Danny Frederick on Refutation vs Rejection

A while back I had a long Facebook conversation with Danny Frederick, a well-known scholar of Popper. We kept talking past each other, so I offered to summarize his view to make sure I was understanding him. I made a summary and he changed a single word in it. I updated it and he accepted it. Here is what he accepted as a summary of his view on refutation vs rejection of theories: (see also this post for further discussion.)

Danny believes that there is an important distinction to be made between ‘refutation’ and ‘rejection.’ We will use the perihelion of Mercury as our example.

Danny calls any observation that doesn’t match a prediction as a ‘refutation’ or ‘falsification.’ In his mind, this refutes or falsifies the original theory (by which he actually means “a theory + background knowledge”) because the number of assumptions made before the experiment (or observation) was a finite list of assumptions due to the Ceteris-Paribus clause whereby we assume “there are no other significant forces acting in this situation.” Since this is always a requirement for any experiment, the remaining assumptions must always be finite and can be made explicit because they can only be the ones we already know about or which we wish to test. (In this case, 8 planets and the sun plus maybe some assumptions about how light refracts.)

The moment we obtain an observation that deviates from theory (e.g. the perihelion of Mercury) Danny claims we have now refuted or falsified “the theory.” However, it may be possible to save Newton’s theory proper by challenging any of our assumptions in the background knowledge – possibly even the Ceteris-Paribus clause itself. For example, we may conjecture that there is actually a ninth planet (e.g. Vulcan) and work out what size it must be and then work out how to observe it by experiment. If we fail to observe it by experiment, we’ll be forced to conjecture a new set of assumptions or even a whole new theory.

If we do fail to challenge the other [1] existing assumptions we may decide to challenge the Ceteris-Paribus clause itself. Danny agrees with me that absent the Ceteris-Paribus clause that “there is a potential infinity of theories.” However, to do this, we must conjecture a new assumption (e.g. let’s say that there is a body of dark matter exerting forces on Mercury that we can’t observe, other than its gravitational forces, due to the nature of dark matter.) But once we make this new conjecture, we then have a new set of finite assumptions and the Ceteris-Paribus clause again, in this new form, applies. So we always have, for any given experiment, a finite set of assumptions to work with when we do an experiment.

If all such attempts to challenge the assumptions fail, we may need to challenge Newton’s theory directly. However, to ultimately ‘reject’ Newton’s theory would require an alternative theory (say General Relativity). Until this new theory exists, it is not truly possible to ‘reject’ Newton’s theory. But once the new theory is conjectured, we can then ‘reject’ Newton’s theory and replace it with General Relativity. This is what Danny refers to as ‘rejection’ of the theory and he agrees with me “All rejections are within theory-to-theory comparisons.”

Therefore Danny see ‘Refutation’ or ‘Falsification’ as what happens the moment we have an observation that doesn’t match what ‘the theory’ (when combined with our finite set of assumptions we’re currently testing) predicted and ‘Rejection’ as what happens once we have a new theory to replace the old one.

Danny replied: “I accept your summary of my view, though I would word it differently in places.”

Danny made it clear that he feels this is what Popper himself meant by ‘refutation’ but that Danny himself coined the term ‘rejection’ for when we actually drop the theory.

The reason why this is significant is because David Deutsch uses the term “refutation” for what Danny calls “rejection” and uses the term “problem” for what Danny calls “refutation.” Thus we really have two different understandings of what a ‘refutation’ is. If Danny is correct that Popper understood ‘refutation’ the way Danny does [2] then Deutsch accidentally codified a common misunderstanding of Popper’s concept of ‘refutation’.


[1] Adding the word “other” is the only adjustment Danny had me make before accepting my summary of his view. Restating someone’s view is a technique I’ve tried a few times. I’ve never once had a person accept my summary on the first try. They always insist on making at least one small change like this.

[2] Nathan Oseroff in “Addressing Three Popular Philosophic Myths about Karl Popper’s Demarcation Criteria” covers this in that paper. This is an excellent paper that every Popperian must-read. He produces strong evidence straight from Popper’s own writings that Danny Fredrick is correct: Popper understood “refutation” as a basic statement at odds with the combination of the theory plus the background knowledge. We have been misusing/misunderstanding the term “refutation” this whole time.

