# Why isn't Manyworlds obviously correct?

1. Jul 10, 2014

### GofG

First, Schrodinger's Cat:

Photon emitter, pointed at a half-silvered mirror, such that 50% of the wave hits a sensor that activates a gaseous poison emitter inside a box that has a cat in it.

Standardly, we think about it this way: the photon hits the halfsilvered mirror and decoheres into a superposition of "straight" and "reflected", then the sensor decoheres into a superposition of "hit" or "not hit", then the poison emitter decoheres into a superposition of "activated" or "not activated", then the cat decoheres into a superposition of "dead" or "alive"... so the cat is both dead and alive, etc.

Isn't it obvious that upon opening the box, the human decoheres into a superposition of "sees an alive cat" and "sees a dead cat"?

Isn't that blindingly obvious? If the cat can decohere, why can't the human?

And that's all Manyworlds is, is assuming that decoherence doesn't suddenly "stop" at macroscopic levels.

Can someone explain to me why this isn't selfevident?

2. Jul 10, 2014

### UltrafastPED

Because none of it is observable. There is no experiment which will prove/disprove "Many Worlds".

PS: Shroedinger invented his cat puzzle because it was ridiculous. It was never intended as a serious problem for QM. Rather it is taking simple ideas to ridiculous extremes.

3. Jul 10, 2014

### GofG

We've observed decoherence up to ridiculously large systems (I believe 45nm was confirmed recently?), why on earth should we assume it doesn't apply all the way up?

Why would decoherence suddenly "end" just because a system gets large enough?

I feel like we should assume macroscopic decoherence until we see a counterexample. The math doesn't describe a point where decoherence stops happening; if we take schrodinger's equation at face value, then waves of amplitude *is* reality. Why postulate, on TOP of S's equation, that huge chunks of the wave get pruned away by an unknown, nonlocal, faster-than-light process?

Or are you going to wait for macroscopic decoherence to get verified at the 1mm level? The 1cm level?

What if your theory is right, but decoherence actually stops at, say, the 1km level, so we aren't actually large enough to collapse the wavefunction? Doesn't it seem suspicious to you that, in all three of these scenarios (macroscopic decoherence, decoherence stops at some point between an atom and a human, decoherence stops at some point larger than a human) we observe the same experimental outcome?

I guess my question is: what reason do scientists have for believing that, against the math, at some point of size decoherence no longer applies?

4. Jul 10, 2014

### atyy

Last edited: Jul 10, 2014
5. Jul 10, 2014

### GofG

The only real complaint I see listed on that page is that MWI doesn't explain the Born probabilities. This is true, but the Born probabilities are a mystery, not just for manyworlds, but for all interpretations, including Collapse interpretations. That is, the fact that we experience a given state by the squared modulus of that state's thickness in the wave is not *evidence against* manyworld.

In fact, that page pretty well defined my own stance: manyworlds are simply a part of Schrodinger's equation, period.

Are there any objections to macroscopic decoherence, other than the fact that it doesn't explain the Born probabilities?

Last edited: Jul 10, 2014
6. Jul 10, 2014

### atyy

7. Jul 10, 2014

### Jano L.

Using this approach, bodies should interact instantaneously too, which contradicts electromagnetic theory and experience. To insist that every math concept of the theory has counterpart in the real world is unreasonable.

8. Jul 10, 2014

### GofG

Huh? I don't see how this "approach" (taking the math at face value) says that bodies should interact instantaneously. Can you give me an example?

I just finished reading that paper... I don't quite understand what you're saying. QM as described by Schrodinger's Equation is deterministic; we know exactly how thick each slice is. This doesn't suggest that each Everette branch is clearly distinct from each other Everette branch; if this were the case, a single photon wouldn't be able to interfere with itself. I don't think I'm claiming that the branches are distinct, but when you get up to the macroscopic level of humans, there isn't really any interference from one branch to another (ie, the slice-of-the-wavefunction which is you-seeing-the-dead-cat doesn't causally affect the slice-of-the-wavefunction which is you-seeing-the-live-cat).

