Thursday, July 06, 2006

Is science a sandwich?

I got into an argument recently about whether, if Einstein hadn't described the theory of relativity when he did, someone else would have inevitably done so at some later point. The claim had been made that "what Einstein did was find something that was already there. Like a sandwich. Now sure, Relativity is a really well hidden sandwich, but if it is true, and it could be found, it would have been eventually." The discussion got pretty repetitive after a while, with the other party just sort of reiterating his point (which I'd already established I didn't agree with!) instead of arguing it, but since it taps into some of my most strongly-held philosophies about science, I thought it was worth giving the digested version of my points here. I also realized that I'd written a paper back in undergrad (!) that dealt with some of what I was trying to say, so I've cribbed liberally from my own work in this post.

For starters, I think it’s important to debunk the idea that science consists of discovering things that are already there, as though the Actual Rules of the Universe were just out there waiting for someone clear-sighted enough to perceive them. Don’t worry, I’m not going all Parmenides; there may be something out there, and our science may closely approximate it. But in no case can we interact with truth directly without using science as an interpretive tool. In effect, truth is a theoretical entity, something it seems reasonable to believe in but with which we cannot interact. It is more useful, then, to talk about the verifiable qualities of a theory –- how well it predicts, how well it explains –- than to question whether it "describes reality." If science amounted to drawing a complete map of a well-defined territory, then we could critique a scientific theory on the basis of how well it accords with reality. But to launch such a critique, we would need access to the territory. Instead, we must (and do) treat science as our best means of navigating a territory we cannot see. And it’s foolish to imagine that there’s only one way to do this, or any significant difference between the best version of one method and the best version of another. Very good echolocation versus very good feelers? The distinction is minimal.

To imagine that all scientific discoveries thus far would have inevitably been made, changing only the order, assumes that there is a way that science must look. This assumption isn't justified if you view science (as I do) as "the best explanations we can manage" rather than "an explanatory system that is coextensive with reality." Of course we always try to get theories that have more predictive value, but we can do that in a number of ways; there's no "true" version of science. And since a lot of discoveries are predicated on others, changing the timing could change a great number of things. For instance, we might have discovered relativity without Einstein, but we might just as easily have come up with a theory that worked equally well but wasn’t exactly the same. In fact, given that the scientific context would be so different -– imagine getting through the early 20th century without relativity, if you can cope with counterfactuals! Imagine, just for starters, the state of technology! –- it’s reasonable to assume that this other theory wouldn’t look exactly the same. It would give us another method of navigating the unseen territory. The only way we could imagine that 1950s relativity would look exactly like 1910s relativity would be if we assume that science is purely a process of discovering things that are Really Out There, of making our map look more like some verifiably existent landscape.

Saying that "science is developing towards an understanding of the world" is like saying “the human race is evolving towards an organism that is well-adapted to its environment." It is understood that the evolution is not complete, but also that there is no one thing that "complete" would mean, no one thing that this well-adapted organism must look like. There is no reason to imagine that the current state of the species represents the only way we could have adapted to the environment on Earth. Why must the smelling mechanism be on the face? Why round pupils, and not oval (like a cat) or rectangular (like a goat)? What’s so maladaptive about a tail, or feather-like plumes on the head? They’re not as harmful as an appendix, that’s for sure, and yet we’ve got one of those. The human race could have had any number of different traits without significant harm, and there are any number of ways in which we could have achieved the useful traits we have today (for instance, it’s important to have jointed limbs, but why must the joints only go one way? Why one ball-and-socket and one hinge per limb?). We are adapted to our environment because we evolved this way, but we didn’t evolve this way because it was the only way we could have ended up adapted to our environment. To say so is effectively to say that there is a blueprint for an effective human race (not necessarily that there is a designer, mind you, but that there is only one good design). Either evolution is teleological, or it could have happened differently.

It’s easy to look at some of the hypotheticals I’ve been posing and to say, for instance, "well, pretty much all mammals have one ball-and-socket and one hinge joint per limb." This is important. Because we evolved from other mammals, we retained certain characteristics of those animals that were useful or at least not maladaptive. A change somewhere far back in the game -– maybe Tiktaalik grew an extra leg, for no really good reason except that there was no reason not to -– might have altered all subsequent development. Make it two extra legs, actually, because symmetry is useful. Natural selection would then select among six-legged forms, not four-legged ones, and subsequent developments would be based on that model. It would be pretty far-fetched to imagine that you could somehow get a bipedal human out of that! And yet we would have evolved based on exactly the same processes.

