Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
We encourage researchers to utilize the full-text search on this page to navigate our oral histories or to use our catalog to locate oral history interviews by keyword.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Abner Shimony by Joan Bromberg on 2002 September 9 and 10, Niels Bohr Library & Archives, American Institute of Physics, College Park, MD USA, www.aip.org/history-programs/niels-bohr-library/oral-histories/25643
For multiple citations, "AIP" is the preferred abbreviation for the location.
In the interview Shimony discusses his undergraduate years at Yale in mathematics and philosophy; influence of C. S. Peirce, A. N. Whitehead; reactions to Hume; studying under Robert Calhoun and Paul Weiss; the bases of Shimony's physical realism; Army service at Ft. Monmouth, 1953-55; physics Ph.D. at Princeton; reading EPR; interaction with Eugene Wigner; teaching and doing research on the philosophy of quantum mechanics at MIT in the 1960s; first reactions to Bell's 1965 paper; collaboration with J. F. Clauser, M. A. Horne, and R. Holt on tests of Bell's inequality; the 1970 Varenna summer school on Foundations of Quantum Mechanics; the researches on hidden variable theory and on quantum mechanics of von Neumann, G. Mackey, J. P. Vigier, C. Piron, J. M. Jauch, E. Specker, and S. Kochen; the metaphysical implications of quantum mechanics: potentiality and nonlocality; the search for non-linear modifications of quantum mechanics; neutron interferometry; interactions with C. Shull, A. Zeilinger, and D. Greenberger; devising measures of entanglement; plans for future research.
I think we might want to start with the 1950s. Does that sound right with you?
I will inevitably go back quite a bit earlier than the 1950s.
Well then let's start where you might go back to.
Yes, I will try to do everything chronologically. I'll answer your questions and then there'll be some references to earlier times. But look, here question one is about Max Born's Natural Philosophy of Cause and Chance and whether it first turns you to philosophy of physics. That's not true. It did influence me very much, and it sort of triggered my decision to go back to school and get a doctorate in physics. But it certainly didn't turn me to philosophy of physics. What happened was in the winter of 1952–53, I was working on my doctoral thesis in philosophy, which was on probability. Now, it was on probability in the sense of inductive logic, reasonable degree of belief.
I was an eager student and I read literature on probability, including things that went beyond what I really needed for the thesis, and I did some reading on probability in physics. For instance, I read about ergodic theory, which I didn't use in the thesis, but I was interested in it. ergodic theory is one of the possible foundations of probability. I read Born's book. I don't think I made any use of it for the thesis, but it was fascinating. Born is a wonderful expositor, and I became very interested in both classical statistical mechanics and quantum mechanics. Not that I hadn't been interested in quantum mechanics before, but my interest was revived.
Then I was in the process of typing up my thesis (I typed the technical part and my wife, Annemarie, typed the prose part) and I told her after I read Born's book, "When I finish this thesis and get my doctorate, I'm going back to school to get a doctorate in physics." Any normal wife would have said, "It's about time for you to get a job." She didn't say that. She said, "If that's what you want to do, that's what you should do." I thought it was wonderful. I told her, "That was your finest hour," in Churchill's phrase. So Born's book didn't turn me to physics. When I entered Yale in 1944, I declared, as my prospective major, physics. As a high school student I read a lot of popular science.
Not just physics; I was very interested in genetics. I read Thomas Hunt Morgan's book on genetics. I thought that was wonderful. But I never remember thinking that I would like to be a geneticist. I always kept coming back to physics because it seems to be the basic science. I very stupidly never took a course in chemistry when I was an undergraduate, nor as a high school student, because I thought chemistry was derivative from physics. It is and it isn't. Of course, in the process of making applications to molecular systems and crystals and polymers and so on, lots of new ideas came in, and I had to learn a little chemistry later on on my own. But I certainly was in the grip of the reductionist dogma that [once] one understands physics one has all the natural sciences. We'll get to that. I have various modifications of that later on.
Anyway, when I entered Yale I wanted to be a physics major, and my physics course the first year didn't turn me on very much, but I liked the mathematics course quite a lot, and I changed my major to mathematics. I think, as a matter of fact, the mathematics department at Yale was very good at that time. Then I took my first course in philosophy, which was "History of Western Philosophy." I happened to get into the second semester, "Modern Philosophy," taught by Robert Calhoun. Have you heard of him?
If so, I don't remember.
No, he's very little known outside of Yale; he didn't publish very much.
Unless you mentioned him in your—
Well, I mentioned him as a great influence in my big introduction to volume one of Search for a Naturalistic World View. He was charismatic. I think he was the only charismatic teacher I had. In fact, he was so good I went up to the Divinity School his primary appointment was in the Divinity School and I took his course "History of Christian Doctrine," which I had no intrinsic interest in; I just wanted to hear him talk about it. I tell the story in the introduction, thinking about the homoousia/ homoiousia controversy, as I rode my bicycle down from the Divinity School, and I rode the bicycle into a curb and was thrown from the bicycle and I had shock.
Let me write down what that controversy is because the typist can surely not pick it up from the tape.
Not in Greek because I don't know Greek. The words are almost spelled the same. "Homo" means same; and "ousia" means substance; and "homoi" is "similar" and then "substance". The question that troubled late third and early fourth century Christian theology was whether the son was of the same substance as the father or a similar substance. The Arian heresy maintained that their substances were similar, and Athanasian doctrine, which became the orthodoxy, was that they were identical, and that led to the doctrine of the trinity. Now this was not what I was interested in. I was interested in anything that Robert Calhoun said. So I had my accident thinking theology on the way down. But Calhoun actually, at one point, influenced me on substantive philosophy. I believe the first time I heard a discussion of Whitehead was in Calhoun's "History of Modern Philosophy," the second semester of "History of Western Philosophy." And I still remember Calhoun saying something like this, "In every generation there is some philosopher with primacy, and in our time, in my opinion, it is Whitehead." I thought I would follow that up. Now, when did I? I think I tried reading Science in the Modern World, but Science in the Modern World was the first of Whitehead's expositions of his mature philosophy, and it's a very bad exposition.
I read it a long time ago. Too long to remember.
Well, it has four or five very nice chapters on the history of thought the Greek achievement and then rationalism, the romantic reaction. All that is very readable. Then he uses the romantic reaction, which insists that mentality is implicit in nature, or at least some of the romantics certainly thought that. He uses that to introduce his own version of mentalism, from the philosophy of organisms. But his exposition of his own philosophy is almost incomprehensible. You're drawn in by this very nice, very readable discussion of the history of Western thought, and then abruptly you have this giant cliff to climb. However, that was just the beginning of my involvement with Whitehead. Paul Weiss came to Yale I think in the Fall of 1945.
He had gotten his doctorate with Whitehead, and I had a course with him in which we read Whitehead's Adventure of Ideas. Science in the Modern World was written, I think, in 1925, Adventures of Ideas something like 1934 or 1935. By that time Whitehead had worked things over and had developed a much better way of expressing his philosophy, and I was very impressed. I became sort of a Whiteheadian, so my Whiteheadianism had some input from Paul Weiss. I was a kind of disciple of Weiss for a while, but never completely a dedicated disciple. I remember when Weiss' book Nature and Man came out. I wrote a review of it for the Yale literary magazine called "Homo machina", which was critical of certain things in it.
Weiss, who thought I was disciple, wrote a reply to it with the Latin title "Et tu, Shimony" that I had stabbed him. [laugh] Well I had. But one of the things that I learned from Weiss, which I'm very grateful for, that remains after very little of the doctrine remains, was to be self confident and think things out for oneself. In other words, I was a disciple to the extent that I learned it's not so good to be a disciple. [laugh]
All this time you're still a mathematics major?
No, I switched then to be a joint philosophy and mathematics major. That existed. The thing is, as an undergraduate I was interested in everything, and I thought philosophy enables me to get a view of the whole world. For instance, I was very interested in political philosophy. I was as green as one can be. You can't imagine how green, but I will give you an example of how green I was. I took a great course, "History of Political Theory," with a man named Coker. No importance for your purposes, but it was a delightful course. We spent quite a lot of time on readings on the social contract and criticisms of the idea of social contract. Now, bright little boy "thinks". I'm not going to understand the social contract unless I understand what a contract is, so I went and sat in on a course in contracts. I was just an undergraduate.
I think the class had started several weeks before, so I was plunging into the middle of it. Then I listened to their analyses of cases of contracts, and Professor Kessler, I think his name was, would go from one student to another asking what this means, what this means, what the law is on this, "No, no, no, no, no." Then somebody would have it right. I couldn't see the difference between what was right and what was wrong. They all sounded the same to me. And this was supposed to be legal logic.
But I was a logic student. I had taken Fred Fitch's courses in logic and I studied a lot of logic by myself, and I couldn't see where the legal logic had anything to do with logic the way I knew it. Of course, by now I understand there is some type of rationality in back of this. There's the proper use of precedents, there's the proper discrimination of instances which ones fall under a law and which ones don't. I didn't understand all of that. I'm just giving you this as an instance of how green I was.
On the other hand, it sounds as if you had a really exciting time at Yale.
I had a very exciting time as an undergraduate, but you see why I was drawn to major in philosophy? Because it enabled you to do everything. Of course, later on I did learn that if you knew everything and don't know anything very well, there can be some superficiality in what you do. I kept being nagged in the back of my mind by the rigor of mathematics and the magnificent achievement of physics. I did continue to take a few physics courses. I took a course in classical mechanics and a graduate course in quantum electro-dynamics, the first course in which I had an exposition of special relativity theory. So I continued to be interested in physics.
I never took Margenau's course in philosophy of physics, which was a mistake. I'm sorry I missed something. I did come to know, I think in about senior year, Adolf Grunbaum. I may have met him at Weiss' soir‚e, but I did meet him and I was very impressed by the way in which he put together his knowledge, particularly of relativity theory, with philosophical questions. So I saw that it could be done. And I remember it was always in the back of my mind that I would someday do some philosophy of physics, even though the work I did for a while was more in philosophy of mathematics. With Fitch I did logic, and then Fitch taught philosophy of mathematics of a kind of platonic sort. Closer I guess to G”del than to anyone else. I think I talked in my preface about G”del's influence on my thinking.
Yes, and you talk about Margenau too, it seems to me. You didn't take his course?
No, I just knew about his work, but I didn't He certainly didn't influence my thinking in philosophy of physics. I just didn't have a course from him. But I probably had some indirect influence via Grnbaum, but I'm not sure. Anyway, I knew from their example that there was a field that's fascinating in putting together physics and philosophy, and I had it in the back of my mind. I even did a terrible thing; I took Reichenbach's Raum-Zeit-Lehre along with me on my honeymoon. [laugh]
What did your wife do to you?
She was angry with me.
Oh, really?
Yes, she didn't like that. [laugh] But that shows that I was sort of preoccupied with this idea of philosophy of physics. Anyway, it always was in the back of my mind. Let me tell you a few more things about undergraduate influences, okay? Very important for—
Well, we're still here but that's fine. Because I think that what is in the literature is not very clear about your undergraduate work.
Yes. One, in a way, negative influence and then one very important positive influence. The negative influence was Hume. I read Hume's Inquiry and it troubled me as, after all, it troubled very many people. I was troubled by his argument concerning necessary connection, and I was troubled by his skepticism about induction. I knew it couldn't be right. If it were right, we couldn't get by with the inductions that we use in ordinary life and we couldn't have the magnificent achievements in science. So I was using a kind of crude hypothetico-deductive method against Hume. That is, if Hume were right, then the inferences that are made in the natural sciences would— If they worked sometimes it would be coincidental, and a series of coincidences, as long as those presented to us by the natural sciences has infinitesimal probability.
So I had a kind of rough hewn argument against Hume. But I kept looking around for solider arguments. That is, after all, I felt that Hume really should be faced on his own terms. To some extent, Whitehead supplied an answer to Hume. Whitehead has a lot of discussion of Hume in Adventures of Ideas and much more in Process and Reality. The ingression of one actual occasion in another is a strongly non-Humean step at his philosophy. For Hume, each matter of fact is what it is, logically independent of any other matter of fact. For Whitehead that's not so. Matters of fact can be ingredient in other matters of fact. There's a kind of entanglement of matters of fact.
When I use that word, of course I'm anticipating a term I learned from Schr”dinger, and it was important in physics. But going back, using that terminology, I would say Whitehead's philosophy answers Hume by saying there is such a thing as entanglement in matters of fact. It doesn't mean that one has a monism. I think Whitehead was aware that his theory of prehension the ingredience of one occasion in another might lead to one substance in the world. A theory like Spinoza's. He tries to balance things. One epigrammatic remark he makes I think in Process and Reality that the formation of a new occasion is a monism passing over into a monadology. What does he mean by that? What he means by that is that every new occasion prehends all occasions of its past, so they all enter into it, and that's a sort of monistic treatment. Not of all occasions for all times, because one new occasion doesn't prehend a future occasion, but only prehends the past. But it's almost monistic/semi-monistic. Then when they're all integrated, some somehow dismissed, some enhanced, some of them compared, some of them integrated with new imaginations and conceptual prehension that's a monadology.
So Whitehead was trying to have a balance between a point of view in the world where there were internal relations among entities. After all, he grew up in Victorian England, which was very strongly influenced by the neo-Hegelians Bradley and Bosenquet. He mentions Bradley at one point and says he's influenced by Bradley, but obviously his influence was selective. So at one pole of his thinking is this monistic element inherited from the British idealists. On the other hand, there's a strong influence from Leibniz, a monadologist. And also, after all, Whitehead was a scientist. He was very impressed by the evidence for the atomicity of matter, and he knew about Planck's work. He knew about the early quantum mechanics. He knew about—
Einstein?
Well, Planck and Einstein. He didn't know this in detail, but he knew that light is granular. In fact, his picture of elementary particles is a funny kind of granularity because it's granular not only spatially but also temporally. One electron is a chain of occasions. I think he thought of one wavelength in the wave train as being an occasion. I don't know that he ever mentions the de Broglie.
What did he do with the Fourier series?
He never mentions it. He must have known about it but he didn't integrate it into his thought. I wrote an article, it's in volume two of my collected papers, on quantum theory and philosophy of organisms. I think at one point I say, "Well, he's very vague about radiation that happens to have many Fourier components." He doesn't say. He talks as if any radiation really has pure frequency, and therefore you can talk about each successive period as an occasion. So a photon is a temporal chain of occasions.
Now is this all stuff that you were studying at Yale, or is part of this Whitehead stuff that you began to study only later?
Well, I wrote a paper called the "Animate Individual" probably in either second semester of sophomore year or first semester of junior year for Weiss, and it was very much influenced by Adventures of Ideas. I liked Whitehead's mentalism. I liked the idea that mentality of some low variety was present at a very primitive level in nature, and—
In animate, or in both inanimate and animate?
Well, he doesn't think that the occasions that make up the life chain of an electron are conscious. Consciousness is a very high-level kind of experience, but the occasions of an electron are proto-mental. He does apply the term "experience" to them. In fact, his doctrine, one of the ways in which he answers Hume, is to say postulate re- experiencing when one occasion prehends an earlier occasion and reenact the experience of that occasion, that might be one particle being scattered by another particle. In Whiteheadian terms, that would be somehow the new particle the new occasion, the occasion in the life history of the second particle has ingredients from the occasions in the life history of the scatterer. But what kind of experience?
Well, we don't know, because the experience we know is either conscious experience or the dim experience as we fall off to sleep or in drunkenness or under anesthesia, and maybe we have some evidence of subconscious experience from a Freudian analysis, and we can conjecture that lower animals have some of that. Well, then try to extrapolate to amoeba, try to extrapolate down to molecules, try to extrapolate down to electrons. Do you know what you're talking about when you attribute a kind of experience to them? In fact, there's a famous criticism of Whitehead by Lovejoy. I think it's in "Revolt Against Dualism." He considers Whitehead to be one of the rebels against dualism, because the dualists maintain that physical reality and mental reality are toto caelo different. Both Descartes and Locke agree on that.
But the rebels against dualism, like Leibnitz and then in our time, Lovejoy's time, like Whitehead are, say, "No, there's something in common to mental reality and physical reality, because those entities which we call mental have a kind of experience." Now you see, that appealed to me very much because I was a strong evolutionist. It seemed to me if creatures like us are evolutionary products, then our mental faculties must be products of evolution, not just our bodies. If our mental faculties are products of evolution, then there must be something mental-like from which the faculties evolve.
The evolutionist part of your thinking is already being formed in your undergraduate years?
In my high school years.
In your high school years.
Oh, long before. I was a strong evolutionist in Memphis, Tennessee, and I fought with some of my teachers. In fact, I had to change from one geometry class to another because I defended evolution in a geometry class.
That's very interesting, because evolution is something that is a constant in your writings.
