January 26, 2005

And Prospero broke his soap box

Posted by Curt at 04:26 PM in Science | TrackBack

I may have bored everyone to death about this topic, but I have my last exam tomorrow, so here is my final thought about what distinguishes science. Most of the descriptions of science that I know of don’t really explain how science progresses without falling into a quaint mythology about approaching some metaphysical truth. Kuhn doesn’t, Popper doesn’t, Pierre Duhem doesn’t, and I myself have neglected to account for it to some extent.

I think the key is that science, at least experimental science, is essentially concerned with predicting the future. Every hypothesis, in essence, is a prediction about the future. What distinguishes science from other forms of prediction is the emphasis on verification, the insistence on framing predictions in such a way that when they are tested they can be decisively answered positively or negatively. In other, the goal is not to not be wrong but to achieve a definitive positive answer. Even a definite negative answer is preferable to none at all.

Some philosophers, like Duhem, claim that individual hypotheses can neither be verified nor falsified, because a whole body of theories and assumptions lies behind, and is implicated in, every hypothesis, and thus one can never be sure just what has been validated or failed. While that’s true, it is also nonetheless true that when the result of an experiment does not match a hypothetical prediction the hypothesis has been proven invalid as it stands. In other words, no matter what went wrong, the body of theories and assumptions that led to the hypothesis do not work as they now stand. Thus, things will have to be changed until they produce accurate predictions. Conversely, if a hypotheis is corroborated with a positive answer, the theories behind it stand validated until a hypothesis receives a negative answer.

In other words, experimentation does not serve to lead by induction to new theories, but rather theories serve to make possible specific predictions about the future which can be verified decisively. This at least is the goal. The goal is not a description which is true or corresponds to the truth, or at least that is not the immediate goal. When the facts or events are given, anyone can interpret them, and the fact that these events are known can mask the relative merits of the theory which interprets them. The idea that theories are validated by their correspondence to experimental results is tautological: the first condition of any theory is that it accounts for the experimental results that gave rise to it. But the only way to determine whether it is simply a theory to fit the facts or whether it is truly generalizable is to test it against unknown facts via prediction. Of course, predictions are almost always only approximately true, so the specific point of acceptability is not provided for by the general concept, but, at least in theory, decisive verification of predictions provides a simple, clear, and immensely useful criterion by which to evaluate theories. In my opinion, this explains much of the evolutionary capacity of science (I mean evolution in the more contemporary sense of diversification and selection rather than the old idea of teleological perfectibility).

If experimental prediction is the mark of science, this leaves the question of whether purely descriptive disciplines like zoology and areas like quantum theory where predictions are inherently statistical and ambiguous are scientific. Zoology and the like I think are, because hypothetical prediction inherently implies classification. In other words, by saying “under these conditions, such an event will happen,” one classifies, in other words sets parameters. The goal of zoology seems to be not simply to describe members of a group but to describe all the characteristics which define the group, set the parameters of the group, which is the first step towards making predictions about the group. So it is an element of science, but incomplete. As for quantum, I avow my profound ignorance of it, so let my opinion be taken in that light. As far as I understand, the stastical laws in that realm allow predictions in aggregate, so I am inclined to view it as still within the domain of science, at least in spirit, but of course the lack of decisiveness of statistical predictions gravely weakens the predictive power of science in this area, and I have already suggested that the rise of relativity and quantum in my view are intimately tied to the waning of the scientific age. Finally, it should be noted that while making correct predictions is the goal of science, that should be qualified by saying that the predictions are intended to answer general questions concerning the nature of things and establish specific knowledge. Optics or engineering, for example, are not science, although they once were, because all the major questions have been answered, and they no longer concern gaining further knowledge of the future and the universe, but rather in applying that knowledge to constructing specific objects.

