probability to be thought of in the objective ``propensity'' sense (e.g., as recommended by Popper) or as a feature of what is known when someone knows less than everything that could be known. To make the idea of probability precise one usually gets involved with `infinity', which may be more of an idea than an actual feature of nature, and with infinite imaginary ensembles. The large collection of future possible experiments to which the probabilistic predictions of quantum theory refer will, for the most part, never be actually performed. > Maybe this is one explanation for our clash of intuitions. I see the > probability waves in the Copenhagen interpretation as something real, > external, objective. They are there even if nobody is measuring anything; > they are what the world is in some sense really made of. Its a very odd > picture compared the the classical one, but it doesnt require one to posit > anything psychological or epistemic about them. They are *physical* > probabilities (of a particle having a momentum or an energy, etc.), not > epistemic probabilities. > When Heisenberg describes ``what really happens'' (See chapter 3: The Vulgar Copenhagen Interpretation) he definitely uses this objective tendency (propensity) idea. But he does distinguish that from what the probability function of (Copenhagen) quantum theory represents, namely our knowledge, which suddenly jumps when our knowledge changes. Popper has written a strongly-worded denounciation of the Copenhagen interpretation (calling it `The Great Quantum Muddle') in the book ``Quantum Theory and Reality'' (ed. Mario Bunge/Springer-Verlag) which is basically an attack on the use of the subjective interpretation by the Copenhagenites, [a fact that he amply documents] and a polemic in favor of using, instead, the propensity interpretaion. But my position on the nature of the quantum probability is intermediate. I do not adhere to the Copenhagen idea that the quantum formalism is MERELY a tool for scientists and engineers to make predictions about what their future experiences are likely to be. I ontologicalize, and give no special ontological role to human beings, and make the probability function objective: it is tied to an objective sort of knowledge to which all ``observing systems'' can contribute, be they human or not, or biological or not. The probability does thus the become objective, since the relevant knowledge is objective. Yet it is a representation of ``knowledge'', and it evolves as this knowledge grows, with a sudden jump associated with each new ``observation'', insofar as an observation means a new ``knowing''. > > ...And the fact that this probability suddenly changes when a > >new knowing occurs also makes the state more idea-like than matter-like. > > But is that really true? Isnt it rather that the probability suddenly > changes *for the observer*. But if someone else includes that observer into > their external universe, then the first observer's observation makes no > difference to his calculations. You wanted to make the probabilities objective, not subjective/epistic. Yet I would agree that BESIDES the objective probabilities there can be subjective (Copenhagen-type) ones that reflect only the subset of knowings that a community of communicating human observers possesses, and that they can use to make practical predictions about what their future knowings are likely to be. This subjective probability associated with this communicating community of human being reflects only their knowledge, and changes only when their knowledge changes, whereas the objective probability changes when any observing system has a knowing. Thus my approach allows the epistemological Copenhagen quantum theory [which is weakly objective in the sense that the probabilities do not depend upon any particular member of the community of communicating observers] to be naturally imbedded in a [strongly] objective ontological structure of the same general form. > As Stan keeps reminding us, one can draw > the boundary wherever one likes. So there is no objective reality to the > 'quantum collapse'; all these 'collapsings' are just our various individual > ways of perceiving the flux of probabilities which constitute the real, > physical universe. Stan has emphasized that a key difference between him and me is that he wants a sliding boundary, whereas I seek a definite ontology. But I think he says the boundary has to be put somewhere. In any case, my approach is to take seriously the von Neumann/Wigner idea that each collapse occurs in conjunction with an experience, which is in accord with the Copenhagen view. But the von Neuman/Wigner approach can be deemed ontological, whereas the Copenhagen approach is pragmatic and epistemological, and anti-ontological. I do not claim that every quantum physicist would agree that ontologicalization is a desiratum, or that the ontologicalization that I propose, which stays as close as possible to the orthodox ideas, and provides an imbedding of them in an objective ontological structure of the same general form, is the best way to go. But I think it is coherent, and it does address the mind-matter question in a way that is both compatible with, and suggested by, contemporary physics. > The thing that needs to be explained, then, is why we > all feel so vividly that what we perceive doesnt consist of probabilities > but instead seems more like the classical world which we all once thought > it actually was. Why do we have this mass illusion, as it were? > > Pat Hayes Well, we do not perceive the probabilities or propensities: we see the Geiger counter either fire or not