Oh yes, it just seemed to me so obviously true the first time I heard about it. I never ceased to be an evolutionist. But Whitehead, of course, was writing mainly before the days of pre-biotic evolution. I think Oparin's work was around 1920, but Whitehead never would have heard of Oparin. Whitehead didn't read much anyway. He certainly didn't read obscure authors like a Russian chemist. But it's just part of Whitehead's philosophy of organism that high order mentality is the evolutionary development from low order mentality.
Are there any other people besides Grnbaum who were students at Yale who were very important in your development, that we ought to mention at this point?
I was a friend of Ted and Dave Calhoun, who were a son of Robert Calhoun. Ted Calhoun was a kind of Whiteheadian for a while, and then he had a conversion. I think he had many conversions in my time of acquaintance with him. I couldn't understand that at all. Either Ted or Dave went to England for a year and became impressed by English analytic philosophy, which was a technique of somehow dissolving philosophical problems by linguistic analysis. I hated that. I thought there were real problems. [Tape 1, Side B] The mind-body problem is a real problem; the problem of whether there are things independent of our thoughts, contrary to Berkeley, that's a real problem. The problem of induction; that's a real problem. You can't dissolve these by words. I mean, that was my emotional reaction.
But it took a long time before I was able to articulate better In the course of articulating, I made some use of a philosophy that I had read, but read without internalizing, that of Plato. I read Plato very early in my career at Yale. It didn't enter deeply into my thinking. Later on (I think this may have been partly the result of courses at the University of Chicago, partly my long and close acquaintance with Howard Stein) I realized that Plato was on to something very profound. That is, the clarification of ideas and positive progress in the content of knowledge go hand in hand. And neglecting that is one of the things that is shabby about English analytic philosophy, in spite of the real brilliance of some of the leaders. I have no doubt that Wittgenstein was a real genius. He was a brighter man than most of the people I agree with, but he may be a wronger man than they were. Partly it's a neglect of the way in which clarification and positive progress in knowledge go together. Much later on I read three books by Hao Wang. I only read them in the last few years.
I must have read them around 1996 to 1998, after Wang died. There was a memorial for Wang, and I was asked to talk about Wang's philosophical ideas, because most of the people in the memorial service were going to talk about his contribution to logic. So I read three books of Wang's: one called Beyond Analytic Philosophy, one called Reflections on G”del, and one called A Logical Journey, from G”del to Philosophy. He has a real polemic against both the British analytic philosophy and logical positivism for somehow detaching semantic analysis from the content of mathematics and the content of the natural sciences. Now, to be sure, the Vienna Circle is very interested in the natural sciences. Carnap had a degree in physics, and Carnap knew a lot of classical physics. One of the purposes the reforms the Vienna Circle envisaged was to clear away nonsense from our language in order to have a language suitable for the natural sciences, and a language in which one can reason with rigor in the natural sciences, and present the discoveries of the natural sciences without a metaphysical overlay.
So it's true that they were interested in the natural sciences. Nevertheless, when you look at the kind of semantical analysis earlier, syntactical analysis which Carnap did, and then the semantical analysis, the tools used are hardly influenced at all by the contents of the natural sciences. What I learned from Howard Stein, whom I was very close to, was that you cannot just semantically analyze, say Newton's treatment of time and space, and his idea that there is such a thing as absolute motion; in particular, absolute acceleration. (The treatments of absolute velocity and absolute acceleration have to be different, which Newton didn't get quite right. But you cannot analyze these just by reference to ordinary use of language or ordinary experience, as Mach tried to do.
Mach, after all, was a precursor of the Vienna Circle. The Vienna Circle was, for a while called the Mach Verein. When Newton did the water bucket experiment, which he actually performed it wasn't just a thought experiment, he did it and when he analyzed the motions of the planets around the sun, his detailed treatments of these motions were inseparable from his clarification of the idea of an absolute time and an absolute space and an absolute acceleration. Something he didn't get, because his mathematics was not advanced enough it took the late nineteenth and early twentieth century to realize that one could have absolute acceleration that would be the same in every inertial frame even though velocities would be different in different inertial frames.
Newton didn't quite see that. Nevertheless, the exposition of time and space, and the terms that he used, like "flowing equably without reference to any external thing", (which Mach dismisses as nonsense) are not nonsense. It's not nonsense if you watch how Newton associates these rather elliptical phrases with definite processes of analysis, and with definite bodies of physical law.
Now, it sounds as if we're at Chicago by this time. Are you already—
See, I told you I have to be apologetic, because I don't always know when I learned certain things. I suspect that in many cases I had earlier dim adumbrations of points of view which I never was able to articulate until later on, partly because I then had more reading, partly because I had good influences. So I came to college very much oriented towards the natural sciences. I never ceased to be interested in the natural sciences. I remember once at a soiree at Weiss' house when Mrs. Weiss said, "You're interested in time. You will learn more about time waiting for the telephone to ring when you're expecting an important call than you will from the theory of relativity."
I remember being very disturbed by that, and thinking, "Yes and no." There is something right about what she said. That is, the elementary phenomenology of time is indispensable to understanding time. But also the subtle thing that is found out by the Michelson-Morley experiment, that also is indispensable for the understanding of time. And that two events that are simultaneous in one inertial frame are not simultaneous in another inertial frame, that also tells us something about time.
What good philosophy has to do is keep in balance the phenomenological component and the scientific component. I'm saying that now; I can say that after 50 years. But it took me a long time to articulate what I just said.
Now, at this point you're coming to Yale. It's at the very end of the Second World War, and you're going to live through—
Not quite. No, I came in June 1944. I expected to go in as a soldier. I came to Yale at age 16. I had left high school a year early to get a college education in before going to war. But, as a matter of fact, the war ended before I was old enough.
I'm wondering, just looking a little bit outside the actual courses and books, at the context of what was going on in the world, what influence that might have had on you at this point. You lived through the atom bomb at this point, you lived through Roosevelt's death, and Truman coming in. I'm wondering if that had anything you do with anything that we ought to be talking about.
I am sure that the issues of World War II very much influenced my whole ethical and political point of view; no doubt. But I don't see how those things had very much influence on either my epistemology or my ontology or my ideas of scientific methodology. It seems to me they didn't overlap very much. I'm certainly not a professional ethicist, but such thoughts that I have on ethics are partly naturalistic in character. I believe that we can't understand ethics properly without looking at human beings from an evolutionary point of view. But how do you derive an "ought" from an "is" by use of evolution? I'm not sure you can do it. Even though there is a naturalistic component in whatever ethical theory I've tried to articulate, it doesn't completely handle all the basic problems of ethics.
That's fine. I just want to make sure that as we go along, if there's something outside the—
I was trying to get to something very important that happened in my undergraduate time and didn't get there because we got off on other matters. I was trying to tell you the philosophers who influenced me most, among teachers but also among philosophers who were not my teachers. I told you that I learned about Whitehead from Calhoun and then more from Weiss. I learned about G”del from Fitch. The third man who influenced me more than either of those was [Charles Sanders] Peirce, and I didn't know about Peirce until I had a course with Weiss.
Weiss and Hartshorne were the editors of the first six volumes of Peirce's collective papers. I read lots of Peirce's papers, and I loved Peirce. I love Peirce to this day, and I think my point of view is closer to Peirce than to anyone else. Let me just tell you a few of the things. First of all, he thought that an adequate philosophy has to take into account science. He was himself— He was a working scientist on some parts of science that I have very little interest in. He was a kind of geographer. He was interested in the variation of the Earth's gravitational field at various places on the surface of the earth. He said, "I was a pendulum swinger." [laugh] Looking at the period of the—
Humboldtian science.
That's right. The thing is, he knew the science of his day very well. He was very impressed by the kinetic theory of gases. So to say that I didn't get interested in statistical mechanics until reading Born's Natural Philosophy of Cause and Chance couldn't be more wrong, because I already knew about the kinetic theory of gases from Peirce. I certainly heard about kinetic theory from other people while I was an undergraduate. After all, there were people teaching at Yale who had been students of Josiah Willard Gibbs. I was one generation away from Gibbs. That's an amazing thing to feel.
So I knew dimly about statistical mechanics at the time. The idea of statistical mechanics, which is to understand the macroscopic behavior of composite things in terms of the dynamics of the components, is a very powerful idea. It's an old idea. It goes back to Democritus and Lucretius. The only thing is they didn't have the dynamics. All they had was the idea of the parts and the whole, but they had no idea of the dynamics which connects the parts and the whole. But in what was almost my time one generation or two generations before my time there was a detailed theory which enabled one to analyze heat flow in terms of the behavior of molecules, and the behavior of molecules governed by Newton's law. This is an amazing thing.
And I expect that people at Yale thought about Gibbs and felt that you were sort of [inaudible].
[Henry] Margenau did. I know that, and I heard little bits of this from [Leigh] Page. Page was the theoretical physicist at Yale. I heard about these things, even though I did not have a course in philosophy or physics as an undergraduate. It was a great gap. Anyway, a little bit more on Peirce. Peirce had a philosophy which was grounded in science. In volume one of the collected papers he said something wonderful about why he has devoted so much of his life to the study of history of science.
He says, "So that I may not neglect any pathway to the truth" that is, if I study what was historically done, and what was historically done was successful, that must be a pathway to the truth. I cited that passage in my essay on Kuhn. Kuhn in my opinion studies history of science for the wrong reason: in order to bypass, in order to emphasize the disagreements instead of to emphasize the systematic approach to the truth, which is I think the signature of the progress of science. Anyway, let me go on a little further. What else was impressive in Peirce? He was a fallibilist. He said no proposition can be asserted with absolute confidence. Does that make him a constructivist or deconstructivist or relativist? Not one bit. At the same time that he was a fallibilist, he believed very strongly in reasonable degree of belief, and the use of probability theory.
He had many essays on inductive method. Now, lots of them are his own redoing of the hypothetico- deductive method. So as I say, from way back, from high school days, I must already have dimly known, as anybody who studies the sciences dimly knows, about hypothetico- deductive method. Everyone knows you don't deduce the laws of nature from phenomena, but by taking those laws as conjectures and deriving consequences and comparing the consequences with the data, one can confirm or disconfirm. The whole machinery of probability theory is to turn that very qualitative framework which I just sketched into something quantitative. So I was very impressed by this, and I thought that what Peirce had was one of the answers to Hume; a useful answer if you can't answer Hume by dismissing Hume's statement that any impression is logically independent of any other impression. That may or may not be true.
As I said, Whitehead really had a different analysis of Hume. His different analysis was that one experience can be ingredient in another. But what Peirce is saying in his essays on scientific method is quite apart from ultimate ontology. You have an answer to Hume by use of probability theory. Now, as far as I know, Peirce doesn't mention Bayes, but he must have known it; Peirce knew everything. He certainly knew Laplace, and Laplace makes a lot out of Bayes' theorem. So it's possible that Peirce did. All I know is that when I learned about Bayes' theorem at the University of Chicago— I read Nagel's little book on probability theory and then I took a course in probability with Carnap it seemed to me that the Bayesian formulation of probability is the natural framework within which you can do the hypotheto- deductive method.
I was rife for incorporating the Bayesian point of view into my work. Now I would have to jump ahead to my doctoral thesis, where I have some modifications of Bayesianism.. Anyway, reading Peirce certainly prepared me for the point of view that probability is absolutely essential in our epistemology, and that judgments of very high probability in favor of one conjecture and against another are quite compatible with his overall fallibilism. So I thought he had the makings of a balanced epistemology. Balanced between dogmatism on one hand, (and you want to avoid dogmatism on the one hand, which of course was Hume's greatest enemy), and excessive skepticism on the other hand. I really felt that Peirce came closer to striking the balance of these two polar errors than any philosopher that I have ever read before. Now let me go on to a few more things in Peirce. He also really anticipated so much of the epistemology of the latter half of the twentieth century.
He has an essay called "Critical Common Sensism" and he was against taking any statement of fact as unalterably and undeniably true. Any perceptual report may be erroneous, and reasons may come up for us to doubt our senses, to doubt our evidence, and therefore to revise the perceptual reports. So that was essential. Then, unlike Hume in the last section of The Inquiry, Hume's chapter on our knowledge of the external world, for Peirce the existence of things with careers of their own is almost beyond doubt. That is, one can have no real doubt of it. We can be fallible about that too, but any doubt that you express as a skeptic, your heart's not going to be in it. You really have no real qualms in asserting that the tables and the chairs and the sun and the moon and other people have careers independently of you.
So he's an objectivist. Now, let me just go on, going very quickly through some theses of Peirce which add up to a wonderful world view. In current philosophy of science, you have scientific realists who maintain that not only the sun and the moon and the tables and chairs are real, but also electrons and protons and photons and so on are real. You also find very frequently in this package presented by the scientific realists a general physicalism. That is, you believe that these have independent existence just because the physics formulated in terms of things with these independent careers has been so successful.
Another application is the hypotheto-deductive method. But if the world is made up of these things, there's no place for an independent mentality. There is mentality, we know that phenomenologically. But there is mentality just as there are can openers. That is, pieces of matter with particular configurations which are adapted to particular purpose.
Like neurons?
That's right. Neurons are particular configurations of matter with problem solving ability, with adaptabilities of various sorts, and there is nothing in the behavior of a neuron that cannot be understood in terms of fundamental physics. We don't quite have it yet, because we don't even have the physics of macromolecules completely. But the problem is generically of the same sort. Peirce will have none of this. Peirce is an objectivist, a scientific objectivist, but he thinks, like Whitehead, that the ultimate entities in the world have a mental component. One can be an objectivist, and nevertheless a mentalist . And this is very important.
Among people who oppose scientific realism were romantics or religious people who one way or another were influenced by Berkeley. There were Wordsworthians who saw mentality throughout nature, and then there were religious people who followed Berkeley for Berkeley's own reason that Newton is the main enemy of a spiritual religion. Now, Peirce's mentalism is not of that sort because the mentality that he sees as pervasive through nature is not the mentality of human beings, nor necessarily the mentality of a super rational being like a god.
It is like the Whiteheadian mentality low order. There's a continuum from where we stand, or maybe from higher up, beyond the point where our imagination can even conceive. But the whole evolutionary point of view demands that we, with our faculties, must have come from something that was endowed with proto-mentality.
Now this suggests to me to ask you, are religious views having any role in the formation of your thinking at these early stages?
Not very much, except— It's more the other way around. That is, what does my general philosophy tell me that overlaps with my religious background? It tells me something. One is the importance of a sense of wonder. When I look at the most inspiring passages in the Bible, they are passages which celebrate a sense of wonder. The heavens declare the glory of the Lord. Wonderful. Beautiful. In the last book of Job, God comes to Job out of a whirlwind and asks him a series of questions. Where were you at the creation of the world? Can you draw out leviathan with a hook? And so on.
And the whole book is an expression of wonder at the world. Well, my feeling is no matter how far off the dogmatic content of my Jewish religion might be, or anybody else's religion might be, the sense of wonder which was inculcated, or ought to have been inculcated and it still is inculcated by some people, is very important for spiritual nourishment. So that's one thing, the natural wonder. The other thing is that once one has a mentalistic view of the components of the world, if the world is suffused with mentality, then there is something analogous to a god who is spirit. Now, I don't like any of the dogmas. Jewish, Christian, I don't know many others, that seems to me to say it right. But a view of the world which is not just physicalistic, which sees mentality suffused throughout the world, and makes it possible for our human mentality to resonate with that suffused mentality, that seems to me, first of all, likely to be the true story, not just a story made up to get one solace.
But it's very likely to be something like the truth. And then as a byproduct, it may give one inspiration. So I'm not saying that my religious upbringing led me to these philosophical views; I don't think it did. But the philosophical views the mentalism, the scientific objectivism those have enabled me to sort of come to terms with my religious upbringing, but to come to terms with great latitude with complete freedom from individual doctrines. And so it was worthwhile going over influences in my undergraduate training. Because I came to quantum mechanics with these philosophical ideas in mind, and I never felt that I had to reverse any of them. What was involved was fine-tuning. Can I talk about that for a moment, or do you want to do that at another time?
Sure. No, I think that's absolutely relevant.
Through much of Peirce's career, he defended the idea that the primary sense of probability is the frequency interpretation of probability. So probability really refers to an ensemble of similarly prepared systems. Then somewhere around 1890 or so he wrote a series of papers saying, AI was all wrong. I was immature, unripe". The frequency theory, the frequency interpretation is derivative; it doesn't apply to the individual case. You may have an ensemble of one entity and no replicas are made. You may destroy the die and, nevertheless, when the die was in one piece it had a certain probability of turning up one, two, three, four, five or six.
Destroy it; you have no ensemble anymore. Therefore, the fundamental notion of probability has got to be a would-be that applies to an individual case. So Peirce had the idea that a statistical interpretation is simply derivative. When one has the same would-bes, or a large number of entities similarly prepared, then one has a statistical interpretation of probability. But the would-be is well defined in an individual case, even though you can't test it without having the ensemble. It's meaningful in the individual case. That's exactly what Popper said in his papers on propensities in the 1950s. Now, my feeling is that Peirce was the first to articulate the propensity interpretation, though it's clear to me that Laplace had the idea but he didn't articulate it so well. Now, where does a propensity interpretation of probability fit in a classical world view? That's a real problem for Laplace, because Laplace was a strict determinist when all the data are given, all the boundary conditions, all the conditions of the whole universe. That's why for Laplace it's not entirely clear that he had a well-formulated propensity interpretation.
