The Many Worlds of Quantum Mechanics

Greetings from Sihanoukville, Cambodia, or at least the waters immediately off. I’m here as part of Bright Horizons 19, a two-week cruise on the Holland American ship Vollendam, in collaboration with Scientific American. We started in Hong Kong and have been working our way south, stopping a few times in Vietnam, and after this we’ll briefly visit Thailand before finishing in Singapore. A fascinating, once-in-a-lifetime experience, even if two weeks is an amount of time I can’t honestly afford to be taking off. Been getting a touch of work done here and there, but not as much as I would have liked, in between dashes ashore to sample the local cuisine. Although the local cuisine has been pretty spectacular, I have to admit.

My job here is to give a few talks about physics and cosmology to the folks who signed up for the package — a public audience, but the kind of people whose idea of a good time while sailing the South China Sea is hearing talks about molecular biology or world history. Mostly my talks are variations of themes I’ve spoken on frequently before — the Higgs boson, the arrow of time, dark matter and dark energy. But to spice things up I decided to throw in something new, so I wrote up a talk on The Many Worlds of Quantum Mechanics.

And here it is — the slides, at least. The content is roughly based on my explanation in From Eternity to Here, with a few improvements thrown in.

Two basic goals here. One is to introduce QM to people who don’t know much more about it than a vague notion of “uncertainty” or “fluctuations.” And in particular, to focus on the conceptual foundations, rather than any of the other perfectly legitimate angles one could take: the historical development, the calculational basics, the experimental evidence, the role in modern technology, and so on. Hey, it’s my talk, I might as well concentrate on the parts I’m most fascinated by. So there’s a discussion of entanglement and decoherence that is a bit more specific and detailed than one would often get in a talk of this type, even if it is enlivened by silly pictures of cats and dogs.

The second goal was to give a subtle sales pitch for the Many-Worlds interpretation. Really more damage control than full-on hard sell; the very idea of many worlds is so crazy-sounding and counterintuitive that my job is more to let people know that it’s actually quite a natural implication of the formalism, rather than a bit of ad hoc nonsense tacked on by theorists who have become unmoored from reality. I’m happy to bring up the outstanding issues with the approach, but I do want people to know it should be taken seriously.

Comments welcome, especially since I’ve never tried this approach in a talk before. Of course by only seeing the slides you miss all the witty asides, but the basic substance should come through.

This entry was posted in Science. Bookmark the permalink.

67 Responses to The Many Worlds of Quantum Mechanics

  1. kashyap vasavada says:

    Sean:I am somewhat sympathetic to the idea of multiverse in the cosmological sense but not at all in the quantum mechanics (Everett) sense. For QM it seems to me that this is total copout because we do not understand QM. My reasoning is this. This idea in QM is so vague and arbitrary that perhaps it cannot even be called science. Suppose a professor asks his graduate student to do a QM experiment next morning. If the student comes to work early then the universe splits. If the student feels lazy that morning and does not do the experiment then the universe does not split!!! After the experiment is done with say a billion electrons or photons the final result is completely predictable. Any other graduate student will get the same answer. Thus all the split universes have to conspire to give the same final answer! Am I missing something in this argument? If the argument is that the universe is split already in heavens before any experiment is done and you are just choosing the branch (by free will?!!), that argument would be too religious and unscientific. I would be surprised if an atheist like you would believe it.

  2. Harry says:

    I liked the slides; a good way of presenting it to general public listeners. This is such a hard subject to really grasp. One can see why people have a hard time believing that quantum theory provides the true nature of the world we live in.

    Being honest with myself, I did not truly grasp this until grad school. Sure, I had the “quantum physics for non-physics majors” in undergrad, but the teacher was uninspiring and I had partying to do. Eventually a good course in applied PDEs and a Mat Sci course heavily based in quantum mechanics cemented the knowledge.

    I feel very strongly now that this material needs to be emphasized in high school physics. Obviously the students aren’t going to be doing time-independent solutions to Schrodinger’s equation, but they could certainly stand to see material like your slide show presents. In my science education as a child, I feel like the implications of quantum theory were the major omission. This is how the universe works, after all!

