Quantum Interpretation Poll (2011)

stevendaryl

I think the point is that in classical (Bayesian) probability, there is an interplay between the ontic and the epistemic. You assume that there is some unknown actual arrangement of cards, and your probabilities are updated to reflect your knowledge about this arrangement. It's hard to understand how you can have epistemic probability without a corresponding notion of what is really the case.

al onestone

The point I have made concerning the information interpretation and the "projection postulate" is one which is strict to the case of quantum interpretation. I am talking strictly about the issue of correspondence between description and system, unlike other interpretations of quantum mechanics which do not address this issue in favour of interpreting the description, the mathematics. They only do so with the assumption that the mathematics somehow is assumed to be the real picture.
In the information interpretation we consider the mathematics to be important in that it reproduces the results of the experiments of quantum mechanics, but does the math really correspond directly to reality as the EPR paper would assume? In the info interpretation the correspondence is provided between system and description by simply stating that the system IS the information which the description provides (which can be formulated into elementary propositions) rather than looking for the description (which is very very abstract, probabilistic, etc.) to be describing the system directly. In this interpretation, correspondence is via information theory and information theory ALONE.
So when we say that the system is an amount of information, don't expect for that information to be anything more than the description : propositions like; the position is X, the momentum is Y. Not propositions about the laws of quantum mechanics, that is information too, but not the specific information that the system IS. So forget about probabilities, and poker, and information like " if X and Y then Z". These are not the info that the info interpretation sais that the system is.

malreux

Sorry to randomly but in: just to interject that I don't think the so-called 'projection postulate' is actually a part of the formalism of QM. Rather, it is better understood as an ad hoc rule by which (in the absence of a more principled treatment [as a card-carrying Everettian, you know where I'd think one might be obtained]) we extract empirical content from the formalism. Further, it isn't even any more a particularly good reflection of practice in physics: any serious treatment of the quantum physics of measurement uses something much more sophisticated.

malreux

Modern treatments of quantum measurement use much more sophisticated tools.

For an introduction to this area see wiki's: http://en.wikipedia.org/wiki/POVM

malreux

Consider the following theory of fossils - we are not entitled to view fossils as evidence for dinosaurs, but rather we have available a certain well-known formalism that we can use to model found-fossils and also make predictions about fossils yet to be discovered. Of course, even creationists would find this theory a little strange. If our theory calls for certain complexes of book-keeping devices that act upon certain other such complexes, we can cut down on inelegant talk by considering naming such complexes as e.g. 'dinosaurs' and, noting such 'dinosaurs' appear to act on, and also be acted upon by, other 'dinosaurs'. (Evidence for which is extracted from examining fossil records) Our theory might in turn be underpinned by a principle that runs something like 'if one real thing (dinosaur a) is acted upon by another thing (dinosaur b) then the latter is also real'. This new theory - of real dinosaurs - seems much more elegant then the complicated instrumentalist approach to a theory of fossils. Incidentally, it also much more in keeping with how we ordinarily view all the other sections of science.

malreux

[1] This doesn't appear to mean anything! Anything MWI has to offer is less crazy and more in keeping with the history of science than the idea that the Potential affects the Actual. On the other hand, every interpretation of QM that doesn't require us to modify the formalism or take an anti-realist stance on the scientific enterprise must incorporate those features interpreted as 'worlds' in the MWI, but simply offer an argument at to why we regards these structures as 'Potentia' say, rather than 'real'. For example, one can introduce Bohmian particles to 'point at' the favoured structure, etc. (The most principled modificatory strategy is dynamical collapse, which makes other branches 'wither away')

[2] This is utter tripe, for a modern discussion see e.g. Wallace (20120

malreux

On this philosophy-of-science point: can this distinction really be maintained? For example, modificatory interpretations of QM actually introduce novel formal devices, and so might be better seen as different theories (different, for example, to unitary quantum mechanics). Even no-collapse interpretations might be better construed as different theories - for example, though dBB recovers the quantitative / empirical predictions of the standard formalism, it too could be seen as a different 'theory' - albeit in this case the hidden variables are idle wheels in both formal and empirical senses, perhaps the difference lies in explanatory motivations... Regardless of the resolution of the latter issue, I dislike the term 'interpretation'. In my example of a theory of dinosaurs above, it would seem strange to talk about an 'interpretation of the fossil record'.

sigma.alpha

Fuchs
...The quantum system represents something real and independent of us; the quantum state represents a collection of subjective degrees of belief about something to do with that system (even if only in connection with our experimental kicks to it). The structure called quantum mechanics is about the interplay of these two things|the subjective and the objective...