But Poincar‚ did. He didn't call it propensity interpretation; he called it chance. Poincar‚ was the man who first developed the ideas of chaos. The idea that even if the world is deterministic we don't know for sure, but even if it is that doesn't mean that if you have only approximate knowledge of the initial conditions on a system, that you will be able to say approximately what the system will be doing ten seconds from now or a year from now. It is characteristic of chaotic systems that if you're off by an ever so little bit from the exact initial conditions, you will be off by ever so much in the final state. We now have several bodies of mathematical theory, ergodic theory and chaos theory, which investigate such things. But what Poincar‚ said was that a roulette wheel can be described objectively by probability because, unless you know exactly what torque the croupier is imparting to the wheel, and exactly what resistance is imposed on the bearings, and exactly what resistance the air will make, you don't know how many times the wheel will go around. Since any little error allows for a large range of possibilities in the final state of the roulette wheel, where it ends up, you have an objective determination of a probability distribution of outcomes. I'm mixing up terminology.
That's how chance arises in a deterministic physical system. It is deterministic but yet unstable. In a stable system, if you're off by a little bit in initial conditions, you're off only by a little bit in the prediction. In an unstable system, chaotic system or chance system, if you're off by a little bit in your initial condition, you're off by ever so much in your final condition. Now, Peirce knew all this. I don't remember that he cites Poincar‚, but he understood all this, and I think he talks something like this when he talks about would-bes. Peirce believed in addition that the world is in deterministic. He said, first of all, we have no evidence; no evidence for strict determinism.
All evidence is evidence concerning bouncing balls or solar systems, and that evidence is fitted perfectly well by laws which are off by a little bit. Secondly, even if you granted that the laws were exact, where do the ultimate initial conditions in the universe come from? They can't be themselves matters of law; that would just give you an infinite regress. Therefore, brute chance must enter somewhere in the world. The Judea-Christian creation legend makes the brute chance be the will of God in starting the whole mechanism going. The 17th and 18th century mechanists kept some of the religious point of view.
That is, God set the whole mechanism going, but they added certain things, such as the deists said that once it was set going, there was no more interference with the machinery. The theists said it is possible for God to intervene in that which He set in motion in the first place. What Peirce is insisting on is that the various mechanists and determinists of the past still have an opening wedge for something nondeterministic; namely, the ultimate initial conditions. If that is so, if initial conditions are not matters of law, then why not take a cosmology in which chance is suffused through the world throughout its career? Instead of localizing chance at the moment of creation, let us suppose that as time goes by there always is absolute chance.
Now all of this, remember, is before quantum mechanics. He did know about statistical physics. That is, he knew about kinetic theory of gases. But there were formulations of kinetic theory of gases entirely within deterministic mechanics. Boltzmann's own formulation was within a Newtonian dynamics. Now, this is what I'm building up to.
He had a point of view which was sitting there waiting for quantum mechanics to be discovered. When quantum mechanics was discovered, there was evidence that the world really does work the way he conjectured in his alternative to classical cosmology. That is, he had the point of view that stochasticity is suffused throughout nature at all times. Then along comes a theory representing particles by waves, by wave functions. When measurements are made or something else happens that requires a definite outcome, the outcome is chance.
What about [tape volume cuts out; few inaudible words] between optics and classical optics. Was that any part of what you were looking at then? I mean, I'm now thinking about Emil Wolf and that kind of thing where they—
Rayleigh and Jeans had statistical optics. But what does that mean? It means that you don't know exactly what radiation comes from the sources, therefore you have to treat the radiation by a probability distribution.
Which sounds exactly like classical kinetic...
That's right; like kinetic theory of gases. That's right. They had some problems, because when they discussed the equilibrium states and they used equilibrium ideas of statistical mechanics, they got the Rayleigh-Jeans statistics instead of Planck statistics. So there was something wrong. But it was not clear when the discrepancy between the Rayleigh-Jeans statistics and the discovery of the actual distribution of black body radiation, it was not clear what the source of the trouble was. In fact, even after Planck introduced his quantization of the radiation field, he worked for many years trying to make the picture deterministic. So it wasn't obvious that the granularity of light entails stochasticity. As things developed, the Bohr's theory of the atom has, in a way, an ad hoc stochastic element. That is, the transitions from one stable orbit to another.
Then, when Heisenberg tried to get a general mechanics that would fit known radiation laws and known phenomena, known data about amplitudes and frequencies, he sort of put in stochasticity by hand. One of Schr”dinger's motivations was to develop a wave mechanics which would avoid putting stochasticity in by hand. Then it didn't work. It didn't work because the spot on the screen is localized even though the wave impinging on the screen is spread out. He tried to explain it by a continuum theory of matter and it wouldn't work. He finally had to give in to the Born interpretation, which made the amplitude of the wave function a probability amplitude was the only way of fitting the evidence.
Now, let's come to Princeton with these things that you have in your baggage: Peirce, Whitehead...
G”del.
G”del. What was Princeton physics department like in view of all these ideas? Wheeler was there. Wigner was there. I'm not quite sure— Einstein was not—
Einstein was at the Institute for Advanced Study, not at the— I entered the physics department in 1955 and Einstein died in 1954, so I would have never have met him. I have exchanged two words with him. I said, "Good morning Professor Einstein," and he said, "Good morning" to me. We met. That was in toto our conversation. Now, I don't know why they took me at Princeton. I remember Gell-Mann was my friend as an undergraduate. He came to Yale one year after me. By 1953, he already was a very famous elementary particle physicist and wrote for me. By 1953 I had decided to go back to school in physics, though that was just when I was beginning my Army career, and IY
So you did go into the Army?
I was in the Army between getting my doctorate in philosophy and beginning physics.
I see. I didn't know that.
I was in the Army for two years, which was very good. That gave me time to read undergraduate physics. I learned some mathematical methods and I learned electromagnetism and I learned quite a bit of classical mechanics.
I see. Well what were you doing for the Army?
I was in a mathematics section, and we did calculations. It was kind of low- level work for the most part. One thing I did that turned out to be useful later on was to teach a course in information theory. I even had a lieutenant colonel in my class. I was a private. I did one piece of research later on, which I probably wouldn't have done if it hadn't been for my course in information theory. I knew Shannon's paper very well, and I wrote a critique with one of my students on Jaynes' maximum entropy principle. I'm not sure I would have been able to do it had I not studied Shannon's paper, and I don't think I would have studied Shannon's paper except that I gave the course in information theory.
Where were you, physically?
Fort Monmouth.
So you gave that course, you studied physics. Is there anything else out of those two years that—
Yes. I learned what ordinary people are like and how to get along with them. It was the one time in my mature life that I was not in a college campus. It was very important. It was very good for democratizing me.
Good. Were you already married at that time?
Yes, but my wife wasn't always there. She was on this Indian reservation part of the time. We had a little house off the post.
Now again, I'm going to ask you a little bit about the world around you. This is in the middle of the Oppenheimer affair. McCarthy. Was that of any importance?
It was. It was dreadful. Fort Monmouth was one of the places singled out by McCarthy as a hotbed of communist infiltration, and the commander of Fort Monmouth was very sympathetic with McCarthy and insisted on pulling the clearances of many, many civilian employees. There were about a hundred of them who were in what were called "the leper colony" while their credentials were being checked. They had to report to work, they could read whatever they wanted, they could play cards, but they couldn't see any classified documents. In some cases, their careers were ruined.
All were cleared except one. That's the statistic I've heard. I had one friend who had been in the leper colony. I saw him in Boston years later and he said, "It was very hard for me to get another job. I wanted to leave Fort Monmouth. It was hard to get another job because I was tainted by the accusations. Therefore, I could stay on the rest of my career at Fort Monmouth, which by that time I hated, but I was a captive employee." Now, it didn't impinge directly on me, because I had left-wing views from junior high school days, from reading my mother's subscription to The Nation, but The Nation was anti-communist, and I was an anti-Stalinist. I was a socialist. A socialist but an anti-Stalinist. So I seemed never to have any trouble myself, but I had some GI friends who did have trouble. One of them became a very famous logician. His name is Sol Feferman and he became president of the Association for Symbolic Logic.
They pulled his clearance. In fact, for quite a while he was working in the craft shop. He was a good craftsman, but could touch no technical material. Then he wanted to get out of the Army two months early, which was offered to anybody who had a job such that the beginning time of the job was fixed, so that you would lose the job if you didn't get out two months early. He was offered a job at Berkeley and started in September. He wasn't due to be discharged till November. This is one of the most stinking things I ever heard. They said, "We'll let you get out two months early if you will accept a discharge with less than honorable conditions." He said, AI won't do that.
I want an honorable discharge, because there's nothing I did that was dishonorable. All my associations, every organization that I belong to, I declared on my clearance form when I came to Fort Monmouth. There's nothing that I hid. There's nothing that you didn't know about me when you put me in the laboratory. There's nothing that I have to apologize for now." He wouldn't do it. I don't believe he was let out early. I think they kept his position for him at Berkeley. He was a case of a man who really was really under pressure. I also met a man whose name I caught as Brower. As I said, my wife was on an Indian reservation when I first came to Fort Monmouth, so I lived in the barracks. At night I used to go to the post library for reading. I remember I saw a man reading a mathematics reprint, and that looked interesting. So I talked to him and I caught his name as Brower. There are a lot of mathematical Browers.
By then I was already assigned to the mathematics section. I told him, "It's a pleasant place to work, a good way to spend your time in the Army." He seemed utterly uninterested. And we talked about other things for about a week, then he disappeared. Then I found out why he disappeared. His name wasn't Brower; it was Browder. He was Earl Browder's son. You know, Earl Browder was chairman of the Communist Party, and as soon as he filled out his clearance form and said, "I'm the son of the chairman of the Communist Party," they shipped him off. I saw him a couple of years later at Princeton he was on some business there or giving a lecture, he was a mathematician and I asked him, "What did you do after you left Fort Monmouth?" He said, "Someone asked one of the French emigr‚s that at the time of the Reign of Terror, and his answer was, "I survived.'"
Fort Monmouth is reasonably close to Princeton. Did you go down to Princeton while you were in the Army?
I went for an interview. Wheeler interviewed me, because I had a friend from Yale who was an assistant, named Jack Macintosh. He told Wheeler about me. Wheeler had philosophical interests. By the way, I saw Wheeler this summer at his place in New South Bristol, Maine, and we spent a day together talking about foundations of quantum mechanics.
Was he already in the early 1950s interested in foundations?
Oh, yes. Well, he had studied with Bohr. I think he had gone to Copenhagen and worked with Bohr for a while. In fact, he and Bohr wrote the famous paper on the liquid drop model. It's a uranium nucleus the nucleus' of an atom that underwent fission. He was a conduit of information to American physicists about the extent to which research had gone on in fission. That's not an answer to your question. He clearly was influenced by Bohr, and remains influenced by Bohr.
Now, when you went to be interviewed, did you talk about Peirce and probability and stuff like that, or what?
A little bit, yes.
What kind of things did they—
I'm afraid I don't remember. I really don't. I can tell you what I do remember about my interview with Wheeler. After I was admitted— I must have conveyed to him that I really am interested in the interplay of philosophy and physics and I believe that they go together, and he agreed with that. I don't think much more happened. My guess is that he did most of the talking; I just don't know. It probably wasn't a very long interview in any case. When I was admitted in the Fall of 1955 my formal advisor was Wigner, and Wigner said, "Why are you here? Why aren't you studying with Alonzo Church if you want to come to Princeton?" Church was the great logician of the mathematics department. I said, "I'm interested in the physics.
He started asking me some questions about physics, and I didn't answer them very well. Then he said several things; one chiding me and one not chiding but something remarkable. He said, which was chiding me, "You know, physics is not just about philosophical questions such as, "Are there atoms?" Physics wants to know specific things like the dielectric constant of water." Then he added, "And when you understand a phenomenon like that, you have an elevated feeling". I thought that was so beautiful. I just went off sort of glowing from it. There was a meeting on foundations of quantum mechanics organized under the auspices of the New York Academy of Sciences. The organizer was Dan Greenberger, a good friend of mine.
He asked me to give a sort of summary of the conference. The conference was dedicated to Wigner and I said, AI want to tell a story about Wigner that he probably has forgotten, and everybody should know. I told him about his saying one has an elevated feeling. I said it was one of the most beautiful things I'd ever heard. Wigner didn't remember, but he seemed very touched by it.
Yes, I should think so.
Anyway, with Wigner, something very interesting happened. Shall we skip to that, because...?
Yes, we're really talking— We're really right in the middle of that, so I think that's a good place to be.
I had started doing a thesis with [Arthur] Wightman. Let me just digress for a minute. I had a course with Wightman on mathematical methods of field theory. I thought it sounded very powerful, and I wanted to do a thesis with Wightman. So several things happened. He went on leave just as I was beginning my thesis. He just told me things to read and the mathematics overwhelmed me. I'd had quite a bit of mathematics as an undergraduate, but the mathematical treatment of several complex variables and the distribution theory I just felt I would not be able to master all of this in time to apply it to the physics problems that he had in mind.
I finally decided not to do the thesis with Wightman, but to do it in a branch of physics where I knew the mathematics better, which was statistical mechanics. I'd had a class with Wigner in statistical mechanics and I'd done very well in that class, and I knew probability theory from my doctoral work in philosophy. Wigner had actually raised some research questions in that course, so I did my thesis in statistical mechanics with him. But let me tell you one more thing about my experience with Wightman. Very interesting. It wasn't what he expected; it wasn't what I expected. While he was away he said, "Now, there's one paper I want you to read to get into the subject.
Read the paper by Einstein, Podolsky and Rosen on an argument for hidden variables, and find out what's wrong with the argument." So that was my first reading of the EPR paper, and I didn't think anything was wrong with the argument. It seemed to be a very good argument. I never saw anything wrong with it. Later on I realized it had premises. Well, I knew then that it had premises, and one thing that could be wrong is that one of the premises is false. In fact, the argument is based on relativistic locality, and that may be the trouble. That is, that the quantum correlations may not be compatible with relativistic locality.
But as an argument, starting where they started, it seemed to me a flawless argument. So I am rather grateful to Wightman for giving me something to think about which was very important in my later work on foundations of quantum mechanics. Now, several things. One is that Wigner was working at that time or was beginning to work on foundations of quantum mechanics himself, and that even though I was doing a paper on statistical mechanics which had nothing to do with the measurement problem, nothing to do with superselection rules and such, nevertheless I read what he said and I heard him talk. I heard some of his students, like Yanase, talk, and I came to know Yanase.
Janossy? I always think of him as stuck in Hungary somewhere.
No, that's Y-a-n-a-s-e, not J-a-n-o-s-s-y. Michael Mitsui Yanase, S.J. He was a Jesuit Japanese theoretical physicist. A very sweet man; very nice man. So, what happened? Some philosophical questions came up, and Wigner asked me to read a couple of his papers. My thesis work was sort of dragging, it wasn't going very well, but I read the paper, a well-known paper, called "The Improbable Success of Mathematics in the Natural Sciences."
I thought it was a beautiful paper, but I also thought that there was some philosophical background that he could look at which would make the paper even more interesting. And I pointed out some papers of Peirce to him, which talked about the abductive powers of human beings. Probably evolutionary development, we don't know the details, but somehow as the result of evolution, we have an affinity for proposing good hypotheses concerning nature.
Wigner loved that, and Wigner cites Peirce in that paper, "The Improbable Success of Mathematics in the Natural Sciences." He thanks me for letting him know about Peirce. He thereafter from time to time, would refer to Peirce. Wigner was a man of incredible intelligence, acuity, and dependence of thought. He was not well read in philosophy, but he knew a lot of philosophy. Of course, lots of philosophical ideas are in the air, so he must have heard discussions of Kant, and discussions of logical positivism.
Especially being trained in Europe.
That's right. And the time that he was in Germany, he must have heard such discussions. So he heard some things, even though he didn't read much. Then secondly, he just thought about things himself. I think he independently discovered some of Peirce's ideas. Now, Peirce spent a lifetime on them, so Peirce articulated these things better than Wigner had. So it was a great advantage to Wigner to see somebody thinking along his lines, and laying out some of the things that Wigner was just groping for. Anyway, also this episode, my being able to show something to Wigner that he didn't know about, was very good for my morale. My research work on the thesis took off at that point. It was very good that I could teach Wigner something he didn't know. Just think about it. Think about the disparity of our situations, and you will understand how much that meant to me. So Wigner was very interested in Peirce's ideas on probability and cause, but Wigner was a mentalist.
Really? I know he follows this Bohr line of phenomena don't really exist.