So the goal and value of science is in predicting, and thus establishing knowledge of, the future, and the scientific method is the means of arriving at correct predictions. This is not to discard my earlier contention about the ideological basis of science, because the efficacy of prediction is based on the relative value of induction, and successful induction relies on the essential regularity and stability of the universe. In other words, in order to draw a general theory from a specific experimental result and vice versa, the universe must be considered as basically the same everywhere and at every time, which in turn implies that it be material, matter being defined as that which cannot change itself and is therefore static. It seems to me that if in quantum theory, for example, phenomena become genuinely dependent on the observeer in ways that are neither generalizable nor predicatable, it cannot continue to remain truly a science. It would seem to me that the branches of physics which are entirely theoretical are for practical purposes basically metaphysics.

This model depends on a linear notion of time. It might seem the opposite, that if the physical laws are eternal and universal time is actually opposed to this insofar as it represents dynamism, change. But in reality the sameness of the universe upon which science is predicated is not a a sameness at any particular moment, but rather a sameness of behavior. In other words, a view of the universe from a materialist perspective at any given moment shows that everything in the univese is different in the sense of being distinct. However, the idea is that under the same conditions all matter (or whatever you call the fundamental substances) will act in the same way. Without the steady march of time, this unity of behavior disappears, and there are simply a million disparate entities. Thus, space (and time) as properties of the universe are essential to science.

As for what the value of science is, I’m afraid I can’t generalize about that. From reading my recent posts one can most likely guess at my views, but I will simply say that one’s view of the efficacy of science in making the universe understandable will probably depend on entirely on whether one a) believes that linear time is a real property of the universe and b) if so, whether true induction is possible.

p.s. I should note that Henri Poincaré anticipates me in seeing the epistemological value of science as consisting mainly of its ability to make predictions rather than its descriptive correspondence to reality. However, he also thinks that theories are conventions and definitions of concepts, not true descriptions of physical phenomena based necessarily on experimental results. He thinks the conjunction of these two make theories relatively independent of their experimental bases, which he regards as a good thing because it creates a body of stable principles in which we can trust. I think that that is neither true nor a good value. The emphasis is on predicting correctly, not creating stable beliefs (if you want unchanging beliefs, what not join the Church?), and if generating true predictions is the goal, theories should be more rather than less sensitive to their experimental roots.

p.p.s. Since my exam was about scientific laws and causality, I should add that while scientific activity depends on a belief in time, not all scientific theories do: the law of conservation of energy, for example, I believe is essentially atemporal.

Comments

There is too much here to comment on all of it but I think science is doing fine at producing much fascinating information. I think the data are exploding so fast that it becomes atomized into numerous areas that nobody can keep up with. Optics, for example may not be advancing, but new techniques of microscopy, using immunological probes and DNA technology combined with computerized analysis may uncover amazing information. The thing that is driving advances is powerful new technological tools open up new ways of exploring nature. I don’t think these new ideas are primarily culturally determined, although some questions can’t be explored for political or ethical reasons. Some new approaches settle old debates and others destroy old theories. For example, I think some of the new info from the Hubble Telescope caused serious shakeups in theoretical physics. (Something I know nothing about.) Much information about the nature of cancer is being discovered and it is weird, unexpected and hard to understand.
I don’t really understand much about the philosophers of science but they seem hung up on quantum theory and relativity which killed the idea of a clockwork world and threw them into a state of cognitive dissonance. Armchair philosophy isn’t where the action is. It’s occurring due to the application of new and interlinked technology which continually expands and modifies knowledge. Relativity and the quantum world don’t really need to be invoked in most things just yet. We still have a long way to go before that.

Posted by: Dave at January 26, 2005 11:13 PM

After spending the last 16 hours doing math, I'm in no shape to comment intelligently, but a couple quick points:

It seems to me that if in quantum theory, for example, phenomena become genuinely dependent on the observeer in ways that are neither generalizable nor predicatable, it cannot continue to remain truly a science.

