I think it might be worth our while to take a little look at what’s at the heart of all this confusion, namely, the way quantum systems evolve in time and what happens to them when they are measured, and indeed why there is such a controversy regarding the proper ‘interpretation’ of quantum mechanics.
In this post, tim314 presents a thorough and nevertheless very readable overview of what’s commonly called ‘the measurement problem’, which I’ll be shamelessly cribbing from; however, I’ll be leaving some things out to provide a little more of a ‘digest’ of the issues at hand.
First of all, let’s look at the behaviour of quantum systems. Essentially there are (or at least, appear to be) two different processes by which these systems – and accordingly, their wave functions – evolve in time, commonly referred to as:
[ol]
[li]Unitary evolution[/li][li]State reduction[/li][/ol]
Unitary evolution is basically what the Schrödinger equation (or its relativistic versions) tells us how wave functions are supposed to behave; in a way, it means something like ‘probability conserving’, i.e. if you find a quantum system with some probability in a given state, and then let it evolve unitarily for some time, you’ll find it with the same probability in that unitarily evolved state.
State reduction, now, is basically what happens when you measure a quantum mechanical system and find it to be in some definite state (decayed or undecayed, spin up or spin down, any of those classic examples will do). If you repeat your measurement, you will find the system in the same state. Thus, it is clear that this can’t be a unitary process – where we first (before the first measurement) had a system that could be found with different probabilities in different states, we now have a system that will be found with certainty in one single state (hence, state reduction): probability has not been conserved; the wave function is collapsed.
Just at a point of interest, this is also what’s really meant by ‘quantum leap’ (since the system ‘leaps’ from a superposition of states to a definite one), not, as people commonly think, the phenomenon of tunneling, where a quantum object crosses a barrier it could not classically overcome.
Now, however, the odd thing is (if we all could convince ourselves for a second that what’s been said so far is not odd) that a measurement isn’t really anything but an interaction with some kind of a measurement apparatus – a Geiger counter, for instance --, and that measurement apparatus is itself just an ordinary thing made up out of protons, neutrons, electrons, photons, in short, quantum objects – which ought to follow Schrödinger’s equation and evolve unitarily! What’s more, the interaction – however general we wish to frame this – between measured system and measurement apparatus should similarly be unitary.
This is what Schrödinger’s cat highlights: if all of the processes involved are, indeed, unitary, then, in concordance with the decayed/undecayed atom, the Geiger counter should itself enter a superposed detected/not detected state, the poison vial become broken/unbroken, and the cat dead/alive, which seems obviously absurd and wildly contrary to everyday experience (and measurement, since we do get definitive results from our observations and don’t enter weird superposed states of ‘happy that the experiment worked’/‘worried how to explain all the wasted grant money’).
Thus, Schrödinger’s cat, the way I see it, doesn’t highlight a problem with the Copenhagen interpretation in and of itself (and indeed, if you ask two different physicists on the particulars of that interpretation, if you probe long enough, you’ll probably end up with an awful tangle of superposed viewpoints), but the need for an interpretation: Unitary evolution tells us that quantum systems may enter weirdly tangled, superposed states, yet, in everyday experience, things don’t seem to work that way – how could this possibly be consistent with each other?
So, what’s happening, there? How do we get this apparent non-unitarity from a bunch of unitary processes (or one single, big unitary process)? That, in a nutshell, is what the measurement problem is all about. How this problem is treated, in my opinion, gives the clearest distinction between the different interpretations of quantum mechanics.
Now, the Copenhagen interpretation (though somewhat depending on whose interpretation of the Copenhagen interpretation you follow; see below) treats the collapse as something real, though it isn’t all that clear to me if it’s supposed to be ultimately brought on by the subjective nature of probability, or by quantum mechanics ending where the measurement apparatus begins. Under the first interpretation, the wave function only represents our knowledge of the system, and that knowledge changes upon making experimental observations, which in turn is responsible for the changes in outcome probability. As an analogy, consider the (in-)famous Monty Hall problem: when Monty opens a door and offers you to switch, the probability of the remaining door concealing the price changes from 1/3 to 2/3, prompted by additional information being revealed about the system, which would be analogous to a measurement having been made; however, one has to be careful in making such analogies not to run afoul of the severe restrictions on the possibility of hidden variables in quantum mechanics. According to this interpretation, the cat is either dead or not dead, and the mathematical description of it as being both is merely an artefact of nobody having checked so far, our knowledge of the system thus being incomplete.
