It’s said that a scientific theory can only be proven false, never true. This is basically so: for example, all the experiments in the world performed thus far can confirm (or rather, corroborate?) general relativity, but for one to show otherwise is all it takes for it to be proven false. But why is this so; why are scientific theories not like other facts, which can be confirmed as being either true or false? Take for example the fact that there is chocolate on my desk. That can easily be either proven or disproven, with certainty. And what if I postulated that fact as a scientific theory? Then could it no longer be proven true? No, because the principle of falsifiability is not on account of science decreeing that a theory cannot be confirmed true; it’s because of the nature of what a scientific theory is.
While physical principles are thought to be immutable “laws,” they’re actually the products of inductive reasoning. That is, from a few examples we infer a general rule. The postulate that there is chocolate on my desk only applies to my desk and right now–not everyone else’s desks, not every time the wind blows east–right now. General relativity applies to a whole class of phenomena. It makes predictions about what will happen in spaces we’ve never ventured toward, in specific scenarios we’ve never encountered. For every possible combination of values you can plug into the equations of general relativity, there’s a unique potential scenario that the theory could be being applied to. (Actually, for every combination of values you can plug into the equations, there are as many scenarios that results could be applied to as would render those particular values as measurements—measurements of, for example, velocities, positions, and masses. Everything else about a given scenario that could vary is irrelevant for the purposes of the calculations.)
How could we know for sure that the results of such equations hold true for every possible scenario that the equations could be applied to? All we have certain evidence for is that certain things behave in certain ways in certain situations—not even that, but that things once behaved that way under observation. We can’t prove that it will happen the same way under the same conditions in the future, let alone in the same general way under different conditions and maybe at another place or at another time.
So how would we make a theory certifiable? We could limit it to a statement about things having occurred that we actually observed happening, but that would make it useless as a theory. It would make it and its certification little more than a reiteration of the evidence that we had originally gathered, or most an insightful relationship applying only to a handful of particular scenarios, yet that evidence was presumably gathered for the purpose of intuiting more general rules.
We could limit it to a statement that something that happened in a specific way under the specific conditions observed would happen again in the same way under the same exact conditions in the future, but even that would be uncertifiable, as we cannot make inferences about the future with certainty (as show by Hume), as well as being nearly as useless as the previous formulation for the same reason: while the scope of extrapolation is somewhat wider, it’s still so small as to be not generally applicable, and the degree to which the scope of its application is wider is the degree to which its truth is uncertain.
All this considered, it seems odd that the laws of physics are supposed to be immutable, absolute, or all-encompassing. General relativity (which is the arbitrary example of an “any-theory” I’m using here), being inferred via inductive reasoning, is analytically on the same grounds as the statement that “all swans are white”. In other words, we have no idea when, why, how or under what conditions the principles of general relativity might spontaneously, or systematically, be infracted, though the theory has stood up relatively well to the tests of time thus far.
I think I’ve read that there are now contending theories to general relativity that predict certain phenomena with a little more accuracy. I think that if any infraction is to be found against the general theory of relativity, it’ll be one of subtle differences in measurements that are predictable and detectable under a determinable subset of possible conditions, as opposed to the infraction inexplicably applying only to the kitchen sink. Basically, it would be the same way in which classical mechanics was usurped by relativity. Why is it like that? I don’t know. Maybe the means by which we determine the meaning of a theory, as far they determine the theory’s scope of applicability and what constitutes or doesn’t constitute an infraction to the theory, somehow categorically exclude types of anomalies other than those characterized above from or comprising infractions to the theory. I don’t know. Probably not.
Some pertinent questions:
- Can a single anomaly constitute an infraction to a theory? Why or why not, or for which theories can it be so, and why?
- Can a class of anomalies, predictable in their anomality but unpredictable in their individual details, constitute an infraction to a theory? Why or why not, or for which theories can it be so, and why?
- What kinds of differences between predicted results and actual results indicate the effects of an interfering entity, rather than a falsification of the theory? How does this differ depending on the theory itself?
- What kinds of differences between predicted results and actual results indicate the effects of a class of interfering entities, rather than a falsification of the theory? How does this differ depending on the theory itself?
- What kinds of differences between predicted results and actual results indicate the interfering effects of a hitherto unknown principle, rather than an imperfection in the theory itself?
- What kinds of differences between predicted results and actual results indicate the interfering effects of a hitherto unknown principle, which could be added alongside all current theories, rather than an imperfection in one or more of the current theories themselves?
- When or how do we conclude that a theory, or a collection of theories, fully accounts for all universal phenomena?
And additionally…
- What avenues exist for restricting the scope of an induction (of the theoretic variety) to something less than all-encompassing but more than a mere restatement of the observed data?
- If such a gradient exists between the two extrema, by what means do we intuit the formula of, and the scope of, its upper bound?
- What benefits might we accrue by explicitly specifying, and/or consciously relegating, the scope of applicability of any given theory?
So, just a few things to consider. Because science without Philosophy of Science is like driving without a license, or perhaps knowledge without understanding. Or a theory without interpretation — or worse, a theory with a totally wrong interpretation..
