Hell hath no fury like a Popperian scorned

My post on falsificationism turned out to be a lot more popular than I expected. I’m particularly glad that many scientists seem to have found it useful, since I primarily wrote it for them. On the other hand, it also attracted a number of criticisms, which I guess was also to be expected after the post blew up on social networks. I really don’t have time to reply to all of them, so I will focus on two lines of criticism that have come up repeatedly. First, many people seem to believe that I’m some kind of postmodernist who thinks it’s not important to empirically test scientific theories, which couldn’t be further from the truth. Second, some people have accused me of attacking a straw man, saying I only attacked a very crude form of falsificationism that neither Popper nor anybody else believes. In this post, I briefly address both of those criticisms.

First, because I argued in my post that, strictly speaking, no theory is falsifiable, many people seem to think that I believe it’s never irrational to stick to a theory in the face of seemingly contrary evidence. But nothing I say in my post implies that and, of course, I don’t think it’s true. The point was only that, because a theory always rely on many auxiliary hypotheses to make observable predictions, someone who is committed enough to it can never be forced to abandon it by the failure of a prediction because such a failure can always be blamed on the auxiliary hypotheses instead of the theory. This means that, as a criterion of demarcation between science and non-science, falsifiability doesn’t really do any work. As we shall see shortly, even Popper realized that, but he nevertheless emphasized the role of falsification because he rejected induction.

Sometimes, when a prediction fails to come true, it’s just more reasonable to blame the theory rather than the auxiliary hypotheses. If you ask someone why, the most natural reason he can give is that, given the available evidence, the theory is just more likely to be at fault than the auxiliary hypotheses. (Here is something that, using examples from the history of science to illustrate, explains how you can use Bayes’s theorem to determine whether you should blame the theory or the auxiliary hypotheses.) But this answer was not available to Popper, because he rejected induction, which in turn means that he didn’t think we could assign probabilities to hypotheses. So he was forced to find other, more roundabout ways to do the same thing and, while this might have been okay if his anti-inductivism had proven to be superior to a probabilist framework for epistemology, most philosophers of science don’t think it’s the case.

Another criticism that came up repeatedly is that I only attacked a very crude version of falsificationism that neither Popper nor anybody else has ever defended. It’s useful to distinguish between two claims here. The first is that Popper defended a more sophisticated version of falsificationism than the view I criticized in my post. The second is that nobody believes the view in question, i. e. that I attacked a straw man. On the first point, I’m willing to concede that I could have done a better job to make clear that Popper himself was aware of the problems with the crude version of falsificationism I point out in my post and therefore defended a more sophisticated version of that view, although for reasons I will sketch below, I think ultimately the problems in question also doom more sophisticated versions of falsificationism. The second claim, on the other hand, is complete bullshit. There are plenty of people, especially among scientists, who believe exactly the crude version of the view I criticized in my post.

One person who made both criticisms against me is Daniël Lakens, a psychologist who does excellent work on statistics and methodology:

Here we recognize the first of the two criticisms I have distinguished above. As I already conceded, it is not without merit.

But it’s also not entirely fair, because I explicitly acknowledged in my post that Popper defended a more sophisticated view:

Given this post’s intended audience, I will not discuss the more sophisticated versions of falsificationism that have been proposed, which often stem from remarks Popper himself made that showed he was aware of the difficulties his theory faced.

This is from the third paragraph of my post, so it’s not as if I had tried to bury the fact that Popper’s own view was more sophisticated than the crude version of falsificationism I criticized in my post.