On the other hand, it’s sometimes said that the Copenhagen interpretation treats the measurement apparatus and/or the observer as classical objects, simply neglecting their quantum nature, and sure enough, this would also account for the violation of unitarity, since it only holds in quantum contexts; however, I’m not actually sure it’s right to claim this, mainly because I’m not so sure what’s taken to be ‘the’ Copenhagen interpretation in this case. However, this is where the paradoxical nature of Schrödinger’s though experiment comes in – essentially, it shows that it is impossible to draw any kind of consistent boundary between systems that are to be considered ‘classical’ and those for which a quantum mechanical treatment applies, by virtue of taking a supposedly classical system – the cat – and placing it in a quantum mechanical context.
Another interpretation of this type – that explicitly accepts the reality of fundamentally non-quantum things – would be the ‘Consciousness Causes Collapse’ conjecture that has been mentioned previously. (By the way, strictly speaking, I don’t consider this to be an ‘interpretation’ of quantum mechanics, since it assumes the existence of things that go beyond QM, and possibly beyond physics in general.) This essentially postulates a rigorous mind-matter dualism, asserting that the rules of quantum mechanics (and with them, unitary evolution) only are valid for matter, and that minds are ‘something else’, and capable of collapsing the wave function. This is the basis for a lot of quantum mysticism along the ‘we create our own reality’-line, and frankly, I think it’s somewhat fanciful, to say the least, and jumps to unwarranted conclusions, promoting confusion about the implications of quantum mechanics that lead to dreadful, dreadful films being made.
However, most people, when asserting the wave function collapse to be an ‘observer effect’, are somewhat implicitly (and often, unknowingly) arguing from such a position, equating (needlessly) observer with conscious observer, and also being under the mistaken impression that an ‘observer effect’ is some mysterious quantum hocus pocus. I have often seen the sentence ‘the act of observation changes that which is observed’ quoted as some almost zen-like, confounding pearl of wisdom, when in reality it’s a perfectly mundane thing – even when you just wire a voltmeter into an electric circuit, you have an observer effect, since the voltmeter is now part of the circuit, changing it, and thereby the very voltage it measures. (Also, observer effects are not the reason for Heisenberg’s uncertainty relation, even though it’s often painted that way, but that just to address some general confusion.)
So, while the wave function collapse would be an observer effect under the CCC interpretation, and under flavours of the Copenhagen interpretation opting for a classical treatment of measurement apparatus and experimenter, in general, it isn’t, and even if it is, it doesn’t have to be anything mysterious.
Another possibility is to reject the notion that a collapse happens at all. ‘Many Worlds’-types of interpretation are of this kind, however, it doesn’t necessarily give rise to any parallel universes or something like that. The idea is, basically, this: the radioactive atom enters a superposed decayed/undecayed state, causing the cat to enter a dead/alive state. Now, an observer opens the box, and, his ‘observation’ being a measurement and therefore, an interaction, immediately becomes himself entangled with the whole quantum system, entering a ‘sees dead cat’/‘sees living cat’ state. So, now, the states describing the system are ‘decayed, dead, sees dead cat’ and ‘undecayed, alive, sees living cat’; these states are decoherent from each other, i.e. there is no interaction between an observer seeing a dead and an observer seeing a living cat. From the point of view of any given observer, everything would appear perfectly consistent – he’d either see the broken poison vial and a dead cat, or he’d see the vial whole and the cat alive. There is no ‘sees dead cat and sees living cat’ or ‘sees cat simultaneously alive and dead’ state. One way of thinking about this is to ‘shunt’ each observer into his own universe, giving rise to the whole ‘parallel universes’ thing, but that’s not in fact a necessary consequence of interpreting things like that, though it’s often presented that way.
Another way of rejecting collapses is to propose something like the ensemble interpretation, which basically states that the probabilities only apply to repeated instances of one and the same experiment – so that, if you were to conduct Schrödinger’s though experiment a number of times, what the wave function essentially predicts is the probability distribution of all the outcomes of these experiments – 50% of the time, you get a dead cat, and 50% of the time, you get scratches on your face, in accordance with what the formalism tells you (well, strictly speaking, the formalism doesn’t tell you that you’ll end up with scratches on your face, but have you ever tried shoving a cat into a small box repeatedly, regardless of whether it wants to or not?). Thus, the wave function doesn’t apply to any individual system, and trying to think about it that way just gets you in an awful tangle.
There are other possibilities, such as the transactional interpretation, which essentially allows for wave functions travelling both forwards and backwards in time, giving rise to quantum mechanical interactions through self-interaction in the manner of a standing wave, where the collapse doesn’t occur at any fixed point in time, or the aforementioned ‘objective collapse’-theories, where collapse happens statistically (though Stranger is right about the difficulties in finding an appropriate measure for ‘macroscopicness’); but this is the gist of it, at least so far I understand things. I hope, by the way, that I’ve gotten things at least sufficiently correct so far; I’d appreciate any corrections.
Anyway, I’m a bit afraid that this essay’s gonna achieve little else but kill the thread, but I needed something to get my mind of a couple of things for a while, and that at least I can say I have accomplished.