That's not correct. He's not a Bohrian. I will give you a paper I wrote and gave at the Centenary for Wigner in Hungary this July. Wigner calls that the orthodox interpretation of quantum mechanics. You would think that when a man uses the term "orthodox" something or other, that's what he believes in. That's not true. It's orthodoxy, but it's not his orthodoxy. He really wasn't a Bohrian. What he was a man who thought we don't understand how events occur. We don't know the limits of the validity of quantum mechanics...
In quantum mechanics, the wave function of whatever is the whole system (which may include apparatus, or it may include apparatus and environment), will just go on evolving deterministically under the Schr”dinger equation. Where does an event occur? Events do occur, undeniably there are events. Therefore, there must be some breaking of the deterministic evolution of the state. We don't know now where that break is. One possibility is that when you have microscopic systems, the Schr”dinger equation has to be revised. In fact, you'll see in my paper that I list a number of possible loci for modification for quantum mechanics. There are people who believe, like Penrose, that the proper locus for modification of the Schr”dinger equation is when the space/time metric becomes involved.
I don't remember Wigner ever mentioning that as one of the possibilities, but he certainly did at least once write down a non-deterministic equation governing the evolution of the density matrix i.e., the statistical operator. Now, one possible locus of the breakdown that he does talk about is when life comes into play. Not mind yet, but life. He has a very interesting paper arguing that the phenomenon of life with reproduction is incompatible with quantum mechanics. It's in his Collected Papers. The reason is that in order to have reproduction, some parameters of the new generation must be identical, or close to being identical, with the corresponding parameters of the preceding generation. He does some counting of the number of parameters that have to be determined, and then he says, "There are not enough constraints in the quantum dynamics to fix those parameters."
If that's so, then reproduction, similarity of the offspring to the parents, cannot be— I think it's tangential, his argument that reproduction cannot be explained physically. But he is unequivocally against a physicalistic treatment of mentality. And he says it is possible that the locus of the breakdown of validity of the Schr”dinger equation is when systems endowed with mentality are involved, like Wigner's friend. That is, the paradox of Wigner's friend still being suspended between having seen a red light or a green light, would be resolved if the Schr”dinger equation doesn't govern the mentality of the friend. So stochastically, one or the other of these possible visions is picked out.
He admits that as a possibility, and if this possibility is true, then two things. One is the integration of physics with psychology, which he believes is inevitable if science will continue. That's the great frontier, of putting them together. That when that integration occurs, it will necessarily involve some modifications of things that are now precious within physics, like the literal truth of the Schr”dinger equation. So that's one consequence of making mentality the locus of the breakdown of the Schr”dinger equation. The other consequence is there will be some similarity to Bohr's point of view. It's not that he started with Bohr's point of view, but this idea that the Schr”dinger equation is not valid when mentality enters would lead to some common ground with Bohr. And he says there still will be subtle differences between the orthodox point of view and myself, and I think he's absolutely right on the subtle differences.
He says the trouble with the orthodox point of view is that it makes the fixed points of physics to be sharp, clear observations made on experimental apparatus, like what number you read on a scaler, or whether a bell rings or does not ring. Whereas in a real integration of physics and psychology, you must take into account the whole range of psychic phenomena, including sleep, the unconscious, peripheral vision, many things that are not sharp.
By the way, that's one of the reasons Wigner was interested in Whitehead also. I introduced Wigner to Whitehead as well as to Peirce, and Whitehead insists on the importance of perceptions that are not clear and distinct, and Wigner liked that. He said the narrowness, the shallowness of the orthodox interpretation of quantum mechanics is that it puts all the emphasis on the clear and distinct parts of our conscious experience, whereas a true integration of psychology and physics would recognize the whole enormous morass of vague experience.
So it sounds as if there were really profitable, wonderful discussions between you and Wigner.
I think so. I think there were.
Did Wigner take any interest in what Bohm had done shortly before in his interpretation? Did you take much interest in it?
He knew about Bohm's work. I think he was skeptical. I don't know why he was skeptical, specifically about Bohm, before Bell's Theorem. Because of Bell's Theorem, he felt that what looked like a promising avenue to solving the measurement problem, namely that the quantum description is incomplete, and if one had a complete description, then the complete description would say whether the atom would decay or not decay, whether the photon would pass through the half silvered mirror or not pass through it.
If one had that, the measurement problem would be solved. But Bell's Theorem says that the price for a hidden variable theory that would supply that extra information is nonlocality. Since the evidence that we have now is for the theory of relativity, one doesn't want to accept a nonlocal hidden variable theory. He realized that Bohm's hidden variable theory was definitely nonlocal. By the way, I believe that I'm the one who told Wigner about Bell's Theorem. He read the paper, and he loved the paper, but he felt it was much too complicated a proof, so he did a proof of his own.
That proof was also independently discovered by Belinforte and you see it in Belinforte's book. I remember there was a time when people were referring to the Bell-Wigner Theorem, and I argued against them. I said, "That's not fair. It was Bell who discovered it. What Wigner did was look at the proof and say, "I can give an alternative proof which is simpler." Furthermore, Wigner is not the only one who gave that alternate proof because Belinforte [Frederick] independently did it.
Were other people engaging in these discussions with you and Wigner while you were doing your graduate work, or was it just the two of you?
I talked to Yanase.
Hugh Everett, was he around?
I heard Everett once, and I think I even asked him a question. I asked him whether the branching that he pictures, does that take place even when consciousness is involved, and he said yes. That is, he clarified what was a somewhat obscure point in his "Reviews of Modern Physics" paper. I started going to conferences on foundations of quantum mechanics, even before I finished my thesis. I got my degree in June 1962, but I started teaching at MIT in the Fall of 1959. Annemarie had an appointment at Mount Holyoke.
We lived at Mount Holyoke, and I remember that if I had something to do on my thesis, I would take a train down to Princeton and discuss it with Wigner. But I was commuting in both directions every month or so. I'd go to Princeton to see Wigner on my thesis, and twice a week I would go to MIT to teach a course. It seems to me very early in my time at MIT, maybe in my second year, I was given a course in foundations of quantum mechanics. That is, I was allowed to choose the course, and I wanted to teach a course in the foundations of quantum mechanics.
By that time, I know I had been to at least one conference, I believe around 1959. It was a conference where there was a lot of discussion of foundations of quantum mechanics. It seems to me this was a conference at Pittsburgh, and [Paul] Feyerabend was there and [Norwood Russell] Hanson was there. I learned quite a bit of the literature on foundations of quantum mechanics at that meeting, and used it in my class in foundations of quantum mechanics in the philosophy department at MIT.
It sounds as if that was a philosophers' conference.
Yes, it was. It was a philosophy conference with some people interested in physics. That's right. I don't remember who else was there. I have to think about it. Grnbaum might have been there, and [Herbert] Feigl probably was there. I just don't remember.
That's a conference I've never heard of.
Then Wigner also liked very much the little booklet by [Fritz] London and Edwond Bauer, I think Th‚orie de la Mesure dans la M‚canique Quantique. Anyway, I read it, and what it is is a sort of popularization of von Neumann's mathematical foundations of quantum mechanics. I thought that would be a very good text for my class, but it was in French, so I translated it and mimeographed it. It's only about sixty pages long, so it wasn't such a big job. Wigner wanted to publish it with an introduction by him. I was very enthusiastic about that. London had died, I think, by then, but Bauer was still alive and Bauer liked my translation. I was very pleased about that.
But Hermann and Company, which owned the copyright, would not allow us to reprint it because they had in mind that they would do it themselves. It is so stupid that they turned down the opportunity to publish it with an introduction by Wigner. That booklet was more explicit about the intervention of mentality in the measurement process than von Neumann is, for a very interesting reason. London's first doctorate was in philosophy. He was a student of Husserl. He was interested in physics.
His brother was a physicist, and got him interested in superconductivity. London did some beautiful theoretical work on superconductivity, and his brother said, "If you turn all that in, you'll get a doctorate in physics." And he did, and he became the greatest expert on the theory of superconductivity of his generation.
As a student of Husserl, there were some residues of phenomenology in the little booklet of London and Bauer. Without giving you the details, in the first quantum mechanics paper I wrote, the one called "The Role of the Observer in Quantum Mechanics", I have a long passage on London and Bauer. That came from reading the book to teach the course at MIT. That course really got me into the general literature.
Therefore, in a way, he got you into the measurement problem in a little bit more detail at least.
Yes. The first conference that I went to entirely dedicated to foundations of quantum mechanics was in 1963 in Cincinnati[1]. It was organized by Boris Podolsky (of Einstein, Podolsky, and Rosen), who was Professor of Physics at Xavier University. That was a really great conference, with great people attending. Wigner was there, Dirac was there, Bohm, Aharonov, Wendell Furry, and some more will come to my mind. Wigner asked them to let me make a talk, and I did give a talk. I gave what was essentially the content of "The Role of the Observer in Quantum Theory.
Dirac asked me a question and scared the hell out of me. I thought he was setting me up by a preliminary question, and then would say something devastating. He said, "What is the meaning of solipsism. I said, "Solipsism is the theory that nothing exists but the knowing subject". He said, "Oh." He was content with that. I thought something terrible would happen after that. Anyway, I think from that meeting, it became known that I was interested in foundations of quantum mechanics. I did send copies of the reprints of "The Role of the Observer" around to lots of people. I suspect that's why I got Bell's paper in the mail.
Who sent it to you?
Well, I don't really know. Bell was on a year leave from in 1964.
He was out at Stanford, wasn't he?
He was at SLAC for about half a year, and he was at the University of Wisconsin for some months, I don't know how long, and at Brandeis for a while. The paper that I got through the mail was mailed from Brandeis. Now, I knew people in the Physics Department at Brandeis, and some of them knew I was interested in foundations of quantum mechanics. Schweber knew, and maybe Grisaru, a man named Falkoff, who has since died.
He knew. I don't know whether Petersen was there anymore. Aharonov was there for a while, but I don't know that Aharonov was there at the time that Bell visited. These are people I knew. Probably, Bell asked, "Well, who is interested in this sort of thing?" I remember it wasn't a question of general interest in the physics community in the mid-1960s. The great interest in the foundations of quantum mechanics came partly due to Bell's work, partly I think because the work in quantum optics showed that there is experimental realization of some of the questions in foundations of quantum mechanics.
Wasn't there "enormous" interest in the whole question of measurement that you were talking about, and certainly that's foundation...
I think enormous is an exaggeration. Yes, people were interested in it. Some people even said that if we realize that not everything can be measured, then we have limitations on the superposition principle. If we have limitations on the superposition principle, then the measurement problem would automatically be solved. That is, it could be that somehow nature resists when one comes to the macroscopic level, of having a pointer both pointing to six and pointing to twelve. He adapted the idea of a super- selection rules. Wigner, Wightman, and Wick had a very important paper on super- selection rule. The only super-selection rule they had a demonstration from first principles for is no superposition of a state of integral spin and a state of half integral spin.
The argument is essentially if that were so, and you gave the system a 360 degree rotation, then the component that had a half integral spin would have sign change, and the one that had an integral spin would have no sign change. Therefore, the new state that would result from a 360 degree rotation would not be physically the same state. It would be 1- 2, whereas the original one was 1+ 2. Those would be physically different. Some people said, and Wigner played with the idea himself, that if the superposition principle would prevent one from ever having a superposition, 1 and 2, one corresponded to integral, and one to half integral spin— If that happened in a measurement context, you would never have a superposition of cat alive and cat dead.
It's not a question of having it and then asking how stochastically one rather than the other term is picked out. So, yes, people did think that if you understood the measurement process better, and were realistic about the measurement process, then maybe the measurement problem would go away. In general, the kind of people who did this kind of investigation of the measurement problem were really conservative. They would say, "Oh, the answer to the hard problems in foundations of quantum mechanics are just the results of our idealizing the measurement situation. If one takes into account the actual details of the measurement situation, such as the influence of the environment, and the fact that you cannot prepare the apparatus in a pure state.
The apparatus unavoidably interacts with environment, and therefore, you have to describe the apparatus initially by a density matrix. If you take these things into account, you will find you are going to get a final state which is what? It will be a density matrix for the final state of the total object plus apparatus. What will be the character of that final state? People were already talking effectively about de-coherence 40 years ago. The term decoherence came later. If you look at Kurt Gottfried's textbook, or Bohm's textbook, you will see they were already essentially doing decoherence, and arguing that the off diagonal terms in the density matrix of object plus apparatus will dwindle away. There will be many, many contributions of opposite signs, and those terms will dwindle away. So you will be left with a density matrix which is diagonal in the basis of the apparatus observable, in the pointer basis.
Now, I remember Wigner was unhappy about that, as were other people who seemed to me better philosophers than the writers of these textbooks. The reason he was unhappy was that a density matrix which is diagonal but has more than one term still represents a state of affairs in which there are probabilities that the outcome is this one, this one, this one. And that is not just a matter of ignorance, that's a matter of fact, if you are saying that the density matrix represents the objective of reality of things. Therefore, when one rather than another of these possible results is what shows up, that is a selection of one term out of many in the pointer basis, and what makes that selection? So I think that Wigner, at that stage, was already antipathetic to what later became the decoherence point of view.
In talking about what impelled all this interest in the foundations of physics, and I was just asking you about measurement as something that was interesting to people. Then Bohm, Everett, de Broglie, Vigier...
What do you mean all? There weren't that many. Yes, there was de Broglie and his school. There was Bohm, and how many students did he have? He really had one. He had Aharonov, who was a great student.
Jeffrey Bub? Is he considered?
Yes. Then later on Hiley. But these are later. In the 1960s, there weren't that many people. Going back for a moment, most of the people who actually did treatments of measurement were not trying to give a radical solution, but a conservative solution to the problem. They were saying if you remove the idealizations in describing the measurement process, you will see that the problem goes away. Read the chapter in Gottfried's textbook and you will see that essentially the problem is solved by being more careful than the usual textbook is about the quantum mechanics of the apparatus. One of the reasons why Bell did something great whereas other people did plain, ordinary things, is that he was discontented with a solution that was hand waving.
He was one of the most rigorously honest men ever, and I never met anything like it, myself. He was awesome. He didn't think that a theory, for instance, that said once you have a diagonal density matrix, the problem is solved because you could never do an experiment that would show superpositions of terms with different positions of the needle. He says that's true, and he introduced the acronym FAPP, "for all practical purposes". Yes, that's true for all practical purposes, but you haven't said how the selection is made. Furthermore, what is measurement anyway? Measurement is an anthropocentric thing. What we are doing in physics is describing the world, and measurement is just one small incidental part of the interaction of human beings, a very small part of the world, with the great physical world.
So it's not enough to have an interpretation of quantum mechanics that it simply saves all of the appearances in measurements, because that isn't the world. I have another paper in which I've given an argument. I don't think you've seen this paper because it only came out a couple of weeks ago. It is called "Reminiscences and Reflections on Bell." One of the questions that I ask, and the main part of the paper, is why did Bell discover Bell's Theorem and nobody else did? Not a hard theorem to prove. Then you could say, well, the hard part is thinking up the question, not giving the proof. But there were other people who were interested in the question, too. I traced a series of steps which Bell, because of his rigorous honesty, was discontented with partial answers.
Each discontent drove him to a further investigation. And there are about a half dozen steps which finally drove him to formulate and to prove the theorm. So my conclusion is that Bell proved Bell's Theorem, and no one else did, because of his character. I gave this talk in Vienna on the occasion of the tenth anniversary of Bell's death. Mary Bell was sitting in the first row, and I was waiting to see whether she would agree with my diagnosis. She said after, "You got it exactly right." I was very happy with that. And I believe that I did get it right. Of course he had a tremendous intellect. But what he had to such a superlative degree was the honesty, the tenacity to push through his questions. [Lunch break]
There was one thing I didn't ask you about in the Princeton period, so I'll jump back to that for just a moment, before we go back to the 1960s. Were there other courses that you were taking, or things that you were doing in the 1950s at Princeton— I noticed that you were looking at the mathematics of quantum mechanics in great detail. Is that correct, or did you already know that?
Not at Princeton. I learned mathematical methods, but the emphasis was on differential equations and integral equations, and operator methods. They were useful for applications of quantum mechanics, but I don't think I learned any of the particular tools that people who do foundations of quantum mechanics look into, like lattice theory and non-Boolean logics, and so I didn't do any of that as a student. I learned straight physics. If the thrust of my courses is to be considered, it was mainly towards quantum field theory and applications to elementary particle theory, even though that's not what I did, but the courses mainly pointed towards it. As I said, I started with Wightman, thinking to do a thesis on quantum field theory but I got discouraged because of the mathematics, and then I decided I would do something which I was interested in, namely statistical mechanics, where I was more at home with the mathematics. That worked out pretty well.
So you later on had to go through all this non-Boolean and lattice theory and so on?
Yes, but it wasn't too hard. I read von Neumann's book.
At Princeton?