My question stands. In order to reject manyworlds, you must posit one of:
1) Once a system is large enough, where 'large' is defined as having a certain number of interacting particles, the system stops undergoing decoherence
1a) and usually we posit that "large enough" is somewhere between an atom and a human, so that we can cleanly state that we aren't constantly being split into multiple copies of ourselves
2) Observation/measurement acausally, faster-than-light, nonlocally, nonrelatively 'collapses' the wavefunction down to a single point.

Both of these points seem absurd to me (the first position seems isomorphic to the second in that it's magic, except dressed up in sciency-sounding terms), so what I'm asking is: are there any other reasons why one might disbelieve in macroscopic decoherence (ie manyworlds)?

Last edited: Jul 10, 2014
9. Jul 10, 2014

### martinbn

I don't understand why you equate decoherence and manyworlds!

p.s. From your first post I see that you use the word decoherence in a unorthodox way. Quantum mechanical systems do not decohere into superpositions.

10. Jul 10, 2014

### GofG

Decoherence is when two blobs of amplitude get separated from each other enough that their interaction approaches zero, correct? So if we have two particles whose positions are highly correlative of each other (either the particles are "both on the right" or "both on the left"), then they've decohered into two states: (Particle A - Left + Particle B - Left) * (Particle A - Right + Particle B - Right).

Now, there's still a very small 'probability' (a very very thin blob of amplitude) linking the two, such that we might end up observing Particle A going left and Particle B going right, but if the two blobs of amplitude are separated by enough space then the chance of us observing that is very slim, because the two blobs are decohered.

Now, if you add a human into the mix, a human observing the two particles, then the system becomes even *more* decohered, because the human brain's final state depends heavily on the state of the particles. So the new system is (Particle A Left + Particle B Left + Human Sees Left) * (Particle A Right + Particle B Right + Human Sees Right).

If the amplitude blobs that are what we think of as the particles can decohere, why can't the amplitude blobs that are what we think of as the two particles plus the human decohere?

In which case, Schrodinger's equation tells us that the particles have entered a superposition of L+L and R+R, and the human has entered a superposition of L and R, with some extremely thin/wispy/light-grey blobs of amplitude left over for stuff like R+L,R or R+R,L.

Edit: For clarification, the reason I am equating manyworlds with decoherence is that, when the 'possible state' of two particles (left and left, or, right and right) decoheres, anything that depends on the value of the state ends up on one side of the fence, or the other. Both possibilities, R+R and L+L, happen; the blobs of amplitude in configuration-space look exactly the same, there's no way to use math to figure out which one of them is "real" and which one of them "isn't real". In exactly the same way, if a human observes this system, then the human's final state is heavily dependent on the state of (R+R * L+L), because the two different states are no longer causally interacting with each other. If they were still causally interacting with each other (as in the dual slit experiment), then we would see both (as we do in the dual slit experiment). But they're not; they're decohered. Which means that the *new* state, instead of being (Particle A Left + Particle B Left) * (Particle A Right + Particle B Right), is (Particle A Left + Particle B Left + Human Observes Left) + (Particle A Right + Particle B Right + Human Observes Right).

There are two humans in that system! One of the humans observes the particles on the left, one of the humans observes the particles on the right, but Schrodinger's equation describes both humans equally well! and says nothing about which one is real. If the human in question believes in the collapse interpretation, then the Human in the amplitude-blob (Particle A Left + Particle B Left + Human Observes Left) thinks to himself, "Oh, so I collapsed the particles to the left and made that outcome 'real'." while the Human in the amplitude-blob (Particle A Right + Particle B Right + Human Observes Right) thinks to himself "Oh, so I collapsed the particles to the right and made that outcome 'real'."

Last edited: Jul 10, 2014
11. Jul 10, 2014

### atyy

That was not your question. Your question was why isn't many-worlds obviously correct. So the question was not about rejecting many-worlds, but about accepting many-worlds. There is no certainty that many-worlds us technically correct, even by proponents as shown in the above links by Sean Carroll and David Wallace.

12. Jul 10, 2014

### GofG

Sean Carroll says, in the page you linked:

Quantum mechanics *is* manyworlds. If you want to believe something other than many worlds, you have to add something to quantum mechanics, like a "collapse postulate".

So what I'm saying is, what evidence is there for adding something like a collapse postulate? Why should we add something to a theory that already perfectly describes our observations and experiences?

