There is Nothing Wrong with the Language of Support

For many years I was bothered by the idea that there was no such thing as verification or support in science. In this post, I claim it isn’t true and instead claim:

Popper’s main concern is not that there is no sense in which we can ‘verify’ theories as at all. Clearly, if we are talking about comparing two (or more) theories, we can certainly verify that one of them is the more correct. This is even true for Popper’s epistemology. What we cannot do is verify that there will never be a better theory. That is to say, we can never verify that a theory is one and the same as reality.

Grappling with this problem, I go even further in this post where I claim that Popper is at fault for why people misunderstand him as advocating for what I at the time called “naïve refutation” by which I meant refutation outside of a theory-to-theory comparison with a rival theory. (Only later did I find out that that term was already in use to mean belief in a refutation that was not tentative but absolutely certain.)

I realize now that my problem came from two beliefs that were not quite correct. The first was that refutation was more important than verification because verifications are absolute and refutations are tentative–what I now call the Absolute Verification Fallacy. (You can see hints of this fallacy in the first quote above.) The second was that I could not conceive of refutations being possible outside of a comparison with a rival theory.

Now I am not alone in holding these two positions. For example, Deutsch clearly advocates for the idea that a refutation is impossible outside of having a rival theory:

But in any case, the existence of a problem with a theory has little import besides, as I said, informing research programmes – unless both the new and the old explicanda are well explained by a rival theory. In that case the problem becomes grounds for considering the problematic theory tentatively refuted.

The Logic of Experimental Test, p. 10

And he makes it clear that refutations are always tentative but ‘support’ is entirely impossible:

The asymmetry between refutation (tentative) and support (non-existent) in scientific methodology is better understood in this way, by regarding theories as explanations…

The Logic of Experimental Tests, P. 8

But taken together, these two ideas form an obvious problem: there is just no reason to not call a ‘refutation’ (which only happens when there is a rival theory anyhow!) a form of ‘support’ or perhaps even ‘verification’ since you can just think of an experimental outcome that matches one theory and not the other as ‘supporting’ the theory it matched in comparison to it’s known alternative(s). You might even decide to think of a positive experiment as ‘verifying’ that one theory is more correct than another. [1] So why are we always emphasizing refutations over support/verifications? I asked around a lot of Popperians and never got an answer that satisfied me.

So I was relieved when I found out that Popper explains the asymmetry between falsification and verification. In a previous post, I at length quote Popper on his own reasoning for his preference for ‘refutations’ over ‘verification.’ This discovery was a major breakthrough for me in clearing up my misunderstandings of Popper’s epistemology.

To summarize Popper’s view: Universal statements are more powerful than existential statements and thus of more interest to scientists. Since universal statements in logic can only be refuted, it makes sense that scientists will be more interested in falsifications than verifications. (Though scientists are interested in verifications, particularly as verifications of an experiment’s failure to refute a theory, i.e. corroboration. See also this post for discussion.)

Where Verification Fits Into Critical Rationalism

So it is not that there is anything wrong with the language of verification or support!

There are many cases where the language of verification and support might be preferable. In past posts, I used the example of the Eddington expedition experiment and how it just as much ‘supported’ General Relativity as it ‘refuted’ Newtonian physics due to it being a crucial test between the two theories. In such a circumstance, it just doesn’t matter if we use the language of refutation or support. Indeed, I don’t think that calling such an outcome ‘support’ is any different than what Popper would call “corroboration” of the theory. This is because within a theory-to-theory comparison one might rightly think of us as ‘supporting’ the new theory as ‘more correct’ than the old theory rather than as us ‘refuting’ the old theory as less correct. So within a theory-to-theory comparison, support and refutation are logically identical and are merely the inverse of each other. So saying “this experiment supports this theory” makes perfect sense if we’re doing a crucial test between theories, so long as we remember we’re really only saying the experiment favored the theory that didn’t have a problem due to the experiment.

Moreover, there are specific instances within Popper’s epistemology that require the language of verification or support. For example, all experimental predictions take the form of singular existential statements [2] so they must be verified. (i.e. Eddington verifies that the stars during an eclipse really do move positions just like General Relativity predicts.) For those that think I’m abusing Popper here, here’s Popper himself on the subject:

…certain singular statements–which we may call ‘predictions’–are deduced form the theory [being tested]; …if the singular conclusions turn out to be acceptable, or verified, then the theory has, for the time being, passed its test: we have found no reason to discard it.

LoSD, p. 10 (Italicized emphasis is Popper’s.)

Likewise, Popper’s view of corroboration is a form of strengthening or supporting a theory. The above quote from Popper continues:

…It should be noted that a positive decision [in a test] can only temporarily support the theory, for subsequent negative decisons [in future tests] may always overthrow it. So long as theory withstands detailed and severe tests… we may say that… it is ‘corrobarated

LoSD, p. 10 (Italicized emphasis is Popper’s. Underlined is mine.)

For those still not convinced that Popper thought verification was entirely possible for singular statements, here is what he put in his index for LoSD on p. 513: “Verification of an existential statement, possible 10, 48-9&n…”

There is just nothing wrong with the language of verification and support in such situations and they are an integral part of Popper’s own epistemology. [2] So I kept wondering why there was such a huge concern over the language of verification and support.

Notes

[1] The post that follows this one will challenge the idea that a positive experiment really ‘verifies that one theory is more correct than the other.’ However, at a minimum, a positive experiment does ‘verify’ that one theory is more correct than another for that particular experiment.

[2] Oseroff defines something as intersubjective observable when it is an “expressions specifying the existence or nonexistence of entities at certain limited spatiotemporal regions, the occurrence of nonoccurence of specific events at space-time regions…” (Oseroff, “Addressing Three Popular Philosophic Myths about Karl Popper’s Demarcation Criteria”, p. 8) He goes on to quote W.H. Newton-Smith to connect this view with Popper’s concept of basic statements:

Such statements, which [Popper] calls basic statements, are characterized not epistemological but in terjs of their form and their role. The form of a basic statement is that of a singular existential statement where this means that they are existential assertions about some definite spatio-temporal region.

Oseroff, 3 Myths of Demarcation, p. 8

The key point here is that predictions are, in Popper’s view, singular statements that become basic statements by being intersubjective (i.e. regularities repeatable by anyone in the community.) But they are specifically existential statements (i.e. when you perform X test you’ll have the following happen.) And existential statements must be verified not refuted as per Popper’s own explanation of the logic of universal laws vs existential statements. So it is impossible to have Popper’s epistemology without the concept of verification when applied to testing.

This may seem counterintuitive at first because for Popper theories or explanations are about universal laws and thus are modeled in logic as universal statements and thus can only be refuted. But to test such a law you must use it to make a prediction. And predictions can only be modeled as singular existential statement which are always existential statements and can thus only be verified. (i.e. the stars on the Eddington expedition actually do move during an eclipse.)

There is no contradiction here. One is about the theory itself and one is about how to test the theory. But this might come as a bit of shock to people that understood Popper only in terms of the Absolute Verification Fallacy and thought verification was literally impossible because it was always about certainty. The way you verify the outcome of a test–tentatively–in Popper’s epistemology is by having a full community that can intersubjectively repeat the test. If that whole community agrees they can perform the test and it really is a regularity, then for Popper’s purposes we consider the basic statement (observation) ‘verified’ and we accept it as a real observation. Of course, it might later come to light that the whole community was wrong, so this does not violate fallibilism. This is an example of how Popper saw verification as also tentative just like refutation.

Leave a Reply

Your email address will not be published. Required fields are marked *