Now, it’s true that I also say that, even though he was aware of the problems I discuss in my post and tried to address them, the crude version of falsificationism I criticize is “roughly what he believed” and this was somewhat misleading. However, it’s not entirely misleading, because Popper was not always totally clear and apparently even endorsed the crude version of the view in print on at least one occasion, according to Sven Ove Hansson’s SEP’s entry on Science and Pseudo-Science:

However, in what seems to be his last statement of his position, Popper declared that falsifiability is a both necessary and a sufficient criterion. “A sentence (or a theory) is empirical-scientific if and only if it is falsifiable.” Furthermore, he emphasized that the falsifiability referred to here “only has to do with the logical structure of sentences and classes of sentences” (Popper [1989] 1994, 82). A (theoretical) sentence, he says, is falsifiable if and only if it logically contradicts some (empirical) sentence that describes a logically possible event that it would be logically possible to observe (Popper [1989] 1994, 83). A statement can be falsifiable in this sense although it is not in practice possible to falsify it. It would seem to follow from this interpretation that a statement’s status as scientific or non-scientific does not shift with time. On previous occasions he seems to have interpreted falsifiability differently, and maintained that “what was a metaphysical idea yesterday can become a testable scientific theory tomorrow; and this happens frequently” (Popper 1974, 981, cf. 984).

So, if Popper’s view is often misrepresented (which I do not deny), it was arguably in part his fault.

But I agree that, if we want to be fair to Popper, we shouldn’t ascribe to him the view that falsifiability was a sufficient condition for a theory to be scientific. Indeed, in his most important work on falsificationism, he made it pretty clear that it was only a necessary condition but that it wasn’t sufficient. Here is for example what he wrote in section 6 of The logic of scientific discovery:

It might be said that even if the asymmetry [between verification and falsification] is admitted, it is still impossible, for various reasons, that any theoretical system should ever be conclusively falsified. For it is always possible to find some way of evading falsification, for example by introducing ad hoc an auxiliary hypothesis, or by changing ad hoc a definition. It is even possible without logical inconsistency to adopt the position of simply refusing to acknowledge any falsifying experience whatsoever. Admittedly, scientists do not usually proceed in this way, but logically such procedure is possible; and this fact, it might be claimed, makes the logical value of my proposed criterion of demarcation dubious, to say the least.

I must admit the justice of this criticism; but I need not therefore withdraw my proposal to adopt falsifiability as a criterion of demarcation. For I am going to propose (in sections 20 f.) that the empirical method shall be characterized as a method that excludes precisely those ways of evading falsification which, as my imaginary critic rightly insists, are logically possible. According to my proposal, what characterizes the empirical method is its manner of exposing to falsification, in every conceivable way, the system to be tested. Its aim is not to save the lives of untenable systems but, on the contrary, to select the one which is by comparison the fittest, by exposing them all to the fiercest struggle for survival.

What he is saying in that passage is that it’s not enough for a theory to be scientific that it be falsifiable, because strictly speaking no theory is falsifiable since one can always conjure up auxiliary hypotheses to explain away seemingly contrary evidence (as I explained in my post), it must also be the case that the proponent of that theory adopt a certain methodology that Popper describes later in the book and which prohibits this kind of strategy.

What this means is that, despite what the passage quoted by Hansson above suggests, Popper’s criterion of demarcation applies not just to theories, understood as sets of sentences, but also to the attitude of their proponents. So there is no doubt that Popper defended a more sophisticated version of falsificationism which is not so easily refuted and, even though I acknowledged that in my post, I agree that I could have made that much clearer. To be clear, I don’t think even this more sophisticated version of the view is correct, I think it’s completely misguided. But it would take a lot more than a blog post to argue for that claim and I didn’t think it would be particularly useful to get into that kind of details, because as I say at the beginning of my post, “it’s the crude version that most people, especially scientists, have in mind when they talk about falsificationism”. This brings me to the criticism that I was attacking a straw man, because nobody actually believes this crude version of falsificationism. Before I address this criticism, however, I want to say a few words about why I think even more sophisticated versions of falsificationism are hopeless, so people don’t think I’m making a purely gratuitous claim.