  3. miller says:

    Incidentally, I just made a presentation on QM to my friends, and several parts are identical. I couldn’t use cats and dogs though, because part of the point was to emphasize how decoherence occurs on very small scales; cats are dogs are well beyond that point.

    I find that Many Worlds is an important thing to introduce, because most people aren’t aware of any legitimate scientific discussion on the subject. As a little activity, I mixed a bunch of quotes from physicists and pseudoscientists, including one quote by Sean on MWI. My friends thought that one was pseudoscience.

  4. vmarko says:

    Hi Sean,

    Nice slides! :-) I do have three comments, though. :-)

    Slide 34, “this can’t be tested” objection. Define “this”, please? In your response to this question you seem to assume that “this” stands for “MW interpretation of QM”, but then go on about the testability of QM itself. My point is that there indeed is *no way* to test the MW interpretation against the Copenhagen interpretation — they are both formally equivalent, since they both interpret the same theory. The fact that all sort of experiments test and support QM itself has nothing to do with the testability of MW interpretation against Copenhagen. Your audience might confuse the two things being tested. Incidentally, if you define “this” as “MW interpretation”, the objection is not silly at all.

    Slide 35, bottom row: “The theory is completely deterministic.”. This can also be misleading. The theory is *not* deterministic, because the objects being determined by the theory are not measurement results, but *probabilities* of measurement results. QM deterministically predicts probabilities only. That makes it explicitly non-deterministic, as opposed to a deterministic “clockwork” classical theory. Your audience might get confused by that too. I’ve seen it happen, too many times. :-(

    Slide 35, top half, “How do classical worlds emerge?”. This question is not only reasonable for MW interpretation, but for the whole QM, i.e. all interpretations. Moreover, it is a very serious question, down to the point of showing a fundamental gap in the structure of QM. Your comment “Roughly, the answer is because the interactions are local in space” may sound reasonable if you don’t consider the quantization of space itself, i.e. of gravity.

    But as soon as you enter the domain of quantum gravity, you must allow for superpositions of the “wavefunction of space”, and locality goes away in a puff of smoke. This leaves the question of emergence of classical world to blow up straight into our faces, uncovering a glaring hole in our understanding of QM.

    This glaring hole was ignored for quite a while because quantum gravity is a tough nut to crack, and not many people have been asking those questions over the years. But sooner or later, we need to face the fact that quantization of gravity undermines some answers that we thought were reasonably good, like locality and decoherence. In the context of QG, the measurement problem and the Schroedinger’s cat problem come back to haunt us, showing that our understanding and formulation of QM is conceptually very incomplete.

    So your answer that locality specifies a preferred basis may sound plausible, and we may have no better explanation for the time being, but I want to stress that such an answer is certainly not the end of that story. It’s just a patch — one that is likely to burst wide open once we confront it to quantum gravity.

    Best, :-)

  5. John Call says:

    kashyap vasavada,
    I am not sure I entirely understand your argument, this is what I am getting from it, Student wakes up, does experiment. Universe splits because he awoke (other possibilities = other worlds). Or Student sleeps in and doesn’t do the experiment. Now this is where you lost me, “then the universe does not split!!!” I think this is the problem in your argument (as I understand it). The universe “splits” in either case. What is happening is no matter what the student does all options happen in a “world”, which option splits off is dependant on ones perspective.

    The experiments outcome is a different system than the students awaking or not. If said student preforms experiment or if it is another doesn’t matter. One worlds outcome will be X while another worlds is Y.

    Every time more than one posibility presents itself, there exists a world where each posibility occurs. Hopefully you get what I’m saying, please let me know if I misunderstand you.

  6. Mitchell Porter says:

    So far as I can see, Many Worlds has two things in its favor:

    1) It sounds simple. People are already familiar with one world, the world they experience and know, so it sounds like they just have to imagine that there are more of the same.

    2) QM seems to contain coexistence of possibilities (superposition) and interaction of possibilities (interference), and Many Worlds again provides a simple interpretation of this: all the possibilities are actualities.

    But when you get into the theoretical details, Many Worlds is a dog’s breakfast of nonsense, evaded questions, and inconsistent positions.