While I was a student at Princeton. That's right. I didn't read it for any course. I just read it for my own edification. So I learned functional analysis from von Neumann's book. But von Neumann's book is very well presented. I think he wrote it, mercifully, for the physicist. It's not von Neumann's very highest power of exposition. It's done patiently in quite a lot of detail. So I found it a very readable book.
You mentioned one man who was teaching there at the time, who wrote, as I remembered, on von Neumann's mathematics of quantum mechanics. I think he was a mathematician who wrote on the mathematics of quantum mechanics.
[Rudolph] Haag?
It might have been.
But I didn't have a class with Haag. He was a visitor. I mentioned Yanase, but Yanase either completed his doctorate with Wigner or he came as a post-doc; I'm not sure which. So I heard him talk, and I learned some of the standard techniques of quantum mechanics from him.
So I wanted to tell the tape that Professor Shimony had looked at the Gleason paper, and then had received from a friend, with great happiness, the Kochen and Specker paper. I just want to make sure of that. Then at some point, you got Bell's proof in 1966.
I got that. So let me summarize what I knew as of 1964. I had probably taught foundations of quantum mechanics twice at MIT at that time. So I knew some literature. I knew the Einstein-Podolsky-Rosen paper, I knew Bohr's answer to Einstein- Podolsky-Rosen. I knew von Neumann's no hidden variables proof in his book. I knew Bohm's model, and I think Bohm also criticized von Neumann's additivity of expectation values. I don't remember whether I knew Jauch and Piron at that time. I'm not sure. But since I knew Kochen and Specker, I knew their criticism of von Neumann, which was essentially the same as Jauch's and Piron's. And I knew of Gleason's theorem, I knew what he asserted, though I had never studied the proof.
That probably was the extent of my knowledge as of 1964. Then what happened? The paper by Bell came through the mail. That paper said two things. We now know that there are no hidden variable theories in the standard sense, of assigning definite truth values to every projection, if the dimensionality is greater than three. But he presented a model for hidden variable theories for a spin one-half system. Locality doesn't enter it quite yet. We know that hidden variable models are possible with two dimensional systems. Then, this is so peculiar, he sketches very briefly a result which he gives in detail in his 1966 paper, which he wrote before the 1964 paper, but the paper was lost in the Reviews of Modern Physics office. Therefore, it came out chronologically out of order.
In that paper, what Bell says is that none of the no-go theorems about hidden variables pay attention to another possibility, namely, that the value which a state assigns to a proposition doesn't just depend on the state and the proposition, but on what other things are measured along with that proposition. It later came to be called the "context". I think when I wrote up something on Bell's contribution, I made the distinction between contextualistic and non-contextualistic hidden variable theories. That is bad English. Later, there was a book by Beltrametti and Cassinelli, who took my terms, and left off two syllables each. They had "contextual" and "non-contextual".
Why didn't I think of that, instead of leaving it to non- English speakers to simplify the words? Anyway, a non-contextual hidden variable theory is the kind that von Neumann was talking about. It was an assignment of a definite true value to each proposition, given the state, regardless of what else is being measured along with it. A contextual one says that if you have two different contexts in which a certain proposition is measured, the value assigned to that proposition by the state may be different. For instance, you may be interested in the total angular momentum of an atom.
The operator representing the total angular momentum, J2, commutes with Jz and it commutes with Jy, even though Jz and Jy don't commute with each other. So if you say you are going to have a hidden variable theory which assigns a definite value to J2, independently of whether I measure J2 along with Jy or Jz, that would be a non-contextual hidden variable theory. Bell says that's not physically very plausible because the procedure I would use, the orientation of the Stern- Gerlach apparatus that I would use for measuring J2 along with Jy is different from the apparatus I would use for measuring J2 along with Jz. And if what we want to do is pay attention to the influence of physics on our theories, we should enlarge our view of hidden variable theories to permit contextual hidden variable theories as well as non-contextual. So what Bell did in his 1966 paper, which wasn't available when the 1964 paper came out, was to say first of all, AI have blocked a loophole in von Neumann's original no- go proof.
On the other hand, I've opened a new loophole by pointing to the physical plausibility of contextual hidden variable theories." So, there's a see-saw. It's down for hidden variable theories with the no-go theorem, and then up for hidden variable theories when contextual hidden variable theories are acknowledged as being physically plausible. Great. Now, it turns out if you look at the end of the 1966 paper, written before the 1964 paper, with its great Bell's Theorem, you'll see that Bell already raised the question, "What if you impose some extra reasonable physical conditions? Can you still supply contextual hidden variable theories for a general quantum system?"He already in the 1966 paper says, "You may not be able to do that," for reasons which were dimly adumbrated by David Bohm. Namely, the price of doing this may be to introduce nonlocality.
Now, we can go to the 1964 paper, which says look if you wish, at a contextual hidden variable theory, do it for a pair of particles, both with spin of one-half. What's the Hilbert space? Four dimensional. You are not going to be able to have a non-contextual hidden variable theory for that system consisting of the two spin one-half systems. That's already excluded by Kochen and Specker, by Bell's earlier work, and by Gleason. What about a contextual hidden variable theory? Then Bell says, "Let's try to construct one." That is, let the values of a pair of— Well, he actually does it by taking three different observables, A, B, and C. He does A and C for Particle One, and B and C for Particle Two. He has a duplicate, the same quantity C, spin in the x direction might be used both for Particle One and for Particle Two, and two other directions of spin for A and B. Clauser, Horne, and Holt and I changed that when we redid Bell's Theorem. Anyway, he takes two particles, and he tacitly assumes that you are using contextual hidden variable theories, and let's see if it will go through.
Then the proof is that any theory in which the probability of getting a and b, say, that result, can be written as the product of the probability of getting a for Particle One, and b for Particle Two. That would mean the outcome over here is independent of the outcome over there, and also independent of what variable over there you measure, what quantity you measure. Again, this wasn't so explicit in the original paper. Later on, it was clear that what Bell was assuming was two kinds of locality. One is what came to be called "outcome independence" and [the other is] "parameter independence". That is, the probability of an outcome here is independent of what variable you measure over here, and it is also independent of what outcome it has, modulo fixing the hidden variable.
That is, once the hidden variable is fixed, that sort of screens out anything else about what is done over on the other side. So Bell's locality condition in his 1964 paper is tacitly the conjunction of two locality conditions. He later on made that explicit himself, and Jon Jarrett independently made it explicit. Both of them showed that you really have the conjunction of two different locality conditions. The net result of the theorem of the 1964 paper is to show the indispensability of a feature of Bohm's model from 1952, the feature being that nonlocality is used in order to assign definite outcomes to all quantities.
That feature of needing nonlocality was not just the idiosyncrasy of Bohm's model, which might be removed if we could only think of a more clever model. It is inevitable. No matter how clever you are, you are not going to be able to think of a model which yields the quantum mechanical predictions, and at the same time is local in the senses that I just said. It violates the conjunction of parameter independence and outcome independence.
Something that I'm getting from what you said so far is that when you saw Bell's paper, you already didn't think much of Bohm's solution, and Bell's paper seemed very convincing to you, even before you did experimental...?
I had several reactions when Bell's paper came. I thought, "Here's another kooky paper that's come out of the blue." I'd never heard of Bell. And it was badly typed, and it was on the old multigraph paper, with the blue ink that smeared. There were some arithmetical errors. I said, "What's going on here?" But I re-read it, and the more I read it, the more brilliant it seemed. And I realized, "This is no kooky paper. This is something very great." And the great thing is that the nonlocality was not just a feature of one particular model that Bohm was clever enough to think of. It was inevitable in order to recover the quantum mechanical predictions and yet to have the hidden variable structure. That's a very powerful result.
Did you immediately start thinking how you could test this, or did that come slowly?
Almost immediately. As soon as I understood what he had done, I thought, "Now, that's really interesting." Bell sort of assumed that local hidden variable theories are out now. Why are they out? Because they disagree with quantum mechanics, and quantum mechanics has been so well confirmed. I remember thinking to myself, "Has quantum mechanics been confirmed in just the sort of circumstance Bell is talking about namely, when one has two well separated particles, of the sort talked about by EPR, but correlated in a certain way quantum mechanically, correlated in a way we call entangled? Have the predictions of quantum mechanics been examined carefully in such situations?" Then I thought I knew one other relevant piece of literature.
I had become an acquaintance of Aharonov, much more than of Bohm, though I knew Bohm slightly by then, but I had read papers by both of them. One of the papers I read was a paper from about 1959, on the Wu- Shaknov experiment. The Wu-Shaknov experiment, as far as I know, was the first one to look at entangled spatially separated particles. Well, I'll tell you what the Wu-Shaknov experiment did, and then I'll tell you what Bohm and Aharonov did with it. They were interested in the parity of the ground state of positronium. Wheeler had suggested that you're not going to be able to look at that question directly, but in those days it was assumed that parity is a conserved quantity. That was before Yang and Lee.
So if you look at the daughter particles from positronium annihilation, two photons that are produced when the positron and the electron annihilate, and you measure the parity of those two photons, you can infer the parity of the positronium in the ground state because almost all the positronium that will produce pairs of photons will have been from the ground state. That was Wheeler's suggestion. So Wu and Shaknov did the experiment, and they found that the parity of the ground state of positronium was negative. That they did I don't know, in the early 1950s.
It was published in 1950.
Then, Aharonov and Bohm cleverly asked another question. Given their data, can we decide whether the pair of photons that are produced could be represented as product states of single particle wave functions, of single particle states, of photon one, photon two.
Uncorrelated, essentially.
They did not realize that the question of whether entanglement persisted when particles separate had been asked by Schr”dinger in 1935. When they raised this question at the beginning of their paper, they call it Furry's hypothesis. Now, it's not Furry's hypothesis for two reasons. One is, Furry didn't believe it. Furry said that the loss of entanglement, if entanglement is indeed lost, is in disagreement with quantum mechanics. That was the main thing that Furry pointed out.
And he didn't even make the statement that we don't know whether entanglement is lost or maintained in such situations. Whereas Schr”dinger, who I thought was more reflective about it said, "We are at the verge of our experimental knowledge. Nobody has ever done an experiment which looked for entanglement when the particles are well separated." Entanglement is well known to hold safely for the two electrons in a hydrogen molecule or a helium atom, but those are right on top of each other. So anyway, even though they got their history wrong, as almost everybody does we know by now, they still said, "We now have an opportunity to test whether entanglement persists when the particles are well separated." So they did a calculation whether any mixture of products of one particle wave functions for the two photons would agree with the data of Wu and Shaknov. That is, every term in the mixture is a product of the quantum state of one times the quantum state of two. Now, you take a mixture of those. Will any such mixture yield the experimental scattering data that Wu and Shaknov [got].
They said no. They showed that it is not so. I don't know whether Mike Horne used the wonderful phrase at that point, but he did at one point say that this using of an old experiment, digging up the data, and using that data for settling some other question, is "quantum archaeology". I love that expression. He should be remembered for that expression. What Bohm and Aharonov did was quantum archaeology on the Wu-Shaknov experiment. I knew about the Bohm-Aharonov paper, and it immediately occurred to me that the Wu-Shaknov experiment also concerns pairs of particles. They are not pairs of spin one-half particles, as in the Bell 1964 paper, but pairs of photons. And things you can say for spin one-half particles, you can translate into things you can say for photon polarization. The polarization space is also a two dimensional space.
So I thought, "You know, I think we can fill a great gap in the literature." I don't know how long I thought to myself, by using quantum archaeology. At first, I thought you could do that. But then when I looked more carefully at the Wu-Shaknov experiment, I saw that the directions in which they were measuring polarization were only parallel to each other or perpendicular to each other. I knew from Bell's 1964 paper that if you just took parallel and perpendicular directions, the predictions of quantum mechanics will not violate Bell's Inequality. You need to have other angles besides zero and ninety degrees. In fact, I think Bell uses zero, 45 degrees, and 90 degrees. I'll come to what Mike, Clauser, Holt, and I did later on.
Anyway, at first I thought that we simply could dig out the data they had, and we could already see whether quantum mechanics or hidden variable theories is supported by the actual data. Two things happened. As I mentioned to you a minute ago, one is I very soon saw that no, it's not going to suffice to use their experiment. You really have to redo the experiment and have at least three different angles between the axes along which polarization is measured for particle one and for particle two. The second thing is I talked to Aharonov. He may have been at Brandeis at that time.
He knew about Bell's paper by them, and I said, "You know, wouldn't it be worthwhile to do an experiment testing Bell's Inequality? Just because you have a discrepancy between quantum mechanics and local hidden variable theories, that doesn't mean that the local hidden variable theories are ipso facto wrong. This may be just the place where quantum mechanics is limited." He said, dismissing me, "It's already been done. That's what Bohm and I did in our 1957 [or 1959] paper."
Aharonov is a very fast thinker and a very fast talker, and I was in awe of him, and thought, "He's right. Maybe he's right. But maybe he isn't right." The more I thought of it, the less convinced I was. What I think he did, and in fact, I'm quite sure this is what he did, he was saying that no hidden variable theories of a particular kind are going to be able to recover the data of the Wu-Shaknov experiment. What is that particular kind? It is a hidden variable theory of composite systems, like a two photon system, which assigns a pure quantum state to this one and a pure state to that one. That is, indeed, a hidden variable theory. But it's not the only kind of hidden variable theory. You could have non-quantum mechanical hidden variable theories.
That is, ones in which the treatment of the individual particles is not a quantum mechanical treatment. That simply hadn't been looked at yet. So it didn't take me very long, but I don't know how long, before I thought, "Aharonov hasn't settled the question. I believe there is still something to be done." I think Howard Stein was living in the Boston area at that time. He was living in Newton. He had read Bell's paper. I think I gave it to him and said, "This is a wonderful paper." I said, "Let's try to devise an experiment to test it. And let's see if we can test it by a variant of the Wu-Shaknov experiment, the variant being using more than two angles between the polarization axes." Howard Stein is the most meticulous reader and reasoner that I know. He was interested in the problem, but he wanted to know the whole background.
In the background of the Wu-Shaknov experiment was a paper by Goudsmit and somebody else, I think. This paper looked at the probabilities of joint scattering of— No, I think just looking at probabilities of scattering of individual photons by electrons. I think that was it. So what they were doing is looking for some generalization of Compton scattering. Now I am afraid that my memory is slipping. It may have been that they actually did look at probabilities of joint scattering in different directions. The paper is very general and very difficult. Howard read, and read, and read, and he found it very difficult, but he wouldn't give up reading it. I said, "Let's accept that this paper is true, and let's go on from there. Let's just do our calculation, modulo the assumption that this is true. If we get something interesting, we can always go back and check more." I'm afraid I'm not quite as rigorous a thinker, and maybe I don't have the same sort of moral scruples that he had. His qualities are absolutely admirable, but they drove me crazy. In the end, we simply abandoned the problem.
He wouldn't give up reading the background literature, and I said, AI don't understand the background literature. I just believe that it's correct. I want to go on from there." And we were at an impasse. That's how things were. It must have been 1966 or 1967. So I put the whole thing on ice because I really wanted a companion to do this. I thought it was a big, ambitious project, and I wanted to do it with somebody. And I left it alone. I don't think I talked to anybody else until the summer of 1968. Then, I had accepted a job at Boston University, because I wanted to be in physics as well as in philosophy. My friend Charles Willis at Boston University said, "We have a man who just took his qualifying exams, and he needs a thesis advisor. He's interested in statistical mechanics. He doesn't have anybody to work with." I think Willis said, I have too many students. I can't take him on now. Would you take him on?" I remember saying that I had just begun to teach physics, after this interval of just doing philosophy.
"Can't you give me a little time to get habituated?" Willis is a shrewd man and a good friend, too. He said, "Talk to him, and see what you think." So I talked to Mike, and I said, AI just don't have a good statistical mechanics problem. That's what Willis said you were interested in, but I've been brooding about this other problem. It's been bothering me for a long time." I explained to him what the problem was, and I showed him Bell's 1964 paper, and he understood it. He was very enthusiastic. He said, "Yes, I'd like to do that." What I set him to do was to see if we could do a modification of the Wu-Shaknov experiment. He worked at it for a couple of months, and he finally said, AI don't think it's going to work." I don't know who articulated the reason. He did, or I did, or both together.
Ask him, because his memory may be better about this than mine. What turned out to be the trouble is that the photons that come out from positronium annihilation are about half MEV each. Those are very energetic, hard x-rays. How do you measure the polarization of these things? You're not going to do it with a piece of Polaroid. They will go right through. And you are not going to do it with a calcite prism either. They go right through. They are not going to choose between an ordinary ray and an extraordinary ray. You need something else. In fact, you need what Wu and Shaknov did, compton polarimeters. The compton scattering of the photons is sensitive to the polarization. If you have the photon coming in this way, into the region of the polarimeter, and the polarization is say, horizontal, you'd have a whole ensemble of such photons. Well, you'll have a whole ensemble of scattered photons, and that will not have spherical symmetry. It will be distorted in a certain way.
