Another Look at Law and Theory

There’s a lot of philosophical discussion about what, precisely, constitutes a law or a theory in scientific practice. There’s also a lot of usage of the terms that has come to us over several centuries of not-quite-consistent application of terms.

What I’d like to offer here is not a scheme to try to make past usage fall out consistently; I think that is a task doomed to failure. Instead, I’d like to express a view of the various terms that makes sense to me and is how I have used the terms myself. Hopefully, others will also find it useful, even if only as a spur to discussion.

First, I’ll present a Venn diagram of how I broadly see things, and then add some comments.

I see a qualitative difference between facts and explanations. So the broad non-intersecting categories here are “Facts” on the left, and “Testable Explanations” on the right. I don’t want to go into the qualifier too much at this point, but essentially I see an explanation as delivering knowledge only if it can be tested and applies broadly, that is, inter-subjectively. That’s why I start with “Testable Explanations” rather than an unadorned “Explanations”. You can imagine a further inclusive superset of “Explanations” on the right if you wish, but I won’t spend any more time on that.

Within the realm of facts, one can notice consistent, persistent patterns or interrelationships that occur between facts or classes of facts. These are our “Laws”. One can test a law in the sense that one confirms that the proposed relationship is consistent and persistent. What one does not get from a law is an understanding of why the relationship occurs. Historically, people have received accolades for scientific work in discovering and publishing such patterns, though modern practice seems not to hold such work in good esteem.

When one turns to explanations, things quickly get more complex. “Hypothesis” as shorthand for “scientific hypothesis” is straightforward enough: it is a testable explanation of phenomena, whether or not the tests have occurred yet. But “Theory” is the difficult term to deal with, given that past usage has been so highly variable, confusing and conflating the term not only with hypothesis and the lay connotation of a “guess”, but also where “theory” has been applied to laws or law-like constructions.

It seems to me that when it comes to “theory” it makes more sense to try to make future usage better than to try to reformulate what has gone by in the past. We’d like the usage we settle upon to be broadly applicable to past usage. But we should not be afraid to simply abandon usage that cannot be made to fit a rational view of “theory”, though.

Part of what many people use to distinguish “theory” from “hypothesis” is the status of testing. For hypothesis, testing may not have happened yet. But for “theory” to apply, people generally want testing to have happened already. This seems clear enough to apply as a property of “theory”.

A more problematic property is extent. By that I mean that people will refer to a small-scale explanation as a hypothesis rather than as a theory. One runs into the heap paradox with this, since there is no bright-line rule for where explanations stop being “small” in extent, and thus should be referred to as hypotheses, and where they are “large”, and should be referred to as theories.

It seems to me that a better property to reserve for “theory” is that a theory should be productive, and by that I mean that by application of the theory, one should be able to generate further testable hypotheses.

So for myself, I use theory as a referent for a testable explanation that has been tested and is capable of generating further hypotheses (or has generated further hypotheses). That generally takes care of the problem of “extent” as well, since an explanation that is small in extent is less likely to be productive.

I think this sort of scheme is internally consistent and could be used in teaching, where it should minimize confusion for students. It explains why theories do not become laws (they are in separate categories of concepts) and why laws are more of a starting point for scientific inquiry than an end in themselves, since understanding why the relationships seen in laws happen requires explanation.

It is also why I’m not on board with the move to simply shift terminology around and refer to theories as laws. It seems to me that this is confusing and doesn’t help communication with the public. If we need to deploy different terminology, then make it really different. Where this whole issue of clarifying terms comes to application is in what to call evolutionary biology. At the level of some science organizations, there is a move afoot to simply refer to this as “evolutionary law” as a replacement for “evolutionary theory”, on the grounds (as I understand it) that the reality is closer to the public connotation for law than for the public connotation of theory (the “just a guess” thing). But it seems to me that this essentially is abandoning our responsibility to keep to accuracy if we simply capitulate to lay usage. Another possibility would be to use the insight that what distinguishes theory is that theories provide mechanisms by which things happen, and refer to things as “evolutionary mechanisms”. It at least doesn’t come laden with the baggage of past usage. But evolutionary biology incorporates knowledge that falls into the “facts” category and the “testable explanations” category (as noted by S.J. Gould), so I think a better alternative is simply to make it a broad term and refer to it as “evolutionary science”. This doesn’t permit the easy dismissal of “just a guess” and takes a step away from the whole law versus theory morass.