4 Replies to “Danny Frederick on Refutation vs Rejection”

  1. My understanding, from Popper, is that no falsification is ever definitive, meaning that an inconsistency between a basic statement and the combination of a theory plus background knowledge is hard (or impossible) to prove. To say as Frederick does that a theory is falsified “WHEN it is inconsistent with an accepted basic statement that describes a reproducible situation” abbreviates, it seems to me, “a theory can be REGARDED as falsified ……”.

    Frederick’s definition of what is (to be regarded as) an instance of falsification/refutation ought rather to be elucidating a practical convention. It should not be regarded as a definition of a true or actual falsification/refutation. For all his avoidance of truth as an aim of science, Frederick is constricting refutation to his view of what constitutes true refutations.

    Of course we do say in ordinary discourse and even in scientific discourse that many a theory has been falsified or refuted. This, usually innocuously, abbreviates “in our considered opinion about the results of testing (which includes reproducibility as a consideration) the theory has been discounted by the evidence”.

    1. You make quite a few spot-on points here and, as always, show a strong command of Popper. You’re like my number one goto resource on getting perfect Popper quotes or clarifications of Popper’s meaning.

      However, in Danny’s defense, I think you are really saying the same thing he is but in different words. (Also, keep in mind this is my summary of his view. Yes, he bought off on it, but only as ‘good enough for this discussion.’)

      For example, you say: “WHEN it is inconsistent with an accepted basic statement that describes a reproducible situation” abbreviates, it seems to me, “a theory can be REGARDED as falsified”

      Now certainly this is true when we are talking about ‘the theory proper.’ When we have a ‘refuting’ case we may (or may not) *regard* the theory we’re exploring as ‘refuted.’ And, yes, as per Popper, it’s not absolutely certain and we may need to possibly even revisit if the basic statement in question was even correct. Danny would (from my conversations with him) definitely agree with all of that.

      However, his point (or at least the contextual one he was making to me at the time) is that a ‘refutation’ (or perhaps we should instead call it a ‘counter example’ instead as this is more accurate of a term) is always of the theory PLUS the background knowledge. So at any point in time, it’s a logical fact that a basic statement does or does not represent a contradiction to the theory PLUS the background knowledge. This is the “objective fact”–or rather inter-subjective fact. (What I really mean is that is the observation that needs to be explained. It’s the problem to be solved.)

      Let’s use the example of Adam et al. 2012 finding neutrinos going faster than the speed of light. This turned out to be a problem with the optical cable. So it was not ‘refuting’ General Relativity after all (the theory proper). In fact, it would be accurate to say that the basic statement itself was wrong.

      But this is exactly Danny’s point! The theory PLUS the background knowledge includes a theory about how optical cables work and about if they are set up correctly, etc. It’s the combined theoretical system that is ‘refuted.’ So having a temporary basic statement that neutrinos were traveling faster than the speed of light *was* an example of refuting General Relativity PLUS the background knowledge. It just happens to be in this case that the theory that ended up being refuted wasn’t General Relativity but was instead the theory about the optical cable being hooked up correctly. But the entire *combined* theoretical system *was* ‘refuted’.

      As Popper put this:

      No single hypothesis, it may be said, is falsifiable, because every refutation of a conclusion may hit any single premise of the set of all premises used in deriving the refuted conclusion. The falsity to some particular hypothesis that belongs to this set of premises is therefore risky… …we can indeed falsify only systems of theories and that any attribution of falsity to any particular statement with in such a system is always highly uncertain…

      (See Realism and the Aim of Science p. 187. Emphasis Popper’s)

      So Danny will always be correct (when understood this way.) We don’t have to ‘regard’ the combination of GR plus the background knowledge as ‘refuted’ it IS ‘refuted’ in this sense of the word ‘refuted’. We just don’t know which part of the theoretical system we’re testing is the problem yet. It might be nothing more than a problem with our instrument (i.e. our theory about our instrument working when really it isn’t.) This is why I think it would be more accurate to call it a counterexample rather than a refutation.