If you have photons coming in vertically, a whole ensemble of them, again, you will have a scattering sphere distorted in another way. If you look at the two scattering spheres, they are distinguishable. They are not very different, but they are different. So you can tell for whole ensembles of photons, if you know that each ensemble was polarized in the same way, either all vertically or all horizontally, and you look at the scattering results from the compton polarizer, you will be able to say with virtual certainty, "Yes, they were all polarized this way, or all polarized horizontally." Suppose you want to know photon by photon if it was polarized horizontally or vertically. You get practically no information. You get one hundredth of a bit of information in information theoretical terms.
That's not enough to do a check of Bell's Inequality. You need to know photon by photon the polarization. So there we were with our idea broken down, but not completely broken down. This must have been by November of 1968. Who said what to whom, I don't know, because we talked to each other about it. We realized we had to have low energy photons for which you can do easier tests of polarization. Either tests with Polaroids or tests with calcite prisms. We didn't know about the pile of plates method of measuring polarization, which Clauser and Freedman later used. But calcite prisms would have been good enough for us. Then started our enterprise of scholarship, which took the form of asking people questions. "Where can we get correlated pairs of photons?" The polarizations are correlated, and they need to be low energy photons. I asked my colleagues who did some atomic and photonic physics, and nobody could give me a good answer.
I can remember giving a talk in Case Western Reserve, and I asked people there, and nobody could give me a good answer there. Then, I knew that Costas Papaliolios at Harvard and the Smithsonian Observatory had done an interesting experiment testing the Bohm-Bub theory, and I thought, "Wow, he's a real experimental expert. He may know of this." So I wrote him, "I'd like to ask you a question about a proposed experiment..." He was in the middle of an experiment and put off a meeting.
We were actually sort of talking about how you got into the measurement problem, partly through Wigner, and partly through this very stimulating conference that Podolsky sponsored in Cincinnati, and how you first got Bell's paper, we've gone through that.
We haven't talked about that. As I said, there was some activity on the measurement problem, but it looks to me as if most of the activity was trying to make the problem go away by removing idealizations from the measurement process. There were some people who wanted more radical changes, either hidden variables— And the main exponents of those were David Bohm and Louis de Broglie. De Broglie's history was very interesting. Around 1927, at one of the Solvay meetings, de Broglie had given what he called the Guiding Wave Interpretation. Then he was dissuaded from it by others. I don't know who did the dissuading. It might have been Max Born, but I'm not sure. For years, he became a sort of advocate of the Copenhagen Interpretation.
When Bohm revived the hidden variable theory, he revived it much along the lines of de Broglie's 1927 note; de Broglie got interested in it again, and thought there were some possibilities for it. De Broglie and Bohm weren't identical. De Broglie developed something called a double solution, in which the wave equation governs the normal quantum wave, which he thought was a kind of guiding equation for the particle. The other solution was a singular solution with delta functions. Those delta functions would give you the exact positions of particles. Now, Bohm doesn't have that second solution. Instead of a second solution, he has some extra ontology.
He has particles which have positions, but the particles are really fairly classical. In fact, they obey Newton's Second Law of Motion with one modification namely, an extra force depending on what he calls the quantum potential. So even though both de Broglie and Bohm give similar roles to the wave equation itself to guide the particles, they treat the particles differently.
When did you read de Broglie's post-Bohm work? While you were at Princeton?
No. I think a lot of this I found when I was teaching the course at MIT. I had to read the literature, and I found that de Broglie had written new things after the 1952 paper of Bohm. I am trying to recall my exact reactions to Bohm's version. I didn't study de Broglie's papers very carefully. I really studied Bohm's much more carefully. My feeling was that this is too special. The Einstein-Podolsky-Rosen paper was very general. It only argued that because there are quantum correlations, and these quantum correlations cannot be explained without action at a distance unless there are hidden variables, therefore there must be some supplementation of the quantum description. But they made no commitment what that description would be.
When Bohm's paper came out, Einstein was quoted— I don't know when I learned this quotation. I think I didn't learn it until later on from Jammer's book on the philosophy of quantum mechanics. Einstein wrote a letter saying, AI don't like these hidden variables." Now, Jammer made, it seems to me, a rather heavy handed use of that statement of Einstein, saying that he doesn't like hidden variables at all. I don't believe that was Einstein's intention because any theory that says the quantum mechanical description is incomplete and has to be supplemented, is ipso facto a commitment to some kind of hidden variables, if you mean by hidden variables, that which supplements the physical description given by quantum mechanics. What I think Einstein wrote, rather carelessly, was, AI don't like Bohm's own type of hidden variables," which to him must have sounded too much like classical mechanics.
We've left classical mechanics behind, and he was not a reactionary. He may criticize quantum mechanics, but that doesn't mean he wanted to go back to pre-quantum days. So at one point or another, I did hear Einstein's objection to Bohm, and I thought that sounded right, that the EPR paper had a power which is not completely caught by Bohm's model. Bohm's model is one of many. But of course, what Bell did was amazing, again, part of his character and part of his intellect. Namely, he took one feature of Bohm's model, which probably Einstein didn't like, namely its nonlocality, and he asked whether that would be a necessary feature of any hidden variable theory that gave the same predictions statistically as quantum mechanics.
In Bohm's 1952 papers, the pair of them, he notes the nonlocality. But I think somewhere he says, "Well, we hope we'll be able to find models in which the non-locality is removed." He also knew that it was something disturbing. Well, Bell's great contribution was there is no way of removing the nonlocality, if one wants to be in agreement with the predictions of quantum mechanics, and one also satisfies EPR's sufficient condition for an element of reality, namely the ability to predict with certainty the value of a certain property without disturbing the system in which that property is inherent. If you have those two, there is no way of doing so without some action at a distance.
Is there anybody else we ought to mention? Was [Jean-Pierre] Vigier somebody of interest in this early pre-Bell period?
Yes. Vigier was a friend of Bohm's. They did a paper together. And he was also a student of de Broglie. So he was a clear link between de Broglie and Bohm, and it's possible that de Broglie's renewed interest in hidden variable theories was due to Vigier. That's a real possibility. You know, that's a piece of information you could find by a letter. Vigier is not a shy and modest man. If he is the one who suggested it to de Broglie, he will tell you.
And even if he wasn't, he might tell me the same thing.
No, I don't say that. He's a man given to enthusiasms, but I don't know that he exaggerates.
That's a very good idea. Now, this is material that I don't know, so I may be asking a stupid question, but people like Gleason, Jauch, are they important here?
They are very important. A whole other line of investigation. I know that Gleason's problem was posed to him by George Mackey.
That's the name I was thinking of. At Princeton?
No, Mackey is at Harvard. He's a retired professor of Harvard. He's a great mathematician, and a very charming and pleasant man.
That's the book on the math of quantum mechanics?
There are two books by Mackey. Well, there are more than two, but there are two that I know. One was The Mathematical Foundations of Quantum Mechanics, the same title as von Neumann's book, and the other is something like Induced Representations and Their Applications to Quantum Mechanics. Mackey was trying to do a kind of axiomatization of quantum mechanics, a minimum axiomatization from which one would recover the whole structure of quantum mechanics. Now, I think the antecedents of Mackey's work were Birkhoff and von Neumann. Birkhoff and von Neumann did a lattice theoretical formulation of physical theories, and then a specialization of them to quantum mechanics.
Mackey also uses this lattice theoretical formulation, and then he gets to a certain point where it looks to me as if he just has to postulate the thing he most wants. I want to be fair to him because he's a great man and what he did was magnificent. He postulated something that you just can't get by logical or a priori considerations. That is that the lattice of propositions of the quantum system is isomorphic to the lattice of closed linear sub-spaces of a Hilbert space. Now, that's a big postulate. That's very different from postulating, say distributivity, or a weakened form of distributivity— He has a name for it, which I should remember, but I don't right now.
It's postulating an awful lot of the full structure of quantum mechanics, when you bring in Hilbert space that way. The later work of [Constantine] Piron was in some ways more ambitious. Piron wanted to give axioms that have only a logical or experimental justification, and then derive from them the Hilbert space structure of the propositions of quantum mechanics. So that's a more ambitious program. Now, what else did Mackey do? Once he had the structure of the propositions— These are once the yes/no propositions. Does the particle pass through this filter or not pass through? Does it go into the extraordinary ray or the ordinary ray when it passes through a prism. But the projections are binary, they have yes/no answers, or one/zero answers.
So going into the ordinary ray would be plus one, and going into the extraordinary ray would be zero. Suppose the photon goes into the ordinary ray if it's polarized along the x axis, and into the extraordinary ray if it's polarized along the X axis, and into the extraordinary ray if it's polarized along the y-axis. Then the mathematical representation of a device which does this, a binary device of this sort, for any photon that comes in normal to the front surface of the prism would be a projection operator. Now, what does Mackey do? Given the structure of the projection operators, he is able to recover the structure of all observables. Those are, in his terms, "projection valued measures". Given any Borel subset of numbers on the real line, you can ask whether the value of this observable falls in that set, or falls in the complement of the set. Again, that's a yes/no. It's yes if it falls within the set, and no if it doesn't.
That's a projection operator. But if you have a complete set of projection operators for every Borel subset of the line, that complete set tells you what the observable is. Suppose the value of the position observable is seven on a certain scale. That would mean for any Borel set of real numbers that contains the number seven, the answer would be yes, and for any Borel subset that doesn't contain it, the answer would be no. So that means given the observable is fully defined once you say what projection operator corresponds to a given Borel subset of the real line. Now that wasn't new.
That wasn't Mackey's discovery, that was von Neumann's discovery. What von Neumann showed is that every self-adjoined operator, which are the operators used in ordinary, non-relativistic quantum mechanics, they are the ones that are assumed usually to represent observables. They are the ones that have real eigenvalues. So von Neumann already showed this relationship between the self-adjoined operators and the projection-valued measures. That is von Neumann's famous spectral theorm.
So Mackey simply recovers that in his book. Now, he goes further. He says, "Now that we know the relationship between observables and projection operators, what about states?" Traditionally, at least until one gets to the further reaches of quantum field theory, the states were supposed to be— Let's go back to von Neumann's version of it, the states were represented by vectors in a complete Hilbert space. The completeness property is a convergence property. That is if you have a sequence of vectors and the difference between the two of them in norm goes to zero, then the set of vectors converges to some vector in the Hilbert space. In von Neumann's formulation, those were the pure states. In addition to the pure state, there are the impure states, or mixed states, which are represented by statistical operators or density matrices. Mackey asked the question, Suppose we define a state in a more general way.
Suppose we don't assume that states are somehow mapped onto vectors of the Hilbert space, but let's suppose a state is defined in terms of the observables. I said, No, let's do it in terms of the projection operators. That is, given a state, given any projection operator, the state will assign a certain probability that that projection operator will get the answer yes when that projection operator is measured. That is not good enough. You also want, when the projection operator is the unit operator, the operator representing the proposition that is always true no matter what is done, then the state must assign a value of one to that projection. Finally, suppose you have two projections which are orthogonal to each other, which means they both can't be true at the same time. Then, you form sort of the generalized disjunction of the two.
What does that mean? Well, the easiest way to do it, is if you say one projection corresponds to one subspace of the Hilbert space, and the other to another subspace of the Hilbert space, and these are non-overlapping, then the disjunction of the two propositions will correspond to the closure of the union of those two subspaces of the Hilbert space. Anyway, you try to go as far as you can in keeping the structure of classical logic. Now, you then want to say that the state has an additivity property. That is, if the state assigns P1 to this projection, and P2 to this one, and these two are orthogonal projections, then it should assign P1 + P2 to the generalized disjunction of those two projections. Fine. That would be like ordinary classical probability theory.
One more step, suppose you had a denumerable number of these projections, and they all are mutually orthogonal. Then you want the state to assign to the generalized disjunction of all of them, a probability which is the infinite sum of the probabilities assigned to each.
I want to interrupt for a moment because I'm a little bit lost from the point of view of history, or do you want to finish something first?
I'm almost done. All of these concepts were defined by Birkhoff and von Neumann in the early 1930s. Mackey, however, posed a problem. I want to get to Gleason. Mackey posed a problem, suppose one accepts that the projection operators are isomorphic to closed linear subspaces of the Hilbert space. And suppose you accept the definition of a state that I just gave you. What states are there? Are there any states other than the pure quantum mechanical states, which are represented by vectors in the Hilbert space, and the mixed ones which are represented by density operators? He was not able to answer that question.
He suspected that the answer was yes because he suspected that if there had been any other states possible, then somebody would have turned up with a construction. But he couldn't solve the problem. He talked to Andrew Gleason about the problem, and as far as I know, Gleason had not expressed any interest in foundations of quantum mechanics before, but he just took this as a problem in pure mathematics, and he proved that Mackey's conjecture was true. That the only states in the sense of probability measures which have these desirable probability properties are those which are recognized by quantum mechanics. They are the usual pure states, and the mixtures thereof.
Does that rule out hidden variables?
Almost. I haven't gotten there quite yet. Almost. I have omitted one clause in Gleason's theorem. That clause is provided that the Hilbert space has a dimension of three or more. If it has dimension two, this theorem doesn't apply.
Is that something you were already familiar with in the early 1960s?
He did that in 1957, I think. So what does that say about hidden variables? Well, what does a hidden variable theory do, or what does the usual one do? Let's not talk about all observables. Let's talk about projection operators because they are the simplest observables. They are the ones that have yes/no answers. They have only two possible values. A hidden variable theory would assign a definite value, yes or no, to every projection operator. It would do so in a way that is consistent with those probability conditions that I just told you, so that if you have two projections which are orthogonal, they can't both be true, then remember the probability of the disjunction is equal to the sum of the two.
If more than one of those, if both of them had values of one, then the sum to the two would be two, and that would violate the condition that the probability is always a sum number between zero and one. It no longer would permit a probability interpretation. So that means a hidden variable theory must assign to every projection operator a definite value, zero or unity, and do it in such a way that if two projections are mutually orthogonal, it assigns unity only to one of them.
And if the two are mutually exclusive and exhaustive, then it must assign unity to one of them and zero to the other. Gleason's theorem, in the case of three dimensions, implies that there can be no hidden variable theory which satisfies those conditions. That is, it's a corollary of Gleason's theorem that you cannot assign, in this systematic way, compatible with probability theory, definite answers yes and no to every projection operator, and therefore to every proposition about the system. That's not the full strength of Gleason's theorem, but it's the part of Gleason's theorem that applies to the longstanding question about hidden variables, the question raised by von Neumann.
It doesn't quite say all that von Neumann's theorem is purported to say.. Von Neumann's proof, which we've already been through, it not a correct proof because it assumes a physically implausible premise. But if it were a correct proof, it would state something stronger than this corollary, because it would even cover the case of a two dimensional Hilbert space, like the Hilbert space associated with a spin one-half particle. So Gleason's theorem recovers most of what von Neumann was trying to prove, but not quite all of it.
Gleason's theorem is a very intricate theorem, and several other people were aware of the theorem and said, "There must be a way of proving the corollary to Gleason's theorem," namely the non-existence of the hidden variable theory, "even if we can't prove the theorem in its full strength," namely there are no other quantum states besides the standard ones, the pure states and statistical operators. That is what, independently, three groups set themselves to do.
One was Bell, one was Jauch and Piron, and one was Kochen and Specker. They all were interested in proving that there are no hidden variables that will assign definite values to all the projections. In the case of Bell and Kochen and Specker, they were aware that their theorems only go through if you assume that the dimensionality of the Hilbert space is three or more. In the case of Jauch and Piron, I am pretty sure they thought they had a proof that goes through even when the dimensionality is two.
Had you already read these papers before you picked up Bell?
The only one that I knew before Bell's paper came through the mail was the Kochen and Specker paper, even though that paper wasn't published until about 1967, after I received Bell's paper. But one of my friends gave me a copy of it, and said, "This is a really interesting paper because it blocks a loophole in von Neumann's proof." The Kochen and Specker proof is a correct proof. There are no holes in it. It is a rather intricate argument, more intricate than necessary, but it's a good proof. It didn't come out until 1967, I think. Bell's paper on this topic came out in 1966 in Reviews of Modern Physics, which is after the great paper with Bell's Theorem. I will go into the time inversion. Bell's proof is also correct except for some minor slips that are rather obvious to correct, and it's a simpler proof than that of Kochen and Specker.
It's a very nice piece of mathematical and logical reasoning. It's a beautiful proof. Both of them restrict their attention to the dimensions three or more. The paper by Jauch and Piron is a little problematic. At first, it looks better than either of those two. I don't remember when it came out. It came out in the 1960s also. In fact, I think it came out a little before Bell's paper, because he refers to the Jauch and Piron paper. It purports to apply, even to systems which have Hilbert spaces of dimension two, but it has an extra premise, which looks absolutely reasonable at first. Bell, in his 1966 paper, says, "It looks reasonable, but an ace has been pulled out of the sleeve."