fire, in each individual case. You ask why do our experiences have the form they have. That is just what my paper is all about. I think we do agree that your experiences are causally associated more directly with your own body/brain, than with some aspect of nature such as the record player out there that is playing the Verdi opera that you are listening to and hearing, or the performers and musicians that produced the music in the first place. There is a causal chain going back to them, but the aspect of nature most directly associated with your experiences is your own body/brain. I do not think that, in talking to you, I need to produce an argument for that. But why do you hear the music rather than experiencing the patterns of neural firings in your body/brain directly as patterns of neural firings If one follows my notion that your experience is the "sound" of vibrations in your brain, then all that is needed is a particular causal mapping that maps the acoustical vibrations in your ear to some appropriate corresponding neuronal vibrations in your brain, and you would be able to hear the music. Of course, there can be a tremendous accompaniment of other vibration brought about by associations. So our experience can be correspondingly complex. But this possible vibration-sound association is subsidiary to my main argument. The key premise of my theory that the each human experience is associated with a collapse that actualizes a template for action, of the body/brain, and that the experience is the experience of the intention of that action, which can be either to intend or to attend, where the later is intention to update of the body-world-mind schema, which is the brains representation of the body-world-mind. The "mind" part is an extention of the body-world part, and it includes our more abstract ideas. The fixing of an intention is represented in the physical realm [i.e., in the Hilbert-space representation of certain statistical properties of nature] by the actualization of a projected [i.e., intended] state of the body-world-mind schema. The natural selection is supposed to have coordinated the experience (i.e., the knowing) associated with the actualization of a body-world-mind schema to the mechanical evolution that follows from the causal structure of the body-brain in such a way that the caual effect of the actualization of a body-world-schema that is connected to a certain experienced intended action will, under appropriately directed attention, generally be to produce a subsequent state of the body/brain that is likely to have a collapse that actualizes a body-world-mind schema that corresponds to an experiencing of the intended action. It seems to me that only in this way will our experiences be able to play the executive role in the control of our actions and thoughts that they appears to play. Natural selection ought then to hone and mold the connection between the experience and its effect, which is the collapse that brings the wave function of the body/brain---and more specifically the body-world-mind schema, which is the brains representation of the body-world-mind---into this causal accord with the subsequently This naturalistically generated connection between experiences, collapses, and templates for action formulated in terms of the body-world-mind schema explains the connection you required, namely the fact that our experiences seem to be direct knowings of the (perceived or intended) state of our ``body'', as we feel it, or the state of the ``world'' as it is classically conceived, or of ones ``mind'', or of these things in some combination. The collapse is to a state of the brain that actualizes the brains representation of the body-world-mind. But you may ask: how do we get from the actualization of THE BRAIN'S REPRESENTATION of a body-world-mind to an experiencing, not of this representation by neural patterns of activity, but of this other presentation in terms of sounds, colors, perceptions of geometric forms, feelings of warmth and coldness, etc. Even given the presumption that the two representation are isomorphic, why do we experience the experiential qualities rather than their ``physical representations in the brain''? That is the question I have just answered. But this answer hinges on understanding what is real and what is not. In classical physics one imagines that the physical world is real, in and of itself, with no help or assistance from anyone's thoughts about it. But in quantum theory it is the knowings that are the real things, and the knowings associated with human body/brains have the qualities that we human beings experience. These knowing are connected to each other in a way that can be codified, at least in some good approximation, and at least with regard to certain statistical regularities, in terms of a mathematical formalism that involves such notions as spacetime and Hilbert-space vectors, and that allows connections between our knowings to be represented mathematically, at least in some useful approximation, in terms of changes in a certain Hilbert-space vector that can be imagined to represent an objective state of knowledge that each knowing, human or otherwise, contributes to. But just as the mathematical representation of the physical world provided by classical physics turned out to provide merely a partial characterization of nature herself, so also should the Hilbert-space representation be understood as basically not the real thing itself, but rather a partial characterization that captures certain statistical PROPERTIES of the real thing. On the other hand, our human knowings are PARTS of the real thing.