      Now imagine that we checked the optical cables and found nothing wrong. Then we got every other lab to repeat the experiment. At some point, we’d probably decide that it wasn’t a problem with our theory about how the optical cables were (or were not) functioning. We’d then have to start to explore other ideas — perhaps even if we need to tweak or replace General Relativity.

      So we never really ‘refute’ anything in the sense of absolutely certain refutation. What we really do is we take counterexamples and come up with conjectures to explain them. That might be a problem with the optical cable or it might be a problem with General Relativity.

      1. I think Danny Frederick, in our conversations about refutation, scratched his head also about what I was driving at. In revisiting I can see that I was less clear than I ought to have been. My agenda, at heart, was to orient the discussion to problems of testing rather than a drift into problems of definitions such as the definition of the term “refutation” or what formally constitutes “a” falsification.

        “In this respect, the situation is the same as in any other case in which we have to think out a new theory: the decision to ascribe the refutation of a theory to any particular part of it amounts, indeed, to the adoption of a hypothesis; and the risk involved is precisely the same. To meet it, we need ingenuity, daring – and some luck.

        Thus there is no routine procedure, no automatic mechanism, for solving the problem of attributing the falsification to any particular part of a system of theories – just as there is no routine procedure for designing new theories. The fact that not all is logic in our never-ending search for truth is, however, no reason why we should not use logic to throw as much light on this search as we can, by pointing out both where our arguments break down and how far they reach. The fundamental logical asymmetry which I have described can certainly throw some light on this question.

        [*All this is about empirical falsification and its uncertainties. It must be distinguished from the purely logical criterion of falsifiability; that is, the existence (not the truth) of potential falsifiers of a theory. There are no similar difficulties connected with falsifiability. Falsifiability is untouched by the problems that may affect empirical falsifications.]”

        Page 189 “Realism and the Aim of Science”

        1. I love that quote! (Just used it in an upcoming post.) Yes, practically speaking, we might think of a ‘refutation’ as both a counterexample and a tentative conjecture as to what part of the theoretical system has the problem. Then we have to go test that new conjecture.

          You said: ” My agenda, at heart, was to orient the discussion to problems of testing rather than a drift into problems of definitions such as the definition of the term “refutation” or what formally constitutes “a” falsification.”

          So that’s probably true for your conversation with Danny. And I know Danny had a bad habit of dwelling on definitions a bit too much and that this sometimes caused him to not understand what you were saying. (The reason I restated his view was because he just couldn’t see that he and Deutsch were saying the same thing but in different words.)

          But I was trying to understand what Popper meant by ‘refutation’. When I read Popper up to that point I was accidentally reading into Popper a view of ‘refutation’ closer to the way Deutsch defined it as requiring two theories. And that was problematic. There are many passages in Popper that don’t make sense if you are trying to read the concept of ‘refutation’ (or ‘falisification’) as only happening when you have a second theory to appeal to.

          Danny’s explanation really helped me understand what Popper was getting at.

          For what it’s worth–and feel free to disagree with me here–I don’t believe the word ‘refutation’ or ‘falsification’ are good words to describe Popper’s true intent. Those are terms that come with a lot of baggage. I think much of the misunderstanding of Popper that exists is because people just can’t conceive of a ‘refutation’ or ‘falsification’ as idiosyncratically being defined as against the theory plus the background knowledge and thus requiring a tentative conjecture as to which part of of the theoretical system has the problem. I think those words, to most people, mean “you now know this theory is wrong” which just isn’t what Popper meant at all. So I think we’d benefit from dropping the use of the terms ‘refutation’ in favor of calling them ‘counterexamples’ or ‘problems’ or something like that.

          But at the end of the day, it’s just a choice of words, and ‘words do not matter.’ Except that they do if they summon the wrong ideas to the heads of most people. So it’s a bit of a tightrope. We do need to talk about the meaning of words to make sense of what someone else said. We just shouldn’t argue about what “THE” meaning of a word is. That’s a waste of time because words never have “THE” meaning.

Leave a Reply

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