Basically, my view is that once you take seriously the role of auxiliary hypotheses (what is known in the literature as the Duhem-Quine thesis), the alleged asymmetry between verification and falsification is revealed to be largely illusory and you find yourself faced with very similar problems as inductivists but without their tools. (Indeed, as many people have remarked, Popper ends up having to sneak in through the backdoor many concepts from traditional epistemology, sometimes under the same name and sometimes not, to play the same role but in a more contrived way.) Since this alleged asymmetry was the main motivation for Popper’s falsificationism in the first place, I don’t see then why we should find it attractive anymore, which is precisely why I said in my post that once you complicate the view “loses much of its original appeal”.

So while people say that my criticism doesn’t really show that Popper’s philosophy of science is mistaken, I think in a way it does, but to see that you have to take the Duhem-Quine thesis seriously and see what consequences it has for Popper’s anti-inductivism, which would require a lot more work than was possible in that post, which I pitched at non-specialists. Once you reject induction, you are led to a highly revisionary account of science, on which it’s a purely deductive enterprise. This has a cost and I just don’t think that Popper’s alternative picture of science is attractive enough to justify that we pay it. Again, I understand that some philosophers of science disagree with that (although the overwhelming majority of them do not), which is fine. But this is a complicated debate and I don’t think it would have been very useful if I had tried to make that case, because as I already noted, most people only know about the crude version of falsificationism I criticized in my post.

This last claim, however, is also denied by Lakens:

So the claim is that nobody, at least no scientist, believes the crude version of falsificationism I criticized in my post. On this view, I attacked a straw man.

But this is complete nonsense. I’ve had countless conversations with scientists in many different fields who recited to me exactly the crude falsificationist credo I attacked in my post. I’ve also had countless conversations with other philosophers or philosophically sophisticated scientists who told me they’ve had the same experience. Here is for instance what Sabine Hossenfelder, a theoretical physicist, wrote in a blog post about the negative consequences the popularity of this crude version of falsificationism had in the field of particle physics:

Even in his worst moments Popper never said a theory is scientific just because it’s falsifiable. That’s Popper upside-down and clearly nonsense. Unfortunately, upside-down Popper now drives theory-development, both in cosmology and in high energy physics.

Is Lakens saying that we are just making this up or that we are imagining things? Anyone who knows how to use Google can find plenty of example of people who endorse precisely the crude version of falsificationism I criticized in my post.

In fact, in response to my post, many scientists told me the crude version of falsificationism I criticized was uncritically accepted in their field. Here is what a PhD candidate in psychology even said a few months ago:

After I published my post, she shared it on Twitter with the comment that, “in 10 years of higher education, not one person explained the problems with falsification”. Judging by the reactions to my post, and the fact that it was shared by hundreds of scientists, she wasn’t the only one.

This isn’t because scientists are stupid. I used to date a chemist who recited to me the crude falsificationist credo on several occasions, yet she is one of the smartest persons I have ever met in my life, and I have met a lot of very smart people. It’s just that naive falsificationism is all the philosophy of science they have ever been exposed to and, when you don’t take some time to think a bit harder about it (which most scientists don’t because it’s not their job), it’s not obvious that it’s false and it even seems pretty convincing, as I noted in my post. In fact, even after I published my post, there are still people who seem intent on defending this crude version of falsificationism. If you read the replies to my post, you will see that, while some people argue that Popper’s view was more sophisticated, the vast majority of the people who were not convinced by it make no such suggestion and instead seem to think that something like the crude version of falsificationism is correct despite my arguments.

Perhaps Lakens is better acquainted with the view Popper himself defended, but frankly I’m not sure he is:

Not only would it not be hard to find a psychologist who said that “if it is not falsifiable it is not science”, but what makes this tweet really bizarre coming from someone who criticized me for ascribing to Popper views he did not hold is that Popper himself consistently made that claim…

Indeed, the claim that “if it’s not falsifiable then it’s not science” just means that falsifiability is a necessary condition for something to be scientific, which as I already noted is a view that Popper definitely endorsed. Here is a passage in Conjectures and refutations, for instance, where he makes that claim:

A theory which is not refutable by any conceivable event is non-scientific. Irrefutability is not a virtue of a theory (as people often think) but a vice.

