Monday, April 27, 2009

So You Think You Can Philosophize: Episode 2

This week's contestants, Destro and Cobra Commander, consider whether there can be any critical experiments in science. That is, until their boss catches them.

There can be no critical experiments in science, if you believe in structural realism.

There can't be any critical experiment, no matter what your theory, if you ask me.

Well, that depends on what you mean by critical experiment.

So what do you mean then?

I mean an experiment that, if it comes out a certain way, proves your theory wrong. An example would be the 1919 test of Einstein's theory of relativity by observing how far the distant starlight was deflected by the curvature of spacetime around the Sun. If there was no deflection, Einstein would have been proven wrong. That's a critical experiment.

So it sounds like critical experiments in general have to be defined counterfactually.

Hmm, I never thought of it that way, but it would be an interesting thing to explore. What do you mean?

Well, if a theory is successful, it has passed all the experiments, or at least all the trusted experiments, that have been done to test it. If they were critical experiments, then they can only be described as such counterfactually, . I.e. had they come out differently, the theory would have had to be abandoned.

Okay, but that doesn't preclude critical tests of the theory in the future, does it?

How do you mean?

Well, the tests it has passed in the past may have to be counterfactually shown to be critical tests, but there is still the future to look forward to. The theory will be put to more tests. And if it makes predictions that differ from reality then it ruled out. In that case, it was a critical experiment, not counterfactually.

Okay, yes. But still any passed test can only be defined as a critical test artifactually, right? Can we agree on that?

Well, it seems so right now, so I'll go with it. But it may happen in the future that something comes to light that would rule it out. Moving on.

Point noted. So, what was that you were saying about structural realism? I'm afraaid I don't know quite what that is.

Oh, structural realism? It's a nifty way of looking at progress in science. You see, there is a strange tension in the progression of science. Our numerical predictions are getting more and more precisely verified by experiment. Therefore, some reasonable people say that science is getting us closer and closer to the truth. Or, rather, that science gives us the truth, in ever more fine grained detail.

Yeah, that seems right to me.

Ah, but then you have the problem of scientific revolutions. As Einstein put it, "no amount of experiments can prove me right, but a single one can prove me wrong". Now, what do you think he meant by that?

Well, it sounds to me like he's talking about critical experiments.

That's right, I think he was talking about crucial experiments in the last clause. But the first clause is curious, isn't it? "no amount of experiments can prove me right."

He was a very humble man.

Well, I don't know about that. He also was trying to figure out if God had any choice in creating the universe…not so humble if you ask me.

Alright, alright. But he's just saying in that quote that his theory will never gain 100% precision. There will always be some uncertainty, so if certainty is your criterion for his theory being "right", then you are out of luck. But that is a very stringent criterion for a theory to be considered "right". Clearly the theory is right, to the extent that your GPS unit has to use his theory to account for your correct coordinates.

Okay, I see you point. That is one way of interpreting his statement, and now I see why you think he is being humble by saying so. His theory is obviously right to some extent, but he is taking the high road but not claiming to be settled completely. Well, there's another way of interpreting his statement that you may want to hear.

Let's hear it.

Okay, so he's saying that no amount of experiments can prove him right, and that one can prove him wrong because he knows that someday his theory will be overturned. He has said as much elsewhere.


Right. But think about it: his theory was overturning Newton's in a sense, a system of laws that had been taken as gospel for hundreds of years. And yet a single experiment, the eclipse observation of 1919, was sufficient to overturn it. This has happened time and time again in the history of science. A theory is thought to be "right" but then at some point an experiment proves it wrong. So, if it can always be proven wrong at some point, then what does that say about it's chances of being right? Those chances are apparently zero.

Unless you have the correct theory of everything.

And that only makes sense if you are a unificationist, I think.


Nevermind. So, do you see the tension now? On the one hand it seems that are theories are getting us the truth, more and more precisely. Even when a theory is overthrown, its in a regime where the old theory wasn't designed to tread anyway, and the mathematics are continuous by design. Quantum mechanics mathematically reduces to classical physics in the limit of h bar being very, very small.

I see that.