    Consider Sean’s slide 34 – his response to the first “silly objection”. “The number of possible states remains fixed.” What fact about quantum theory is he referring to here? That the Hilbert space remains fixed? Or is he perhaps thinking of a particular basis? Maybe he’s referring to unitarity, as the dynamical conservation of the norm (size of the state)?

    And isn’t this remark in tension with the common understanding of Many Worlds, that the worlds “split”, i.e. increase in number? Is he alleging that there is a conservation of the number of worlds, so that for each split, there is a join happening elsewhere? Does that mean that he actually has a definite prescription such that, given a particular wavefunction, he can tell us exactly what are the worlds in it, and how many there are? If so, could he tell us what his principle for decomposing the wavefunction into worlds is?

    Let’s consider the next slide. We get a mention of the preferred basis problem. In the previous slides, quantum states were illustrated with examples like “cat awake + cat asleep”. Again, the layperson may think that this is all straightforward; it might be strange to think of two worlds coexisting, but one world with a cat awake, or one world with a cat asleep, at least that is a familiar and comprehensible thing… But the meaning of the preferred basis problem is that, confronted with the quantum state “cat awake + cat asleep”, you ask, does it consist of a “cat awake” world and a “cat asleep” world; or does it consist of, e.g., one world that is “99% cat awake and 1% cat asleep”, and another world that is “1% cat awake and 99% cat asleep”?

    If I were to try to chronicle everything that is wrong with the intellectual culture of Many Worlds enthusiasts, I would surely exceed the bounds of this comment as well as my own patience. But to summarize my impressions over the years, discussion of Many Worlds exists in a sort of haze that is inherited from the Copenhagen interpretation. (And by the way, the original Copenhagen interpretation is not “wavefunctions are real and they are collapsed by observations”, it is just “wavefunctions are mysteriously effective calculating devices, and we form no opinion about what happens between observations”.)

    To accept Copenhagenist quantum mechanics as the final form of physical theory, one had to rationalize a disinterest in the reality of what happens between observations, and in the causes of individual events. This required a certain slackness or carelessness of thought, a deliberate weakening of the idea of an objective reality. Many Worlds is an attempt to reinstate ontological realism in physics, but the heritage of careless thinking about such matters is still there, and that is why Many Worlds advocates are so characteristically oblivious to questions like, exactly what are the worlds, and how many of them are there, and when do they split, and why.

  7. kashyap vasavada says:

    @John Call . I am not sure if I understand your argument. Are you saying that the universe was already split into possibilities even before you measure the system and you are merely choosing the branch at each instant? That is too much like predestiny or at least some free will. But as I said after you do experiment with billions of electrons, the result is determined for everyone. That is not satisfactory. Do you really take it seriously that in one universe you have an accident and in another one you escaped from having an accident? What is really the meaning of you having multiple lives? Application of QM to our life may look like a joke, but some people believe that QM is always valid no matter what the size of the system is.

  8. kashyap vasavada says:

    @vmarko: What is the basis of your belief that the problem of interpretation of QM has to do with QG? QM works superbly in situations where effect of gravity is insignificant. So we have to find believable interpretation even in those situations.

  9. I have only one question regarding the quote from Deutsch about quantum mechanics: “…the very suggestion that it may be literally true as a description of nature is still greeted with cynicism, incomprehension, and even anger”.

    Is this true? In my experience I never encountered this and I highly doubt it. However, I do see this applied to the many-worlds interpretation and maybe Mr. Deutsch equates quantum mechanics with MWI, (which is not true given the other interpretations out there).

  10. vmarko says:

    @ kashyap vasavada,

    “What is the basis of your belief that the problem of interpretation of QM has to do with QG?”

    Shortly put, the modern QM resolution of the Schrodinger’s cat problem is via decoherence — large quantum system necessarily interacting with the environment, thereby choosing a preferred basis on orbital Hilbert space (“spatial locality” of the interaction Hamiltonian).

    Once you consider gravity/spacetime as a quantum system, spatial locality is lost, and decoherence argument cannot be implemented. Consequently, the Schrodinger’s cat problem comes back from the dead, with no solution in sight.