I'm not going to try and go through that. You can look at Bell's paper if you want, and you can see where he's critical of Jauch and Piron. In fact, Bell's paper begins with criticisms of two earlier attempts to forbid hidden variable theories.. Von Neumann is one of them, and Jauch and Piron is the other. He says, "Both of them slipped something in." In fact, what they slipped in is assuming that the value assigned to a— I better not go into it. It's a little complicated. So Bell gives his own proof, and he is correct. It is definitely an improvement over Jauch and Piron.
It definitely blocks the loophole in von Neumann's proof. I like Bell's proof because it's very simple, but it's essentially no stronger than Kochen and Specker. And certain things that Kochen and Specker do are very beautiful and go a little beyond Bell. They talk about something called partial Boolean algebras. A partial Boolean algebra is a subset of all the projections, but within that subset, you can have distributivity. That is, An(BuC0=(AnBLu (AnC). So it's a little bit hard to draw it in air.
It's a slight generalization of the standard distributivity in Boolean logic. So the distributivity holds within this subset, and then you have many different subsets, and you can assign different hidden variables to each subset. But the hidden variables in one subspace will not agree with the hidden variables in another one of these partial Boolean algebras. It was a nice contribution quantum logic.
I had found out something interesting at Case Western Reserve. A friend there said, "Well, I don't know the answer to your question, but I have an experimentalist friend at Harvard named Joe Snyder, and he may know." Snyder I remembered from Princeton. He was a graduate student with me, and the only thing I remembered of him from Princeton was, I think, he was the champion tennis player in the Physics Department. I went to see him, and we dimly remembered each other from Princeton. We ran in different circles. I told him what I wanted, and he said, "You know, I think I've seen an experiment recently which worked with polarization correlation at low energies. Let's go up to the library and look."
So we went to the library, and he picked up a stack of recent Physical Review Letters. They were recent because he said it was a recent paper. In about two minutes, he found the issue with the Kocher-Commins paper. Sure enough, they did have correlation of polarization, of pairs of photons produced in a calcium cascade. What were they doing this for? They wanted to find out how long the intermediate state in the cascade lasted. They wanted to know a lifetime. Now, to know a lifetime, one way to do it is to look at the difference between arrival time of the first photon and the second one. And it is going to be variable, but you will find out something about the statistics of the arrival time. How do you know which photon goes with what?
If you can sort of use their expected polarizations, if their polarizations match, then you are pretty sure that these photons belong together as a pair. Say if you get zero angular momentum initially, going to the total angular momentum one, then down to total angular momentum zero, you know that the pair of photons has zero angular momentum, therefore you know that in any direction, their angular momenta are opposite. So they are well correlated, and you can tell whether you have two photons of a pair.
So they wanted photon polarization correlation really as a tool to help them answer another physical question, the physical question being the lifetime of the intermediate state. I later on read Kocher's thesis, and Kocher mentioned Einstein-Podolsky-Rosen. He doesn't mention Bell's theorem.
It seems to me he mentions EPR in his paper, too.
That's right. He mentions EPR. It's exactly so. And then he says, "Unfortunately, our experiment doesn't seem to have anything to do with EPR." They were within inches of being able to handle EPR, but they were so preoccupied with another question that they missed the point. As soon as Snyder turned up the Kocher-Commins paper, I said, "This is what we need. Don't look anymore." I made a xerox of it, showed it to Mike, and "Yes, this is what we need." We just proceeded to do calculations assuming that we had the Kocher-Commins apparatus at our disposal, but they, too, didn't do all the necessary angles to test Bell's inequality. We said, "We'll vary the angles until we get choices of angles in which there is a definite discrepancy between the local hidden variable predictions, and the quantum mechanical predictions."
It wasn't too hard to do. There was one thing that was hard, which Holt contributed. I'll get to that in a minute. Anyway, so we knew what we wanted. That is before my appointment with Papaliolios. He said he would talk to me after he finished his experiment. So I went to see him, and I said to him, AI made an appointment to ask you a question, but I've come to tell you the answer to it." I told him about Joe Snyder's directing our attention to the Kocher- Commins paper, and I showed him what the apparatus looked like. He said, "You know, we've got apparatus like that here at Harvard.
Frank Pipkin has apparatus like this, and he has several students using it, mainly to determine lifetime of intermediate states," the same purpose that Kocher and Commins had used their apparatus for. It shouldn't be too hard to modify their apparatus and test Bell's inequality, just as you want. He said, "I'll make an appointment with Pipkin and his students, and we will see." So what happened next...?
By the way, I did bring along these two letters[2] that you might want to look at. [break]
See, Clauser had independently come upon the Kocher-Commins paper. I knew how he did, because he told me. He gave a talk at MIT about Bell's inequality.
Did you know about that talk at the time?
No, I never heard about it.
It was in January of 1969.
I never heard of it. I probably wasn't looking at the physics calendar. Anyway, he gave the talk, and Kocher was there. That's what Clauser told me. Kocher was there, and Clauser gave the talk. I think he must also have asked for low energy correlated photons.
He said he was originally interested in doing a crossed beam experiment, and that I think it was Kleppner's group at MIT, that was working with crossed beam I may have this slightly wrong. But somebody at that seminar said Kocher has done something...?
Pritchard. David Pritchard said to Kocher, "Wouldn't your apparatus be the sort of thing Clauser needs?" And Kocher thought about it and said, "Yes, I think so." And Clauser, I'm not sure he knew about that apparatus, but he looked up the Kocher- Commins paper, and that's what he wanted. So I think he found out about it about a month before we did. Because he had time to write up a little note to put in the bulletin for the [American Physical Society] Washington meeting for 1969. Mike and I were too late for that. I remember telling Mike, "It doesn't matter.
Nobody is working on this sort of thing. This is so far out. We'll write up a full paper with all of our calculations, and that will be much better than one paragraph in a bulletin." Then the bulletin came out, and there was our work, and we felt pretty low. We really felt pretty sick. So I asked some of my colleagues, I remember asking Wolf Franzen, who said, AA note in the Bulletin isn't a publication. You go ahead and publish your own paper." Well, I didn't really like that idea. I didn't want to pretend that I hadn't seen the note in the Bulletin, I would certainly want to acknowledge it. So I called Wigner, and I called him at home in the middle of the evening. I said, "This is the best work I've ever done, what should we do about it?" Wigner said, "It's happened to me before. I and another person independently discovered the same thing."
He mentioned Bargmann by name, and he said, "We decided to join forces and write up the paper together, and it was very good. The joint paper was better than what we would have done separately." I think he said something like that. "Why don't you do the same? You call Clauser and suggest that." Well, I called Clauser, and at first, he didn't like the idea. He felt he had gotten there first with his note in the bulletin. But, we had our secret weapon after my meeting with Papaliolios, when he told me, "We have apparatus like the Kocher-Commins apparatus here," Papaliolios arranged a meeting between me, and I think Mike also. I think so, but I don't know.
Pipkin was also there, and his student Holt, and I think Nussbaum was there. Nussbaum was nearly done with his thesis, and that's why he dropped out of the picture, because he wasn't going to delay getting his degree another year or two to do another experiment. I think he had already accepted a job at Bell Labs, and he wanted to be done. He thought, "Well, maybe when I get to Bell Labs, I'll do the experiment there," but he wasn't going to delay getting his doctorate in order to do the experiment. But Holt was just beginning.
According to this letter that you sent, you say Nussbaum was now at Bell Labs and told Papaliolios that he has equipment to the experiment, and is eager to do it.
It's more than that. Pipkin didn't understand what it was all about. Pipkin was a very good physicist, but he just didn't catch on as fast as his student did. Holt caught on immediately, and Holt was explaining things to Pipkin. Finally it got through to Pipkin that this is an interesting experiment, and he would allow Holt to do it. In fact, I think Holt did nothing else for his doctoral thesis than the test of Bell's inequality. Well, in the course of it, he found out some information about the lifetime of the intermediate state of mercury, which he was using, not calcium. But Holt was the one who saw it. So anyway, when I called Clauser, I don't think I talked to Clauser until after this meeting.
It may be that I didn't even know about the Bulletin at the time. I think I didn't know about the bulletin at the time that I met with Pipkin, Holt, and Papaliolios. I think not, because that certainly would have complicated our discussion. All I knew was that we had found the Kocher-Commins paper with the help of Joe Snyder, and that we knew exactly what to do. Well, not exactly. I'll tell you one complication, we'll get to it in a little bit. We knew almost exactly what to do. Then I saw the bulletin, and by that time, it was quite clear that Holt was going to do the experiment. Therefore, when I talked to Clauser on the telephone, I was able to tell him that the experiment is underway.
Fortunately for Mike and me, Clauser very much wanted his hands on the first experiment, the first test of Bell's Inequality, because he was absolutely convinced that the experiment was going to come out for the local hidden variable theory and against quantum mechanics, and it was going to be an epoch-making experiment, and he wanted to have his hands on it. So he agreed he would cooperate with us if he could do the experiment with Holt. I think Holt was willing. I don't remember what Pipkin said about it, but Holt was willing to do it. So over the telephone, it was tentative.
We agreed to meet at the Washington meeting, and then we went over the arrangements. I certainly was happy that Clauser would work with Holt, and I was very happy that we just joined forces. I thought it would be a better paper, and certainly, it was the civilized way to handle the priority question. As Franzen said, the Bulletin did not constitute a publication, so it wasn't that he had already published before we did. But that's a borderline case because it was something in print, and it did announce what the intention of the experiment was, and that it would be done with correlated photons from a calcium cascade. So it would have been unpleasant, and suppose if we would have gone on and published independently, there might have been bad feelings.
I'm so happy that that didn't happen, that the net result was that we became friends, which is a nice story in the history of science. We really did become friends. If I can succeed in getting Clauser nominated for the Nobel Prize for his work in the design and the performance, I will have proved my friendship to him. I really want that to happen. He wants it to happen, too. I said we almost knew what to do. Now, why almost? The reason is that in calcium, the initial state, that is the state to which the atom was pumped before the cascade began was a total J equals zero state. The final state was J equals zero, so the cascade was zero, one, zero.
Two photons that come out, have total angular momentum zero. Now, total angular momentum is the sum of spin angular momentum, which is polarization for photons, and orbital angular momentum. Now, if the photons came out like this, at 180 degrees to each other, the orbital angular momentum would be zero. But if they came out that way, the only way to make sure that they came out this way, is to have minuscule little detectors here, and minuscule lenses to catch the photons coming in this direction, and the ones going 180 degrees backwards to be caught again with a minuscule little lens. What kind of counting rate are you going to have if you have infinitesimal lenses? It will take forever.
The rate of production of these pairs is rather small to begin with, and if you throw away all but one millionth of them, you're going to have to wait for a long time to get enough data, and then you are not sure that the production rate is going to be uniform. There can be all sorts of secular changes in production rate. The tube can coat up with calcium so that the rate of photons coming out from the tube changes over time. It's a mess. So what you have to do is say, All right. We'll put in fairly big lenses, and we'll catch lots of photons." Then what will be correlated is the total angular momentum, that is J of this one, and the J of this one, are anti-parallel. So Jz is plus one for this one, it's minus one for that one. Fine. Good enough. But, what is measured with your calcite prism? Not total J, not the orbital angular momentum. It's just polarization.
So what we had to do is integrate out the contribution from the orbital angular momentum, and see what kind of correlation is left in the polarization angular momentum. Is it a strong correlation or is it a weak correlation? Does this opening up of the lenses really spoil badly the correlation of the polarizations? Mike and I didn't really know how to do that calculation, but Dick Holt had done calculations of this sort.
Of course, we were in contact with him because he was doing the experiment. So he didn't do the calculation for us, but he set up the kind of equations one would need, and from then on, we just sort of followed out the rules of the game and integrated as indicated. We were able to get the probabilities of joint polarizations with various arbitrary choices of polarization axes. It turned out that you could have large lenses, lenses with a half opening of about 30 degrees, which is a big lens, it's a nice lens.
You have plenty of photons. It turns out that the depolarization effect of this large opening is negligible. That is, instead of having one correlated with zero— I think for us was one correlated with one/zero, with zero. Almost all the time, it was something like 0.995 of the time. The correlation was still very strong. So the depolarization was not such as to ruin the discrepancy between the hidden variable prediction, which is Bell's inequality, and the quantum mechanical prediction. It's because Holt set us up with that calculation that we made him a co-author in the paper. That is, there would have been a hole in our exposition without that. People don't know it.
Clauser wasn't able to do that calculation?
I don't know whether he could have or would have. He was very busy. He was trying to finish his doctoral thesis at Columbia at the same time.
Having to sail to Berkeley. Oh, he wasn't going to Berkeley yet.
No. He hadn't gotten his appointment. I think he was involved in several things. He was finishing up his thesis. He had to do some writing for his thesis. He was still doing some work on the test of Bell's inequality. He was getting a crew for his yacht. He lived on a yacht in the East River, and it was docked there, and he would drive to Columbia every day to work. But he intended to celebrate finishing his doctorate by taking his yacht down the Atlantic Coast. When he got the job at Berkeley, he decided he would take it to Galveston, Texas, and then ship it overland to California, to San Francisco. I'm not sure he's allowed to take a yacht through the Panama Canal, probably not. And he surely was not going to go through Cape Horn! So he had a complicated life at that time. Whether he could have done the calculation, or knew how to set it up or not, it never arose.
Mike and I were in contact with Holt, Holt showed us how to do it, Mike really worked out the details, and so then it was done. Later on, the next year, I did the whole problem in another way so one didn't have to do it by the approximate method that Holt set up. I was very happy. The exact computation that I did is in the paper— You have my bibliography. Anyway, it's the one in d'Espagnat's volume, Foundations of Quantum Mechanics, the Varenna volume. That's it. That must have been 1971. "Experimental test of local hidden-variable theories." That was in the Varenna volume[3].
There, I did it exactly, and it agrees exactly with the approximation. So it's one of these lucky cases where the first order in approximation was the true answer. Anyway, but by that time, we had published the four-man paper, and the first person who showed us how to handle this extracting from total angular momentum correlation the polarization correlation, which was the experimentally easy quantity to determine, was Holt. That's how he joined us in the publication. I was very pleased. I remember asking Mike, "Would you object to having a fourth man in the collaboration," and Mike being one of the best-natured men in the world said, AI have no objection."
I didn't know whether Clauser would object or not because after all Clauser felt he had made a sacrifice joining with Mike and me, but he didn't object either. It came out the right way, so I was very happy about that. The four of us remained friends. There's another little detail that I want to put in because I really dislike the idea that experimental results are theory laden, that somehow experimenters see what they want to see. Clauser is a very ebullient man, enthusiastic, and so forth. So he thought he was certainly going to find local hidden variable results, and that this was going to be revolutionary. He had bets of quite large amounts of money on the outcome, I think of the order of $500, but you'll have to get from him how much it was. His experiment with Freedman came out for quantum mechanics, unequivocally. Holt didn't make, as far as I know, any bets, but he once told me, "My experiment better come out for quantum mechanics.
If it comes out for local hidden variable theories, well, I've got the Nobel Prize, but Harvard is not going to give me a doctorate." How did his experiment come out? Well, it was an odd kind of borderline result. It did barely agree with Bell's inequality. It was sort of at the Bell limits, you know, the sum of those expectation values has to be between minus two and plus two. It was just at two. And it disagreed pretty sharply with quantum mechanics. So his experiment came out certainly clearly against quantum mechanics, even though he anticipated, and his theory said, it would be quantum mechanical. So here, you have a case in which the two main experimental players in the game each got the result that the other anticipated. So much for theory ladenness of observation. Phooey. [break]
So we got this interesting phenomenon about Clauser proving quantum mechanics, and Holt coming out somewhat for hidden variables...?
In any case, it clearly disagreed with quantum mechanics. I was not a witness to this, but I was told that a lot of the Harvard experimenters were looking over Holt's equipment and trying to figure out what went wrong. Nobody believed that the experimental results were reliable, but they didn't want to deny him a degree, but they wanted to know, "Well, what went wrong?" The conjecture I heard, and I don't know which of the great experimentalists they have there came up with it, was that the tube that the mercury vapor was in a curved tube which was optically active. I think having the glass bent into a cylinder made it optically active so that light that comes out from the tube polarized say, like this [gestures], was rotated. Even a rotation of a few degrees would cause a systematic error in the data.
Anyway, when Clauser repeated Holt's experiment with mercury, I think in 1974, he took pains to avoid rotation of the plane of polarization, and he got perfectly good quantum mechanical results. So that makes it plausible that the rotation of the plane of polarization was the source of the trouble.
Did you get a lot of response to the four man paper? Did you already know d'Espagnat at that time, or he invited you to Varenna because of the four man paper, or what?
I believe d'Espagnat invited me not because of work on Bell's Theorem, because I don't think he knew that we had done it, but for another reason. In 1962, I believe, Wigner wrote a paper in The American Journal of Physics on the measurement problem. It was a very nice paper. What Wigner attempted to show was that you couldn't solve the measurement problem by making the description of measurement more realistic in the following sense.