But on the other hand, there are radical revolutions of theories, in which the old theory is not continuous with the new. Newton's law of gravitation explained the motions of the planets in terms of gravitational force, which emanated from the center of massive bodies and acted over a cosmic distance. In Einstein's theory of gravity, there is no gravitational force. There is only the local curvature of spacetime that massive bodies follow along in the straightest line that they can. Entities such as gravitational forces that were the main players in one theory are completely absent in another.

But their roles are still present. Something is still present that makes the planets move as they do.

In a sense, yes, I think I agree with that. Although a little nagging doubt thinks about Wheeler's book "Spacetime Physics" which says that the natural state of motion in GR is free-float, and it's deviations from the geodesic path that need to be explained. But I'll leave that issue aside for now and just say that I agree with you. For this is what I think structural realism is all about. Science gets at the truth in that we are discovering the true roles, the true structure of the universe. But we may, in individual scientific theories, be getting the players of those roles wrong.

So scientists are writing a play, and their first performance may have had a bad cast, but the screenplay is still great if we can only find the right actors to play the parts.

Something like that.

So back to your original point--how does this preclude critical experiments?

Well, to follow the play analogy, how can we ever really know if it's the actors that are bad, and not the script? If it's really hard to find an actor that does a role justice, is it a problem with the actors or with the role?

What do you think?

Well, I don't know. There may not even be any fact of the matter. But for now, I will assume that there is a fact of the matter and keep running with the metaphor. The director can decide to tweak the script to find a balance with the actor that he's got. Or, more radically, he may take that role out entirely.

But then you have to worry about whether its even the same play anymore.

That's true, depending on the role. If the play were Hamlet, and you took out the role of Hamlet, it seems clear that it's no longer the same play. But if the roles were Rosencrantz and Guildenstern? Then it's not so clear. Perhaps it's the centrality of the role removed that determines to what extent the play has changed. But you agree that it's not an on/off thing?

Well, I don't know. If you change the main character, for whom the play is named, that seems pretty on/off. What if you took King Lear out of King Lear? Godot out of Waiting for Godot? The silence out of "2:00 of silence"?

But on the other hand, if you change the other roles there seems to be varying amounts of grey area. What if you took Brutus out of Julius Ceasar? Or just the chorus?

Okay, so I'm having trouble seeing how this gets back to crucial experiments in structural realism. I got a little lost in the metaphor.

Fair enough. I can see how that might be the case. I am, after all addicted to metaphor, but that's a discussion for another day. If structural realism says that we have the roles correct, as expressed in the explanatory and mathematical structure, but we get the details wrong about ontological entity filling those roles, then I suppose a critical experiment would be one that puts to the test one of the roles in question. And if all we are ever changing in practice is the actors that fill the roles (gravitational force vs. curvature of spacetime) rather than the roles themselves, then we are never doing critical experiments.

Wow, that seems pretty weird.

So if we never put roles on the chopping block, then maybe that means we've had all the roles all along. Which makes me worry that we are just filling out theories constrained by the very structures of our brains.

Well, you are a worrier...I think that's a big leap.

How so?

Well, I'm not sure yet, but I'm trying to work that out. For one, we can't have had all the roles all along. There are new roles introduced by theories all the time.

How so?

Well, quarks, for instance. Higgs particles. Fields. We didn't even know atoms had a nucleus until the last 150 years, so how could the role of keeping the nucleus together been around before that?

That's a good point, but I'm not sure we're talking about the same level as being roles. We might be though. But maybe what I mean by roles is an explanans--quarks play the role of keeping together something that otherwise shouldn't be together. If we didn't think like charges repel then we might never have needed the role that quarks play.

What about astronomical observations? What role is played by black holes, quasars, things we can observationally detect that we wouldn't have dreamed of before?

Well, I'm not sure. Perhaps you make a good point. So maybe my worry was misplaced after all.

I'll give you something to worry about! What is this insubordinance?! Get back to work, you slimes…this, I command!

Serpentor! I should have smelled you coming.

Destro, the tricky details of your thesis have to be done by showing when roles were introduced and by showing to what extent they were really put on the line. But that is a task for another day, my friend.

Long live Cobra Commander!