The view that Popper rejected is not that falsifiability was a necessary condition for something to be scientific, which on the contrary he repeated again and again, but that it was sufficient.

This is important because it’s actually very easy for a theory to be falsifiable once you take into account the role of auxiliary hypotheses. Although people who believe naive falsificationism don’t seem to realize it, any pseudoscientific theory can easily be made falsifiable, so this condition doesn’t really rule out anything once you realize that no theory is falsifiable on its own, but that auxiliary hypotheses are always needed to derive observable predictions. For instance, it’s easy to show that astrology is falsifiable at least in principle, but it’s also easy to explain away apparent falsifications of astrology by introducing purely ad hoc auxiliary hypotheses. Similarly, I often see scientists who work in a STEM field claim that e. g. history is not really a science on the ground that it’s not falsifiable, but that’s just not true. Insofar as physics is falsifiable, so is history. For example, if a historian claims that Julius Caesar was dictator in 40 BC, this can be falsified by pointing out that we have several historical sources according to which he was assassinated in 44 BC. Of course, the revisionist historian could always argue that the sources in question are not reliable, but from a purely logical point of view this is no different from a physicist who claims his measurement apparatus was not reliable to explain why a prediction he derived from his favorite theory failed to come true.

The way in which Lakens tries to show that nobody believes the crude version of falsificationism I criticized in my post is even more bizarre:

Of course, this doesn’t prove that no scientist believes naive falsificationism any more than the existence of Oskar Schindler, who joined the NSDAP in 1939, proves that no nazi was antisemitic… Anyway, I’m sure that Lakens himself does not believe the crude version of falsificationism I criticized in my post (I wouldn’t follow him on Twitter otherwise), but I don’t find his argument that it’s a straw man very convincing.

What he really seems to be getting at in that tweet, which he’d already hinted at in the previous tweet I quoted, is that scientists don’t really believe the crude version of falsificationism I criticized in my post since they don’t behave as this version of falsificationism says they should. But I actually pointed out in my post this discrepancy between the crude version of falsificationism scientists often profess to believe and their actual behavior:

It’s ironic that, although many scientists accept falsificationism uncritically, what they do every day is radically at odds with it.

Now, we could argue all day long about whether, in view of this discrepancy, scientists who recite the naive falsificationist credo really believe it, but that wouldn’t be very interesting. (Do self-professed christians who cheat on their spouse not really believe in the Gospel? I doubt many people would say that, but who cares, they still go to church.) What matters is that they do often recite it and, as Hossenfelder and many other people have noted, this sometimes affects their behavior and has adverse effects on science, which is why I thought it was useful to debunk that crude version of falsificationism.

As I said in my original post, my goal was pretty modest:

I know there are philosophers of science who think that falsificationism can be salvaged and that a more sophisticated version of that view is correct, so they will predictably not be moved by this post, but I’m fine with that since it’s the crude version that most people, especially scientists, have in mind when they talk about falsificationism. I will be satisfied if all I have achieved with this post is to convince you that, at least in this crude version, falsificationism is false. Although I ultimately disagree, it’s okay with me if you think the lesson one should draw from this critique of naive falsificationism is that one should adopt a more sophisticated version of the view, instead of abandoning the project altogether.

Anyway, I hope this post will have clarified a few points that were not clear in my original post, though I’m under no delusion that it will suffice to appease the most fanatical of Popper’s followers, who have been hounding me since last week. (To be clear, I’m not talking about Lakens, who I’m not even sure would self-describe as a Popperian and who in any case just made a few tweets critical of my post.) I’m sure it won’t be long before some of them show up in the comments of this post to castigate me for my offenses against their hero and explain why everything I say is completely wrong. As I joked on Twitter the other day, “hell hath no fury like a Popperian scorned”.