    The argument is conceptual, and it doesn’t depend on the strength of the gravitational interaction, or the very small scales where we are supposed to observe QG effects. It is just a consequence of taking QM seriously, and applying the superposition principle to gravity, i.e. spacetime itself.

    Btw, this argument has nothing to do with “interpretations” of QM, but with the actual formalism of QM itself. IOW, this problem must exist equally in all interpretations of QM, given that the superposition principle does.

    HTH, :-)

  11. Ben P. Stein says:

    Thanks so much. This gave me a better and deeper understanding of many-worlds. The examples in the presentation follow the best tradition of David Mermin’s entanglement explanation in his classic 1985 article, “Is the moon there when nobody looks? Reality and the quantum theory.”

  12. kashyap vasavada says:

    @vmarko: Thanks. But you are opening Pandora’s box! You are saying that the whole formalism of QM and QFT would have to be changed to answer the interpretation question! Of course the well known feeling is that the formalism has worked so well that people may not want to throw it out. BTW do you believe in ST? Are there even hints at solving this issue in it?

  13. vmarko says:

    @ kashyap vasavada:

    The measurement problem and the Schrodinger’s cat problem are not issues of interpretation, but of formalism. In order to be resolved, it is quite possible that some of QM (and consequently QFT as well) needs rethinking. The formalism of QM works so well in everyday and atomic-scale physics. That’s how QM got formulated in the first place. But if you dare apply that same formalism to, say, cosmology — you immediately run into trouble. If QM is to be a fundamental theory of nature, there are various things that need to be ironed out. The contact with gravity only emphasizes those things and puts them into focus. Nongravitational issues of QM have already been ironed out, so it should not come as a surprise that QG is the only remaining relevant problem-highlighter for QM.

    As for string theory, so far it does not have a proper handle on nonperturbative gravity. There are no equations of M-theory to be found anywhere, nobody really knows how the theory actually looks like. All there is are a handful of low-energy effective-theory limits, which do not encapsulate the issue completely. In those effective theories, gravity is treated as a spin-two field in flat spacetime — while you can formally discuss superpositions of gravitons, you still work in flat Minkowski spacetime, and you have no appropriate tools to even formulate the superpositions of geometries, let alone discuss what happens with them wrt. to the above QM problems. So no, string theory has nothing to say about those problems, so far. Maybe some day, when (if) someone actually writes down the fundamental equations of M-theory or something…

    Ehm, we are getting off-topic here… 😉

    Best, :-)

  14. Matt says:


    I really appreciate this presentation. I agree that going beyond the “gee-whiz” popular depictions of quantum mechanics as having to do with “small” things, “fluctuations,” and “uncertainty” is a great thing to do. And I really like that you’re sharing some of the elegant formalism and deeply subtle fundamental ideas with a lay audience. I wish more of my colleagues would do this.

    I like most of your slides, but I have some comments about a few of them.

    On slide 22, you write “Classically, we describe the systems separately. In quantum mechanics, we desribe them both at once.” That’s only if we insist on working in terms of state vectors. If we generalize our formalism by using density matrices, which are the natural quantum generalization of classical probability distributions for the individual subsystems, then we can indeed describe the cat and dog subsystems separately. In a grad-level textbook on QM, we do this all the time.

    On slide 34, in answer to the “silly” objection “That’s too many universes!”, you write that “The number of possible quantum states remains fixed.” What does that mean? Do you mean that the overall system still has just one state vector? That’s true, but its expansion in the preferred basis will have nonzero components along increasingly many basis states. Indeed, that’s the whole point of thinking about the many-worlds interpretation as consisting of many worlds!

    Also on slide 34, in answer to the objection that “This can’t be tested!”, presumably referring by “this” to the many-worlds interpretation, you write “Many-worlds is just QM without a collapse postulate or hidden variables. It’s tested every time we observe interference. If you have an alternative with explicit collapses or hidden variables, we can test that!” Actually, any hidden-variables interpretation without explicit collapse (e.g., something like Bohmian mechanics, rather than, say, a GRW or Penrose stochastic-collapse approach) is also, by construction, going to agree perfectly with all the observable predictions of quantum mechanics. There’s obviously no test that can distinguish between two interpretations of a physical theory that are constructed to give exactly the same observable consequences.