The same is true of science; a change in scientific history might have had a profound effect on our current theories, but it would not mean that we didn’t end up with the most useful theories possible based on the history we had. That’s how science works: as Bas Van Fraasen said, "Do not ask why the mouse runs from its enemy. Species which did not cope with their natural enemies no longer exist. That is why there are only ones who do." Only scientific theories with the best predictive and explanatory power, based on what we know, can survive – but that doesn’t mean that the theories we subscribe to now are the only useful theories we could have developed, any more than the evolutionary need to flee predators means that a mouse couldn’t have had pointy ears.

Viewing science as an evolutionary process makes it clear that we cannot logically conceive of science as aiming for truth, any more that we can consider an individual species to be aiming for a completely evolved form. More importantly for the current debate, we cannot assume that the current state of either science or species represents an inevitable conclusion. The goal is to be adapted and to be adaptable, not to accord with an inaccessible Truth or an invisible Design. There is more than one way to do this, and any change in path would have altered background and context so much that we might end up with something completely alien to what we know now (but, importantly, still about equal in adaptiveness/usefulness!). To claim otherwise, to claim that any scientific advance that’s been made so far would have been made eventually by someone regardless of altered time or background, is to claim that there is only one thing that science can be.

5 Comments:

Bill Hooker said...

Viewing science as an evolutionary process makes it clear that we cannot logically conceive of science as aiming for truth

So are Thomas Kuhn &co. arguing for punctuated equilibrium? That doesn't seem like a bad analogy, come to think of it.

7/06/2006 4:27 PM  
jess said...

Yeah, I think punctuated equilibrium is a pretty good analogy for the classic Kuhnian model, although if I recall correctly he said himself that things didn't happen nearly so neatly most of the time.

7/06/2006 5:00 PM  
Nick said...

I'm not quite sure how much I buy the evolution analogy. I certainly agree with the idea that we never get access to the "truth", even if it is out there. The question is to what degree is this different than saying that you never get access to your desk chair, only sensory perceptions that can be interpreted together as a desk chair? Of course, you certainly can be an extreme skeptic and doubt the existence of the chair, or you can say, "Look, I have so much experience about this at this point that my idea of it is pretty set in stone. I can say this is effectively true."

The point of all this prattling on about chairs is that as you gain more data about something it seems that your ideas about it change less. If the thing you have an idea of actually exists, then each new datum beings you slightly closer to that truth (at least on average). If two different people (who are not in communication) get data about the same thing, initially when the data are sketchy they may form radically different ideas about it. As they get more and more data, won't their ideas be drawn closer and closer together? In the lame chair analogy, if two people see it at a distance they might disagree about what they see, but as they approach won't their ideas converge? Even if you don't presuppose the existence of any "truth" out there to be found, if ideas about the things we perceive (or measure, in science) converge, then perhaps we may define that limit of convergence to be what we call truth. Sort of like how we define the real numbers in terms of convergent sequences of rational numbers (well, actually equivalence classes of Cauchy sequences).

Hmm...perhaps that last analogy is not so helpful, but it's the best one that came to mind.

In Evolution there will certainly be many different sorts of life forms given the same selection pressures. One may be able to explain this as being due to the fact that a) the selection pressures change on a time scale such that natural selection has time to find good adaptations but not the best adaptations before the requirements change again, and b) what adaptation is best will depend on what other creatures are there, since organisms are in competition. In point a), I mean that certainly if a population of giraffes moves to a place where the trees are taller it may result in giraffes with longer necks, but given enough time (100 million or 1 billion years) one might suppose you'd find their descendants as some sort of tree climbing creatures instead. It's just that normally selection pressures don't stay the same that long. Then, considering point b), if there are already tree climbing primates there, it's less advantageous to develop tree climbing abilities since there's already competition there. My point is that there may be features in evolution that seem to lead to multiple solutions to the same "problem", but analogous features may not exist in the case of science.

Of course, putting all this talk aside, we could ask the empirical question of whether two populations independently pursuing science will in fact converge on the same sort of ideas. A lot of the answer will probably depend on what you consider to be "the same". Clearly you have research communities different places developing differing ideas during the history of science, but one may argue this is only a short-lived state that persists while they have but little data about a phenomenon on which they disagree. Also, they are in communication. You'd really like to find an instance where two scientific communities are out of communication for a very long time and develop extensive and useful scientific theories. Then you could look to see how similar they are. Personally, I'm not aware of any such historical incident to look at. The closest thing I can think of is Srinivasa Ramanujan, a genius living in India who, working in isolation and using a textbook about 50 years out of date, developed many of the same mathematical results as the western mathematics establishment. Of course, he'd developed much of his own terminology and notation, so his results were nearly indecipherable at first. However, math and science aren't quite the same game.