That's true, but, of course, the entire raison d'etre of quantum mechanics (at least, insofar as it has one) is to provide descriptions (or predictions, if you prefer) that are generalized up to the limits of our perception and the constraints of the universe. This means statistical predictions rather than "hard" predictions, but there's much less ambiguity than the popular press likes to pretend. In fact, some might argue (and have) that the seemingly ambiguous or even contradictory facets of quantum mechanics arise from the fact that the emphasis is on predictive models (of which different ones "work" in different, but overlapping, scenarios) rather than on theorizing about what is "really" going on (which is, ultimately, metaphysics, at least to the extent that such theories have no basis in experiment and offer no better predictions than the makeshift models).

However, he also thinks that theories are conventions and definitions of concepts, not true descriptions of physical phenomena based necessarily on experimental results. He thinks the conjunction of these two make theories relatively independent of their experimental bases, which he regards as a good thing because it creates a body of stable principles in which we can trust.

As I read it, this opinion reveals Poincaré's fundamentally mathematical perspective. Take that how you will.

Posted by: shonk at January 27, 2005 01:44 AM

Curt,
Can you elaborate on your problems with Popper and Kuhn?
Jonathan

Posted by: Jonathan Wilde at January 27, 2005 11:36 AM

"Armchair philosophy isn’t where the action is. It’s occurring due to the application of new and interlinked technology which continually expands and modifies knowledge. Relativity and the quantum world don’t really need to be invoked in most things just yet."

That may well be true, but I'm not convinced that "the application of...technology" really constitutes science, at least not intrinsically. It seems to me that science is more about generating the theoretical principles which underlie technological applications. You may be right that there's not much use for the leading edges of theory right now in practical applications, but then it seems to me that the "cognitive dissonance" between theory and practice is more fundamental than just that of the philosophers.

Posted by: Curt at January 27, 2005 12:38 PM

Can you elaborate on your problems with Popper and Kuhn?

Well, I've posted on it here, here, and here. In essence, I think that both describe elements of the scientific process at any given moment quite well: Kuhn describes very nicely how acceptance of dominant models and principles affects scientific activity, and Popper how theories are subjected to scrutiny. Neither one, however, in my opinion adequately addresses how theories change, in other words how science progresses. Kuhn's description is too sui generis: some theory is accepted provisionally by everyone in the scientific community, even when it pretty obviously stands in need of revision, and then voilà! some lonely genius thinks up a more useful theory, which gradually becomes accepted. The flash of inspiration may be real, but, assuming it's not an entirely passive process, a better explanation would make that scientist less of a pure anomaly, would locate some reason why he was inclined and capable of breaking with a conception which by the nature of the scientific community seems to require universal acceptance. I'm aware that Kuhn distinguishes "normal science" with science in periods of crisis, but that distinction in itself seems pretty anomalous and discontinuous, and doesn't explain how it is that anomalies are discovered and cause science to go into crisis mode in periods of "normal science," when scientists are supposedly mainly concerned with protecting the theories under which they are operating.

Popper, on the other hand, doesn't even bother accounting for the persistence of attempts at explanation. He apparently seems to think ideas and theories just spring up on their own, or at least he doesn't seem to really care whether they come from or how they are generated--the importance lies entirely with how they are verified. But of course, now matter how flawless a system of verification, if all the ideas and theories with which it is presented are crap, no progress (by which I mean change, not necessarily improvement) can be made. So the ability of science to generate new vast, encompassing ideas which are verified seems to deserve, if not require, an explanation. Of course you might say that nothing can control for the quality of ideas except the verification itself. Which is true, but my idea that scientific hypotheses are oriented towards making true predictions, which are then generalized if validated, provides a focus and a direction for the generation of ideas that Popper never really bothers with. Similarly, Kuhn may be right about the homogenizing effect of paradigms, but if the acceptance of them is not limited in some way, if scientific activity is not oriented towards finding results that may conflict with a theory, then discontinuities in the history of science, "scientific revolutions" as Kuhn calls them, are pretty much inexplicable.

Posted by: Curt at January 28, 2005 05:12 AM