4 thoughts

  1. You are absolutely right: there are prominent people who have tried to sell the ignorant caricature of “naive falsificationism” as something that a) bears any relation to what Popper wrote and b) should be critiqued: Kuhn and Feyerabend. They disingenuously misrepresented Popper, but they, at least, are not straw-men.

    The actual straw-man that you erected, of course, isn’t that nobody believes in it. It is that it has anything to do with Popper’s philosophy. Your protestations to the contrary, that you criticized only a “crude version of falsificationism” and not what Popper actually said, are, pardon my French, pretty hypocritical given that your tweet to publicise it said explicitly that your blog post “explains why falsificationism, Popper’s philosophy of science, is false”. And in the post, you talk about variations of “what Popper said” almost a dozen times (without ever quoting him and without getting him right, unfortunately). You are still trying to have it both ways.

    And now you’re making a big show of “responding to criticism”—while ignoring the only detailed criticism (in the comments) that is actually about falsificationism. In case any of your readers, or you yourself, are interested in a more detailed explanation of what falsificationism, and especially the criterion of falsifiability, actually means, see this comment by me. I’ll happily engage with criticism and/or queries.

  2. Schopenhauer thought Hegel was a fraud because he didn´t really promote the solution of innovative philosophical problems. In the same vein, thinkers who disbelieve in the central role that Popperianism in science are looking at new philosophical problems in relation to culture. To tell you the truth, Popper never even attempted the systematization of his ´sophisticated´ falsification system, for true sentences that apply to empirical theories, e.g., he never really looked at why for example geocentrism became heliocentrism, in a broad sense. He was no true historiographer of science. Carnap came closest to try to do the sentence verification of all possible theories approach, but it never resulted in a bonafide philosophy of science (The Duhem Quine thesis again).

    Great Article!

  3. It seems like most scientists don’t think of falsifiable theories as a classical category that has necessary and sufficient conditions for membership. Instead, I think scientists think of theories as things that range in how falsifiable they are.

    Take some feminist theory (I know you love these =p), and imagine that person X puts forth some observation and claims it counts as evidence against the theory. If person X is male, then theory is further confirmed because men will try and prove it wrong to gain more power (or something..?). If person X is female, then the theory is further confirmed because women who try and show the theory wrong have internalized male power (or something..?). There’s almost no sense in which the veracity of this theory depends on observations of the world. It’s consistent with too much.

    Now take some physical theory. The veracity of the theory depends on observations collected that depend on some auxiliary hypotheses about devices that measure voltage. Person X puts forth some observation and claims that it counts as a evidence against the theory. The theorist says that this doesn’t falsify my theory because the device has malfunctioned. Person X then performs the same experiment with 100 different independent devices and gets roughly the same results. Further, we have engineering specs on the devices like failure rates. A lot of evidence has now built up against the theory. The theorist can’t reasonably claim that the devices are all failing consistently in the same way.

    A typical scientist will consider theories similar to the first one on the far unfalsifiable end of the spectrum and theories similar to the second one on the far falsifiable end of the spectrum. The theories that depend more on observations about the world and have auxiliary hypotheses with few degrees of freedom are more desirable because they are more useful. If the theory doesn’t seem to depend on collecting observations at all, then it shouldn’t be in play. Of course the scientist knows this gets blurry toward the middle, but broadly this seems to be what they mean when they say something is unfalsifiable and, hence, not worthy of consideration.

    1. » Daan:
      It seems like most scientists don’t think of falsifiable theories as a classical category that has necessary and sufficient conditions for membership. Instead, I think scientists think of theories as things that range in how falsifiable they are.

      I mean, if only somebody had written a book to explain falsifiability that contained a whole chapter on “degrees of testability”, right? 😉

Comments are closed.