On the specific subject of Einstein and relativity, if you're talking about Special Relativity it seems plausible that it would have eventually been developed in much the same way, given the related work of people like Lorentz and Poincare. If you're talking about General Relativity, then it's far less clear. The history of science isn't really my thing, but I don't think that many people were thinking in the same direction as Einstein on that topic, and it seems like enough of a leap that it's not too difficult to believe that some other theory might have been formulated instead.

7/08/2006 11:39 AM  
jess said...

I don't think I can allow the chair analogy, Nick... I know it's very popular to use chairs as the primary unit of any philosophical discussion, but we're not talking about some semantic quibble where I say "well you can see the light bouncing OFF the chair but not the chair itself." This is true but it's hardly parsimonious (or useful!) to say so. However, we're talking about something a lot bigger and more complex than a chair, where we're not only trying to perceive it but to understand it. Even keeping it to chair level, this is more like trying to figure out what the chair is for. (Still not a great analogy, because the Rules of the Universe aren't for anything, but we're definitely not just trying to figure out what's there -- we want to know how it behaves.) Sure, if you tested a hypothesis that a chair was for sitting in, it would do pretty well. But, speaking from experience, so would the hypothesis that a chair is for moving a computer across the street, assuming it's one of those rolly office chairs. And if you're committed to the idea that it's a sort of glorified dolly, are you going to try sitting in it? If you do, are you going to conclude that that's its main purpose? I mean, maybe. But also maybe not.

Now imagine it's not a chair, it's the Chrysler Building.

Even if you don't presuppose the existence of any "truth" out there to be found, if ideas about the things we perceive (or measure, in science) converge, then perhaps we may define that limit of convergence to be what we call truth.

I'm not sure how this is different from what I'm saying! My point was just that it doesn't matter whether these ideas converge on something that from some nonexistent god's-eye perspective could be called Objective Truth. That's a non-useful concept, and it's as impractical to worry about whether our agreed-upon understanding is "Truth" as it is to constantly correct people about whether they can see a chair or the light bouncing off a chair.

And I guess the other difference is that I don't see any reason why there has to be one potential point of convergence. Yes, people's ideas are drawn closer together as they get more data, but as you pointed out, they're in communication. This is part of my overall point; the science that's been done affects how we interpret the science that's being done. Again, not deflationary! Our current understandings describe the world in a very accurate way, otherwise they'd have fallen by the wayside long ago, like the mouse that doesn't run. I just think it's disingenuous to pretend that this means we have some kind of hard line on the Actual Truth of the Universe -- it just doesn't gain us anything to think that way -- and I think that's the only way we could justify a conclusion that our data could only have converged on Science As We Know It.

What we have is a perfectly suitable (although necessarily imperfect) scientific epistemology. I'm not knocking it. I like it. It works better, by definition, than anything we've ever tried. I just don't see a compelling reason to conclude that it works better than anything we've ever NOT tried.

I'm not totally sure where the part about selection pressures is going... I think science is under selection pressure at least as often as any organism, considering that every new experiment, every new piece of data, every technological advance potentially requires some reevaluation of what we know.

7/08/2006 6:36 PM  
jess said...

Er, I said "again, not deflationary" when apparently I had taken out my earlier bit about how I don't consider this way of viewing science to be deflationary in any way. It bears repeating, especially when some of my best interlocutors (like Nick) are scientists. Of course I suspect you know this, but let me be very clear that I am NOT taking part in some wingnut-style condemnation of science as some sort of peer review conspiracy. I am a staunch believer in the scientific method and I think we are constantly moving towards more functional versions of how we understand the universe. But part of what makes the scientific method great is that it involves always assuming that you could be wrong, or more accurately, always assuming that there could be better (or just different) explanations. And I feel like there should be no reason to disclude our entire paradigm, and even previous ones.

I should also point out that I'm not saying that there is another way to satisfyingly interpret the kind of data we've got. Fuck, how would I know? I've never lived in a different scientific paradigm. It also doesn't matter from a scientific perspective, I think. But it matters from a historical perspective, when people start arguing that all great ideas in the history of science would eventually have been come upon by somebody, pretty much as-is.

7/08/2006 6:45 PM  

Post a Comment

<< Home