.

Ken G

My point with the cards is that it is not just objective evidence we are using (based on a random deal of given cards, some of which we've seen), but our impressions of the actions of our opponents. Two different players might assess the probability of a hand very differently as a result, and as they are counting different things as important, they will not even categorize hands into the same sets (for checking their probability determinations) as someone else. So both could conclude their probabilities are working for them, and are checked by many trials, even though the probabilities are different. As such, there is no such thing as the "correct probability" that a given hand will win, so there is no such ontic entity as the "actual probability". Even so, there are good poker players, and bad poker players, based on their probability assessments. That quote seems to claim we should view this fact as a "miracle."

Ken G

I think so. It's true that "lines in the sand" are always hard to draw, because the situation requires more flexibility. But there is a pretty clear need to distinguish the predictions a theory makes, which are testable, from how we interpret the workings of the theory, which is not testable. Even Newtonian mechanics admits to multiple interpretations (are there really forces, which then allow us the mathematical invention of action? Or is there really action, which allows us the mathematical invention of forces? Does neither exist, and they are both just ways we can picture our theory, like imagining an image charge to help us solve for a field?) The devices don't make them a new theory, it is their predictions. If they don't make new predictions, they can't be a new theory. Here is where the sands get shifting, because some interpretive devices do sort of make predictions, but they don't guide any new observations, because the predictions are impossible to observe, realistically (like the Hawking radiation of a stellar-mass black hole, or some such thing).
That's because it's not an interpretation, it's a theory. We theorize various dinosaurs, and we make predictions about those dinosaurs. Those predictions are very different from some other theory. The predictions apply to fossils we have not yet found, or were made before the fossils we have now found.

malreux

[1] Your definitely on to something there. However, recalling Einstein's dictum that 'observables' are defined by a theory,* and glancing at the defunct position of say, positivism, one can safely draw the inference that the observable / non-observable distinction is, at best, irrelevant to a functionalist understanding of our best physical theories. On the other hand, the main 'interpretation' of scientific theories in most fields is surely some form of structural realism?

[2] By your lights Darwin's theory of natural selection was not much better than Lamarck's, at least by the standards of his day.

[3] Here you make the nice distinction do a lot of work: e.g. you seem to use the term 'theorize' to mean 'add to our theoretical ontology' whereas most people - scientist or non-scientist - would prefer the term 'discover'. Further, theories can differ whilst predicting the same phenomena - the so-called theoretical underdetermination by data. If you think theories are essentially about prediction, or that this is what makes them scientific, consider replacing a certain theory with a long sentence that omits all theoretical terms so far as it can recover them with observation statements. In practice, you will find you can rarely actually carry this through. This is a very weak argument, essentially arguing that 'if you can't simulate it, you don't know it' - but it sounds plausible enough to me...

*To slightly paraphrase

malreux

I should add that some recent collapse theories do actually make potentially testable predictions, i.e. e.g. if collapse time ≥ decoherence time

sigma.alpha

superposition time vs the time of decoherence.
if last less (sp), nonlinear models are the correct.