The "technical problems" for MWI that Sean Carroll describes are:

He then goes on to say that the "Why certain kinds of worlds and not others" question is solved by pointer states, essentially by the fact that macroscopic systems are robust. The Born probability problem is tied closely with the "why probabilities at all" problem, and neither are solved by MWI, but adding stuff to MWI, like a collapse postulate, does not fix the problem. It's like saying "little angels come in and erase all of the worlds that aren't real in a manner consistent with the Born probabilities." It doesn't explain why the Born probabilities are what they are; from a Kolmogorov complexity standpoint, you've only made the theory more complicated without adding any additional predictive or explanatory power.

Last edited: Jul 10, 2014
13. Jul 10, 2014

### atyy

No, that's rubbish. Something equivalent to a collapse postulate is needed. That is standard Copenhagen. That is problematic, but a cheap claim that many-worlds obviously solves the problems does many-worlds a disfavour, considering the hard and serious work that people like Deutsch and Wallace have put in.

14. Jul 10, 2014

### GofG

What "problems" are you talking about? The Born probabilities? The Collapse postulate does not solve those either. By postulating collapse, you've added something new onto Schrodinger's equation that makes it strictly more complicated without increasing its predictive or explanatory value.

In order to explain the Born probabilities, we need to add something on which explains why the probability of experiencing a slice of the wavefunction is equal to the squared modulus of the thickness of that slice. It seems like our probability of experiencing a slice should just be equal to the thickness; that's what MWI predicts.

But adding on a collapse postulate that reads "when a system gets large enough, all of the blobs of amplitude except one are destroyed, in accordance with the Born probabilities" doesn't solve that problem. It doesn't explain why the Born probabilities are used, instead of something else. That's what I mean when I say it makes the theory strictly more complicated. The theory "Schrodinger's Equation + Born Probabilities" is the strictly simplest theory we have right now, and it implies manyworlds. If you want to *explain* the Born probabilities, and why they're different from what we'd expect them to be, then you can't do it by saying "Something does something, as per the Born probabilities" instead of just saying "The Born probabilities".

I'm not saying it's a complete theory, I'm just saying that basically everything postulated up until this point has been isomorphic to saying 'Little angels come along and destroy every blob of amplitude, selecting one blob to be the 'real' blob, based on the squared modulus of that blob's density'. Nothing has explained why the Born probabilities are used, instead of something else, like the straight thickness instead of squared thickness, or something completely ridiculous like "assign each distinct blob of amplitude a number based on its density, divide each blob's number by the total number of leaves on every tree in the universe, pick the third blob of amplitude whose density, when represented by a real number 0-10, has the third digit after the decimal point being 5."

There's a lot of possibilities for what the probabilities might be; the real question is why pick the squared modulus? And you're never going to solve that by tacking on a mechanism that uses the squared modulus, but could use some other criteria.

Last edited: Jul 10, 2014
15. Jul 10, 2014

### Staff: Mentor

No, that's not the difference between Manyworlds and other interpretations. Other interpretations include decoherence too: they have to, because they have to explain why only certain alternatives are available for the wave function to collapse into. The difference with Manyworlds is that there is no collapse.

16. Jul 10, 2014

### WannabeNewton

Decoherence is part of the QM formalism. It is just good old fashioned physics. No one doubts the validity of decoherence as it is just a framework of calculations derived from standard QM. MWI is an interpretation of QM that uses decoherence but it is not the only interpretation that does so. MWI makes assumptions that are not derivable from QM in order to attempt an explanation of the measurement problem (but not the preferred basis problem). You seem to be conflating the two. MWI is no more valid than any other interpretation of QM that reproduces the predictions of QM and it is certainly not "self-evident" that MWI is the end all be all of the measurement problem.

So before you go about discussing MWI, you should first understand the difference between MWI (an interpretation) and decoherence (just physics of QM). A good place to start would be: http://arxiv.org/pdf/quant-ph/0312059v4.pdf

17. Jul 10, 2014

### Staff: Mentor

I don't think MWI "predicts" this; I think MWI *assumes* this, without proof. But I don't see how the concept of "our probability of experiencing a slice" even makes sense in MWI, because in MWI, by definition, there is a copy of "us" that experiences each possible outcome of any quantum interaction. A particular copy, experiencing a particular outcome, has no way of testing, experimentally, what the relative "weight" of that outcome is, because there is no way of testing what other copies exist and "how much" of each copy exists.