    On slide 35, in addressing the not-so-silly objection “How do classical worlds emerge?”, you write “The ‘preferred basis problem.’ Roughly the answer is because interactions are local in space, allowing some configuration to be robust and not others.” The question of the emergence of a classical-looking macro-reality is not the preferred-basis problem — these are two totally different problems you are conflating here — and, indeed, the classical-emergence problem is a problem for any no-collapse interpretation of quantum mechanics, including those that involve hidden variables. See, e.g., books like this one, or this one.

    The preferred-basis problem, on the other hand, is specially a problem for the many-worlds interpretation, and refers to the fact that we can trivially expand the overall state vector in any of an infinite set of different choices of basis for the overall Hilbert space, where each basis paints a very, very different picture of what those different worlds are and what are the probabilities associated to them. There is no guarantee that most of those choices of basis will involve basis states that look classical, and, on the other hand, there may well be two (or more) choices of basis that give classical-looking basis states and thus inconsistent sets of classical realities that are not related to each other in a classically understandable way.

    The preferred-basis problem is unsolved, and probably unsolvable without somehow adding on more axiomatic principles to the many-worlds interpretation for choosing one basis over all the others.

  15. John Call says:

    kashyap vasavada,
    no, what I’m saying is that when there is the posibility of more than one outcome (if a person is the “particle” in the system), all those playout in one universe or another. I personaly do not choose my path because I am not outside the system, the universe is my enviroment. If there were someone (God, if you will) outside the system, then all posibilities would collapse into one. Sinse most physicists do not believe there is a God (or anything outside the universe to be an observer), the system does not collapse into one outcome, so there is a multiplicity of worlds all playing out every possible scenario.

    At quantumn sizes, every possible scenario plays out (most “paths” cancel out because they are opposites of one another). Once the system is observed all paths collapse into the most likely path.

    I can see how this appears like predestiny or that the observer is choosing the outcome for everyone. But this is not the case, the outcome is determined by what is most probable not by what I see. The only thing that is determined by what I see is that one particular event is occuring, rather than many at once.

  16. John Call says:

    I think we need to all remeber that these are Sean’s slides, not his whole presentation. (I assume he does not read directly off his slides 😉 ). Many of the comments about the presentation have been great; however, some seem to be things that were probably cleared up in the actual lecture. Just my two cents.

  17. Eric Shumard says:

    Bad pun time: are those kitty kets?

  18. I think these slides are really fantastic! I’ll link to them in the future.

    Coincidentally, I just wrote up almost precisely the same thing for a general audience the other day and put it on Quora.

    If I were to make any suggestion, it would be to tie in some more physical example to the cat and dog. Most people have heard of the double-slit experiment at some point. At the end, you can say “In the double slit experiment, we shoot a free electron at a crystal and observe interference at some ammeters. The cat is the electron. The food and post are two layers of a crystal that the electron bounces off. The sofa and table are very sensitive ammeters that detect the electron. The dog is the spin degree of freedom a bound electron in the crystal which interacts with the free electron.” I think this might be helpful since dogs and cats don’t actually interfere, and it leads into a short discussion of why not.

    Also, it might be helpful to drop some of the early content on history in favor of elaborating the cat/dog experiment with more examples. This just keeps things tight and focused for the entire talk, so that people only have to try to learn one thing. For example, you could then do a “quantum eraser” experiment where the entanglement between dog and cat is reversed and the interference for the cat comes back again.

  19. DEL says:

    As a person reared on classical physics + engineering-freshman QM & SR, I could never swallow QM without feeling a great discomfort of the esophagus. That, despite honest efforts to independently continue technical studies of it. (I did succeed to learn GR, though.) Part of the problem, I think, lies in the interpretations: both the Copenhagen and the MW seem outlandish to me.