That is that when the apparatus is prepared, you don't know what pure quantum state it's in, and therefore you have to use a statistical operator. Now, there are some nice simplifications in Wigner's argument. I believe they are simplifications for the purpose of exposition. That is, Wigner was such a virtuoso at calculation, he could have gotten rid of all of the simplifying assumptions, such as non- degeneracy of the eigenvalues of the measured observable. He didn't really need that, but he did assume that for simplicity of calculation.
Because the theorem was proved with a few special assumptions, there were several later papers which achieved his result, that result being this: if you start with the apparatus in a mixed state, and every one of the states of the mixture is such as to give you unequivocal answers about the value of the observable of the object if the object initially is in an eigenstate of the observable with a sharp value of the eigenvalue, and then if you then go to the more complicated situation in which the initial state of the object is a superposition of states with different eigenvalues describing the apparatus by a mixture instead of by a pure state is not going to get rid of that embarrassing superposition at the end.
You will still have a superposition of needle pointing one way, and needle pointing the other way. In technical terms, it means that the density matrix describing...
...object plus apparatus will have non-vanishing off diagonal matrix elements. Well, d'Espagnat had proved a strengthening of Wigner's theorem, which eliminated one of his simplifying assumptions. John Earman and I had another improvement, another strengthening, which eliminated yet another simplifying assumption.
So I think that d'Espagnat knew of my existence from that paper on the measurement theory, but it's possible that Wigner simply recommended my name. I don't know. Then when d'Espagnat invited me, I told him, AI can talk not just about the measurement problem, but also about Bell's inequality." D'Espagnat was, I think, very pleased about that. He had already invited Bell, I believe. There are things I don't know about here.
Well, he was in contact with Bell at this point.
Very early, I think.
Because they were both at CERN or connected with CERN, weren't they?
D'Espagnat was a regular visitor to CERN, and Bell had a tenured position at CERN. Anyway, I don't think I was invited to Varenna for the Bell work. I think I was invited for the measurement problem. As a matter of fact, I gave several talks at Varenna, and one of them was on a paper on measurement — not the measurement problem that Wigner had just written on, but something else, that Howard Stein and I did together. It was limitations on measurement when one has an additive conserved quantity. I presented three papers. One was on the realistic interpretation of quantum mechanics, like my paper on the role of the observer; one on additive conserved quantities in quantum measurement; and the third on the Bell experiment. I also had a poem called Babar et les variables cach‚es.
And you gave that there, too?
No. I didn't give it there. I had written it that summer. We were on vacation in the south of France, and I wrote it in order to show my children what I did, but I used the Babar characters. I brought it to d'Espagnat to correct the language. He said, "Oh, we should put this in the volume." It was one of my bits of foolishness. You were asking something else, though?
I was asking about reactions to the four-man paper.
All right. Now, there were people who were very excited about this. The one man who was really excited was Franco Selleri. He was an advocate of hidden variable theories, and he was absolutely sure that when the experiment is done, if it is done properly, it would come out for local hidden variables.
Is this the first time you had contact with him?
He was just one of the people at the meeting.
So this was an excitement at the meeting?
This was in the meeting in the summer of 1970. Our paper came out in the fall of 1969. So it wasn't long before the meeting. Selleri said, "Call Clauser and call Holt, and find out whether they've gotten results yet." I said, "Look, you are throwing away your money. These experiments take a long time, but I don't mind talking to my friends if you will pay for the phone call."
Is Selleri a theorist?
A theorist at Bari. So I called, and of course, they didn't have results yet. Clauser and Freedman didn't have results until 1972, and Holt until 1973, I think. Now, to whom did we send our preprint, Clauser-Horne-Shimony-Holt, the four-man paper? We sent one to Wigner. I guess we sent one to Bell. I believe we forgot first, and then later remembered and sent one to him. We sent one to Louis de Broglie. This is a very interesting case. We sent one to two men who had written a sort of semi-popular book on de Broglie's point of view, Andrade e Silva and Lochak. There weren't very many people who were interested in the problem, but we knew they were because of their little book.
You sent one to Bohm, I hope?
I think we did. Yes, we did send one to Bohm. That may be all, but there may be one or two more. Now, we didn't know everybody who was interested in it, but these were the people we knew. I said, "If we are going to write to de Broglie, we must have a cover letter, and it must be in French." So I volunteered to write the letter, and I had one of my friends on the Wellesley faculty go over and correct the grammar, and it was meticulous. It was perfect. I got back a letter from de Broglie, in such beautiful old script, it looked like the Declaration of Independence. I answered that.
Again, I got corrections, so the letter was perfect. Then, I got a second letter from him, and I answered it. So in all, I sent three letters to him. Andrade e Silva was at that meeting, and he told me, "Oh, Monsieur de Broglie was so impressed with your letters." He said, "Here is one American who knows how to write a correct French letter. Do you want to meet Monsieur de Broglie?" I said, "Absolutely not!" [laughs] It would be a disaster. And I never did.
But your French is good enough.
Not good enough for de Broglie! But I was invited to give a lot of talks, as was Clauser and Horne. I don't think Holt was invited to give many, but...
But he was still a graduate student. It may have something to with it. Did Horne already have his degree by then?
He got it the next year. His work was done. It just took him a while.
Was the Varenna Conference important for your own thought processes? Did you learn anything there, was Selleri an important person to know?
I met Bell. He was an awesome man. I think I had as great a respect for him as for anybody I've ever met. I then continued to study his work very carefully. So that was certainly important, because I followed the details of his work. That was very important.
Your meeting with d'Espagnat, was that your first friendship with him? Is that the first time you met d'Espagnat?
Yes.
And that was also an important relation for you, wasn't it?
Yes. We are friends. And he invited me to France— This was the Summer of 1970. He invited me to spend a year at Orsay, at the Universit‚ de Paris. I spent a year there. I mainly worked on the measurement problem when I was there. I met both Jauch and Piron at the Varenna meeting, and I got interested in their approach to quantum mechanics via quantum logic. I did some work, probably as a result of studying their papers and Piron's book. So that was influential. Then it's hard to say. There were questions that came up, that I must have continued to think about. That is, particularly, how do you reconcile quantum mechanics with relativity theory. I was interested in the question of whether one can send messages via quantum nonlocality.
Furthermore, you may say, "All right, local hidden variable theories are out. They won't work. Then there are nonlocal ones." You can say, quantum mechanics itself is a nonlocal hidden variable theory, especially if you say, "We're not asking for a hidden variable theory which assigns definite values to every observable. All it need do is assign probabilities, and then the experimental probabilities are really an integral of these probabilities using the distribution over the hidden variables." So in that sense, quantum mechanics itself is a hidden variable theory, but definitely a nonlocal hidden variable theory. You're weakening or broadening the idea of a hidden variable theory.
In fact, I'll say a little bit more than that. Bell was such a gracious man. Here's what happened. When we first met at Varenna, he said, "Have you looked into the question of whether you can derive the hidden variable inequality." He didn't call it Bell's inequality, but that's what we do call it. "Derive it without the assumption that the hidden variables assign sharp values to each one, but instead assign only probability distributions." I said, "Well, I've thought about that problem, and I would like to solve it, and I looked at it a bit, but I haven't solved it." He said, "Well, if you had solved it, I would have let you present it. I want you to have the credit for it, but since you didn't, I will present what I have." If you'll look in the proceedings of the Varenna volume, you'll see that his presentation of Bell's theorem is much more general than his presentation in the 1964 paper. There are two changes.
One is he doesn't just use three directions as in the 1964 paper. He uses two directions, or two values of the parameter for Particle One, and two values of the parameter for Particle Two. That means in all four combinations, A with B, A with B1, B with A1, A1 with B1. So it's much neater when you do it that way, and it's much easier to make a comparison with experiment when you do it that way. That was an advantage of the variant of the inequality that Clauser, Horne, Holt and I derived, by working with four directions instead of three. But the main improvement was that the proof that Bell gave in 1970 at the meeting was what we've come to call a stochastic hidden variable theory.
The hidden variables only assign probability distributions to the various observables. That makes it more powerful. That means even if you banned the determinism, you obtain a conflict between locality and quantum mechanics. I remember there was a time when the literature got filled with rather slovenly remarks. There was a time in which I saw often that one must choose between determinism and locality.
It's not true at all. Even if you give up determinism and accept that even the hidden variables with extra information will not fix the values of the observables, you still cannot get the quantum mechanical results without nonlocality. I believe that kind of remark has finally dropped out of the literature. But it was annoying for quite a while. People just didn't get it right. They should have read Bell's 1970 paper, or the 1971 Varenna Volume, and he had it very clearly there.
I spoke to Clauser. He thought that even in the 1970s, there wasn't a whole lot of interest in this subject, but you are revealing a whole set of people who seem to have been quite interested. Of course, he tried, after his post-doc, to get a job without a great deal of success...
I think he was treated very shabbily. He's a brilliant man, a very good experimenter, and really a good theoretician also. I don't know whether the work he did at Lawrence Livermore Lab, the work on controlled fusion, was done because he didn't get a professorship...
That's my indication.
But how he could have been turned down at so many places, I don't know.
Now, his feeling was that physicists have still not woken up to the interest of the stuff that he was doing.
I think that's right.
But you did find that physicists were responding to you?
Certainly there was a lot of interest, but that doesn't mean there was enough to get majority votes in physics departments to bring somebody whose main credentials were an experiment in concerning hidden variables. I know for sure that— I have a friend who is at one of the California universities who was a great champion of Clauser, and I said, AI know that Clauser applied to your university," I won't tell you the place, "And why didn't he get the job there?" He said, "Oh, my colleagues thought that the whole field was controversial." I said, "That may be. The questions are open questions, but the experiments were solid experiments." Just because the question was controversial before the experiment didn't mean it was controversial after the experiment. Anyway, he was treated badly.
Now, there's these people at I guess Texas, and James McGuire and Edward Fry . Were you in touch with them? Fry did another version of...
Wait a minute. No. I didn't meet Fry until 1976. Where was that? I think that was— I don't remember. Fry was there, and Clauser was there, and what Fry did that was beautiful— Well, he did several beautiful things. First of all, he got his data in I think it was 80 minutes. Clauser and Freedman had taken, I don't know, months. You know that from your interview of Clauser. And Holt took many weeks. I think Fry was the first to use a laser to excite his samples. His source was mercury.
He used natural mercury, not isotopic mercury as Holt did. So he got lots and lots of data, but of course most of it he threw away because most of it was the wrong spectral lines, because there were lines from isotopes he wasn't interested in, isotopes with nuclear spin. So he did very careful calculations of what would be the spectral lines of the isotopes, so that he could use filters to get rid of those. I was very impressed that he could use natural mercury to get data quickly without having to do isotopic separation and so on, and good calculations to know which lines to throw out.
It was a beautiful experiment, and I think Clauser said, "Your talk is a hard talk to follow." Clauser repeated Holt's experiment, and maybe Fry did also, because he used mercury, had in mind to see what would happen if one redid Holt's experiment. Would one get his anomalous result again? I can't remember which meeting this was. I went to so many. I'm sorry. [It was in Erice, Sicily, in April 1976]
I know. I hope nobody would ask me what happened in 1976 because I certainly wouldn't remember. [break]
That is, what did I learn at the Varenna meeting? Well, I certainly began to think very seriously about the compatibility of quantum mechanics with special relativity. That is I thought local hidden variable theories are over, quantum mechanics has won. But quantum mechanics really predicts these correlations at a distance because of nonlocality, because the outcome over here depends on not— This is subtle. It is not determined by what quantity you measure here, but it is determined by what quantity plus the outcome of the measurement over here.
Now, what would it mean to say, the outcome on the left-hand side is determined by the experiment by the choice of the parameter observed on the right- hand side, regardless of the outcome. What you could do is say, "Let's do the run a number of times. There will be outcomes of plus and minus over on the right-hand side," but always— Say you look at polarizations along the axis n. And you average over all of the cases of the outcome of measurement of polarization along n1 on the left-hand side.
But always keeping n the same over here. Now, suppose that the average on the left-hand side is sensitive to the choice of n. If you chose n1, you would get one average over on the left- hand side. If you chose N2, you would get another average. If you did this, then you could send a message faster than light, because you could quickly switch in an ensemble from measuring along N1 to measuring along N2, and then observer over here, when you make a large numbers of observations would be able to say, AA-ha! It was n1 that was used," or, AA-ha! It was n2 that was used." That means that if parameter independence broke down, that is, if you had a dependence of the result over here upon what parameter was chosen on the right-hand side, then you would have superluminal signaling. That would be possible.
I'm assuming this took you some time to work through, or it wasn't immediately obvious.
Let's see what happened. I did only one small piece of work on this myself, but I publicized the work of other people. The only piece of work that I did myself was to show that— It was in one of the pamphlets of the Institut de la M‚thode. There were lots of these. It was an informal newsletter published in Switzerland[4]. Somebody said that quantum mechanics allows one to send a message from the right-hand side to the left-hand side, making use of the fact that the total angular momentum is conserved.
That wouldn't be Lochak, would it? Because there is a paper here in 1976, "Reply to Dr. Lochak." You've got one up here with Horne in the Institut, and one down over hereY Here's another, "Reply to Bell."
Here it is. "An Analysis of the Proposal of Garuccio and Selleri for Super- Luminal Signaling[5]." That was it. And what I did was to show that— Now I remember how it goes. I remembered that there was a flaw in their signaling method. That is they assumed that you knew a total l2 or a total J2, and my argument was that you would not be able to determine the total J2 unless you had data from both sides. It wouldn't— There is no experiment that you could do looking just at the left-hand side to enable you to infer what was done on the right-hand side.
Therefore, the super luminal signaling doesn't take place. Now, that was in a very special case. I only did it for angular momentum. Later on, three different people independently showed what I think is a very important and not difficult result. That is that you do this analysis of Bell locality into a conjunction of two different kinds of locality. One is parameter independence, and one is outcome independence. Then you look and you see that if parameter independence were violated, which is just what I was talking about now, that is the average over here depended on which choice of the parameter you made on the right-hand side, even though you average over all possible results, then you would be able to signal superluminally.
If the result over here depended on the outcome over here, that is it depended on whether a photon passed through a piece of Polaroid, or didn't pass through, there would be no way of using that dependence for signaling. Why? Because you don't control. You can't send an SOS or any Morse Code by making use of processes which are stochastic to begin with. If it's a matter of chance whether the photon passes through the Polaroid or not, you can't influence that in order to send a message over here.
Therefore, even though quantum mechanics is nonlocal in the sense that the outcome over here depends on an outcome over here, that is the violation of locality that is innocuous because it doesn't enable you to communicate superluminally. I said there were three people independently that proved that quantum mechanics does not permit the breakdown of parameter independence.
What paper did I do this? I know. It was in the Japanese paper, "Controllable and Uncontrollable Non-Locality." I sum up the whole situation, and I refer to work by Don Page, By Ghiradi, Rimini and Weber, and [Philippe] Eberhard. I said they showed that if quantum mechanics is correct, you will never have a breakdown of parameter independence, therefore you are not set up to violate no superluminal signaling.
So really it's not until the early 1980s?
That's right. I think Eberhard did the first around 1979, and then Page and then— You can look at that Japanese paper, which is reprinted in Volume Two of my Search for a Naturalistic World View, and you will see the references. I thought this was a very important result. It was not a difficult one, (but importance of results don't depend on whether the theorem is difficult or easy), because it allowed for a peaceful coexistence between quantum mechanics and special relativity. That is that somehow, you could have both of them be true, and the kind of non-locality that quantum mechanics introduces somehow can live with special relativity.
Is this the first time that one can whole heartedly accept nonlocality?
Well, this was my philosophical attempt to reconcile the two theories. The term— Do you remember where peaceful coexistence came in? Do you remember politically...?
It was the Soviet Union and...
It was Khrushchev's term. Khrushchev made an amicable speech at the United Nations, I think, saying that yes, there is tension between the capitalist system and the soviet system, but they can live together they need not have a war. That's what he called peaceful coexistence. So I just adapted that phrase for the relation between quantum mechanics and relativity theory.
[1]See "The Foundations of Quantum Mechanics, A conference report by F.G. Werner", Physics Today 17 No.1 (January 1964), 53–60.
[2]Michael A. Horne to John F. Clauser, April 18, 1969 and Abner Shimony to John F. Clauser, April 20, 1969, from the papers of J.F. Clauser, to be deposited at the Bancroft Library, U.C. Berkeley.
[3]B. d'Espagnat, editor, Proceedings of the International School of Physics "Enrico Fermi", Course IL, Foundations fo Quantum Mechanics (Academic Press, 1971).
[4]Epistemological Letters of the Institut de la Méthode.
[5]This conversation refers to the "Bibliography of Abner Shimony", pp. 247–253, of Robert S. Cohen, Michael Horne and John Stochel, eds., Experimental Metaphysics & Quantum Mechanical Studies for Abner Shimony Vol.1 (Kluwer Academic Publishers, 1997). The paper Shimony mentions is item 56.