18. Jul 10, 2014

### GofG

Alright, I admit that decoherence does not equal manyworlds, but that's not what I'm saying.

I don't see how MWI makes any assumptions. It seems to cleanly solve the measurement problem (by saying that there is no collapse, and that there is a version of you who sees all possible outcomes) without postulating any additional assumptions. That is, if you take the state of a setup with a photon, a half-silvered mirror, a sensor, and a human observing the whole thing, then if you are only using Schrodingers Equation, then you end up with a description of two separate human brains, one in the state of observing the photon get reflected and one in the state of observing the photon go through the mirror. If you could input into Schrodinger's Equation the entire state of such an apparatus, including all of the particles in the brain, before the experiment, then it would output the total state of two brains. It is an additional assumption on top of this to assume that one of these brains is real, and the other isn't real.

Can you point me to where my assumption is?

Even though different versions of you will experience every possible outcome of a given experiment, you can only experience one of those outcomes. This is simply due to the fact that our brains are built on macroscopic physics, and do not function as superpositions of themselves. I think you are confused about the meaning of the word "probability". If I am set up with a half-silvered mirror, a photon emitter, and two sensors, then I have specific rules as to how much I anticipate seeing the photon go left as opposed to go right. I call these differently-weighted anticipations "probabilities", and they are proportional to the density of the amplitude-blobs which describe those states.

I am not saying that other interpretations don't have decoherence; I am saying that, for the most part, they describe some additional postulate whereby the various decohered blobs of amplitude collapse, such that one is made real (and that's the one you actually experience) and the rest are made unreal. Manyworlds simply says that you experience all of the amplitude blobs; that there is no level-of-complexity-or-largeness where collapse happens.

19. Jul 10, 2014

### Jano L.

Consider the Schroedinger equation for hydrogen atom

$$\partial_t \Psi (\mathbf r_1, \mathbf r_2, t) = \frac{1}{i\hbar} \left(- \frac{\hbar^2}{2m_1}\Delta_1-\frac{\hbar^2}{2m_2}\Delta_2 - \frac{e^2}{4\pi \epsilon_0} \frac{1}{|\mathbf r_1 - \mathbf r_2|} \right) \Psi (\mathbf r_1,\mathbf r_2,t).$$

Interaction of the electron with the proton is described by the last term in the bracket. If we wrote down similar equation for large many-particle system, there would be many such terms, but they are always functions of positions $\mathbf r_1,...$ only - there is no delay and no corresponding speed of interaction propagation, no matter how far the particles are.

Projecting this mathematical property of the equation onto the world one would claim that the interaction *is* instantaneous. But in fact, we know that electromagnetic interaction propagates with the speed of light. The electrostatic interaction term is just an approximation.

So be careful about applying mathematical theory in physics - not every mathematical property of the theory is physically valid.

20. Jul 10, 2014

### Staff: Mentor

Well, sure; but that's not the problem. See below.

Yes, and in actual experiments, these anticipations, which are based on the Born rules, are borne out (pun sort of intended ). But I don't see how to get them from the MWI, without *assuming* the Born rules. It seems to me that the MWI predicts that there are copies of me who experience *every possible sequence of outcomes* of, say, the half-silvered mirror experiment you describe; there should be some copy of me, for example, that experiences every single photon going left, even if the underlying amplitudes are equal for left and right.

In other words, if I use the MWI, I can't actually justify deducing amplitudes from observed frequencies, *unless* I add the Born rules as a postulate. Without the Born rules, if I see half the photons going left and half going right, the only thing I can deduce is that the underlying amplitude has *some* nonzero coefficient for both the left and the right state (because if either one were zero, I would only observe the other). But any pair of nonzero coefficients is consistent with what I observe, because any pair of nonzero coefficients will give rise to a "world" in which I observe half the photons going left and half going right. And, conversely, even if I see all the photons going left, I can't deduce that the underlying amplitude has a zero coefficient for the right state, because a state with, say, equal coefficients for both left and right would still give rise to a "world" in which I observed all the photons going left (as above).