    Of course, the undeniable explanatory and predictive powers of QM, and its success in becoming an indispensable basis for physics’ current paradigm, all but eliminate the odds that it might ever be abolished. The interpretations, however, are another matter: since they do not affect the way the formalism is applied in order to produce explanations and predictions, the latter cannot decide their validity. The very fact that nowadays, some 90 years after the completion of QM, there still exist two contesting, undecidable, interpretations, each with its ardent partisans, tells me that the adjective ‘correct’ is applicable to neither.

    I can wait. Maybe sometime in the future some genius will come up with a new idea and a aha! interpretation. I may not live that long, but I don’t really care, because in the meantime I have my own interpretation: The QM formalism is just a mathematical instrument for producing quantitative explanations and predictions in the realm of the very-small—a generally valid instrument, marvelously effective and precise, yet still an instrument. The wave function does not “collapse” upon observation, as it isn’t a collapsible physical object. In a similar vein, the different potential outcomes allowed by the wave function do not “exist” in different worlds—their existence is purely of a mathematical nature and has thus nothing to do with “worlds.”

  20. kashyap vasavada says:

    @ John Call:
    Thanks for your detailed arguments, although I am not convinced by many worlds interpretation.Actually, I am in good company.Remember, Sean’s article “most embarassing graph in modern physics”. There is no consensus on interpretation of QM.
    In fact Copenhagen interpretation got more (42%) votes than many world interpretation (18%). Even Weinberg says that he is not happy with any interpretation of QM. So some more debates and work are necessary!!!

  21. Jack M says:

    You all might enjoy James Ladyman explain that while he intuitively is antagonistic to MWI, he sees that he’s “forced” to take it as most likely accurate.

    For the more advanced, David Wallace has recently published a tour de force defense of MWI against all comers.

    Sean, you once participated in an effort to explain QM in 5 words. Here’s my entry with a decidedly MW flavor: Multiple outcomes from one measurement.


  22. John says:

    It is hard to find a good description of the MWI even after reading a lot of pop physics books. So, I appreciate what your doing here, because I think there has been a lot of unreasonable bias against a lot of them from this. Although, I still don’t buy the MWI from this.

    To me it seems unfair that physicist say that Quantum Physics is extremely accurate, because there have been no experiments that can disprove it. When it accounts for many worlds that are not even in our own, how could it? I tend to think there is just something missing from it that tells us why particles decohere in some instance and do not in others.

    Say, I sent an electromagnetic wave down a half wave guide, the wave will not travel down that path, ever! Did I just send a probability wave into another universe by doing this? No, I think not, otherwise radars would have a lot of unwelcome electromagnetic interference. I think it was just able to interfere with itself before it went down the guide. Time is relative, and for particles traveling the speed of light, they have a complete lack of it from their own frame of reference, at least that is my hypothesis or interpretation of it all.

  23. John Duffield says:

    Sean: I thought the dogs and cats were silly, and that giving a sales pitch for many-worlds is effectively peddling woo. Sorry, but the presentation smacks of quantum mysticism that’s past its sell-by date. IMHO you should talk to guys like Jeff Lundeen, see and clock this:

    “With weak measurements, it’s possible to learn something about the wavefunction without completely destroying it. As the measurement becomes very weak, you learn very little about the wavefunction, but leave it largely unchanged. This is the technique that we’ve used in our experiment. We have developed a methodology for measuring the wavefunction directly, by repeating many weak measurements on a group of systems that have been prepared with identical wavefunctions. By repeating the measurements, the knowledge of the wavefunction accumulates to the point where high precision can be restored.

    So what does this mean? We hope that the scientific community can now improve upon the Copenhagen Interpretation, and redefine the wavefunction so that it is no longer just a mathematical tool, but rather something that can be directly measured in the laboratory”.

  24. John Call says:

    kashyap vasavada
    Yes, I deffinetly agree, there remains quite a bit of work to be done. I myselef am a fence sitter, it seems to make sense to me but I have a hard time swallowing it. It is especcially difficult when applied to macroscopic things. Also, I don’t yet have the mathematical training to study QM in any great detail beyond the concepts presented by analogy. Thanks for the discussion, it has been quite insightful and interseting.

  25. Actually, it was Schrödinger who first showed that the Schrödinger and Heisenberg pictures were the same thing, not Dirac. At least, Moore claims this in his biography of Schrödinger.