Ken G

There is indeed difficulty in separating a "pure" observation from a "pure" theory, as we need some kind of model of what we are doing to even read a dial. But I think it is still not so hard, as distinctions go, to distinguish between "a spot appeared on a screen, so I'm detecting an event happening at that point in space and time" and "a particle passed through one/both/neither slit along the way." The former very definitely seems like an observation, the latter, an interpretation. The structural realism certainly applies to our devices-- we all imagine the screen is real, the spot is real. But the extrapolation from what we participate in, and develop as part of what physics is, to everything that happens along the way, was always a bit of a leap of faith for structural realism. Indeed, I've argued that somehow, what seems "realistic" vs. "antirealistic", has gotten inverted-- it seems realistic to me to expect that the universe does not have to behave, when unobserved, the same as it does when it is observed (since observation is a form of making the universe fit into a particular way of interacting with it and conceptualizing it in our limited human brains), and antirealistic to be carried away on a leap of faith.
I don't think so. Darwin's approach makes lots of predictions, like if you change the surroundings of a moth to a differently colored environment, you will expect to watch the moth color gradually change over by natural selection. There is no equivalent at all in Lamarck's theory to that kind of change, simply because moths are not observed to change color in their own lifetime.
The terms ovelap-- we discovered a theory about dinosaurs. It is considered a very good theory, after some initial modifications, because it became able to predict new fossil discoveries that were indeed later found. There is an important distinction, owed to Popper, between just using a theory to "make sense of" what you observe (which is easily used to fool ourselves, a form of "rationalization" of our findings-- just look at how creationists do it), versus actually making "risky predictions" (predictions that we would have no reason to expect if we had no understanding of that theory). It is the latter that makes us trust a theory, though the former is how we use the theory on a day-to-day basis.
But they cannot predict all the same phenomena, or they do not differ. We don't regard Hamiltonian mechanics as a different theory from Newtonian, though they bear little resemblance at first glance, because they make all the same predictions.
That fails the most important test of all: Occam's Razor. The job of a theory is to predict, the job of a good theory is to unify those predictions into as simple a framework as possible, and the job of an interpretation of a theory is to "make sense" of what the theory is asserting about reality. But the latter two are not unique, and the last one is downright problematic (but we do it anyway, because no one really holds to "shut up and calculate", no one is really satisfied by that).
I believe I understand what you mean, you are saying that the simple theory is more than just a way to get the predictions in a unified way, it is actually the goal itself. It begins to sound rationalistic, like the first step in saying that nature really does follow mathematical rules, and we are just incompletely converging on them. That argument never really washed for me though, it sounds like we are simply taking our converging accuracy and mistaking it for converging on truth. I just don't see any credible argument that leads to that conclusion, it seems like wishful thinking.

Ken G

Yes, I'm not categorically claiming that's impossible, I just haven't really seen anything all that convincing that such different predictions are really plausible. But we can agree that if such predictions are possible to test, then we will no longer have to call them interpretations, we can justify the claim that they have become new theories. It's not unusual for an idea that starts out a kind of interpretation, or even just a philosophical kernel, to get expounded to the point that it does qualify as a scientific theory.

malreux

Thank you for your thorough reply. I feel (hopefully falsely!) that this line of debate doesn't particularly interest you, however.

I won't bore you with scholastic debates regarding the distinction between observation and interpretation. I'll simply reiterate my view that an answer to the question 'what is an observable?' is theory-dependent.

[1] I leapt in with my earlier post without even suggesting what I meant by 'structural realism'. All I mean by this is answered here: http://plato.stanford.edu/entries/structural-realism/

[2] Of course. However I hope I'm not just getting carried away and inferring reality willy-nilly. The way I put it above was - if we have Real Pattern A and find it interacts with Pattern B, then we can award the latter the nice distinction 'real'.

[3] Quite so: Darwin's theory has more predictions to make. However, why was Darwin always amassing evidence for evolution? Not to generate novel predictions, that's for sure. Incidentally, remember Popper at times actually comes off as a bit of an evolutionary sceptic - although one doesn't need to retain this aspect of Popper's thinking to be a good falsificationist, a la Deutsch.

I don't think anyone has ever thought the reason theorizing that fossils were records of animals once extant is attractive is solely or mainly because it might make good predictions about future fossil finds. Or at least, that's a very strange way of putting it.

[4] This is just a matter of emphasis. Explanations that are hard to vary also characterise much of science and what is scientifically good about it.

[5] Untrue, consider some pregnant comments by Wittgenstein on the subject of observation and the helio-versus-geocentric controversy

[6] There is a relevant equivalence between Hamiltonian mechanics and Newtonian physics due to predicting the same quantities, but this is hardly the only or even the most physically interesting equivalence. Also, the former is a representation of (aspects of) the latter.*

Finally, I'd like to assure you that I don't exactly care whether our theories are converging on the truth ('approximating' is the fashionable jargon), I just think that if you look at e.g. fundamental physical research, as practised, and you can actually do quite a bit of work in epistemology, metaphysics and the like. The motivation for doing so is also related to whether you find the following question interesting: 'what must the world be like for our current best theories to be ('aprox') true?'

*In this case, i.e. not generalising to 'Hamilton's principle'

DennisN

Hi, I voted "Other", but what I mean by this is rather "I don't know". I lean towards the Ensemble interpretation or Copenhagen. I also think the Relational interpretation is interesting.