Monday, April 17, 2017
The Fallacy Fork and the Limits of Logic
by Paul Braterman
I had been waiting for a quiet moment to write about this, but there isn't going to be a quiet moment, so now will have to do.
Debaters regularly accuse their opponents of using fallacies. These can be formal fallacies, such as simple errors of logic, or informal fallacies, such as appeal to authority, ad hominem and strawman arguments, among others. If a piece of reasoning depends on any of these fallacies, so it is claimed, the conclusion does not really follow from the premises, and while it might still be true we have not been given any good reason to believe it. And so books that discuss logic, and science-promoting blogs (including one I follow), regularly include descriptions of informal fallacies, with stern instructions to avoid committing them.
In an article entitled The Fake, the Flimsy, and the Fallacious: Demarcating Arguments in Real Life, Maarten Boudry, Fabio Paglieri and Massimo Pigliucci (henceforth BPP) challenge this view. BPP is written for the perusal of trained philosophers, which I am not, but I use it here as a jumping off point, while mixing in further content of my own.
BPP apply what they call the fallacy fork test to accusations of informal fallacy; either the reasoning is obviously erroneous, in which case no one would really use it, or else it is not obviously erroneous in context, and we still have all the work to do. In the first case, formal analysis is redundant; in the second, the facts of the matter need further consideration. So naming and shaming the particular kind of fallacy is either unnecessary or uncalled for. I agree, and suggest that we drop the label "fallacy" for such informal arguments, since to make the label stick we have to show on other grounds that the argument as used really is fallacious. The discussion made me think of my own reflections on formally valid logical arguments, which only work because the conclusion has actually been accepted in advance, otherwise we would not have accepted the premises. In both cases, the formal or semi-formal reasoning, while seemingly at the heart of the argument, is an unnecessary elaboration, and we can cut out the middleman.
Going beyond this, it is becoming increasingly clear that the logical is only one aspect, and not usually the most important aspect, of an argument. More important are the heuristic and rhetorical aspects; will the reasoning point us towards a way of acquiring new knowledge, and how useful is it in the attempt to persuade our opponents to change their minds. (In passing, I suggest that if our objective is to persuade others to change their minds, we are not arguing in good faith, unless we are at least in principle open to the possibility that we too might change ours.)
BPP focus on four particular kinds of alleged fallacy, ad hominem (criticism of person used as criticism of argument), post hoc ergo propter hoc (false claims of causation, just because one thing follows another), genetic fallacy (discounting a statement because of the person who makes it or the way in which it comes to be made), and ad ignorantiam (sometimes called the argument from ignorance; mistaking absence of evidence for evidence of absence).
Post hoc ergo propter hoc: I took the medicine, I got well, and I conclude the medicine cured me. This is induction from a single case, a logically invalid procedure, and maybe I would have got better anyway. I have obviously jumped to the wrong conclusion if I'd taken a homoeopathic medicine, but what if I'd taken penicillin? What if I'd taken penicillin, but it then turns out that I was suffering from a virus, which penicillin doesn't touch? The reason we dismiss the possibility of homoeopathic cure has got nothing to do with the shortcomings of post hoc ergo propter hoc reasoning. It depends on something else altogether; the fact that homoeopathic medicines don't actually contain anything except water and random trace impurities. In other words, we don't believe in homoeopathy because we don't see how it could have done the job, and a single example isn't going to make us change our minds.
There may be cases sufficiently complex that we don't know whether a plausible mechanism exists or not. The Congressman's allergies got better after he was stung by a bee. Congress decided that the National Institutes of Health should devote some small fraction of their budget to exploring such unorthodox treatments, and was right to do so.
Then there is the tragic case of the antivaxx movement. Children had been vaccinated, some developed autism, and it was alleged that the vaccination caused the autism. Never mind that this is bound to be the case, because the vast majority of children get vaccinated shortly before the age at which autism would if present be diagnosed. This claim was uncritically publicised to a mass audience on the Oprah Winfrey Show. Now we have measles epidemics caused by refusal to vaccinate, and the self-righteous apostles of antivaxx have blood and pus on their hands.
Doesn't this illustrate the importance of identifying fallacies, such as post hoc propter hoc? Not really. The antivaxxer movement took off because of a fraudulent paper in a respected medical journal, general suspicion of the pharmaceutical industry and of official guidance, the emotional appeal of the natural (immunity by infection) as opposed to the technological (immunity by vaccination), the chance fact that more cases of autism were being diagnosed because the definition had been broadened, and a handful of passionate and glamorous advocates. A lot more to it, alas, than a simple error of logic.
Ad hominem: dismissing a statement by discrediting the person making it. BPP give the example of climate change. X is a member of a right-wing think tank, and I invite you to ignore X's argument against the reality of global warming. Apply the fork. Yes, I have used an ad hominem argument. But it really is relevant that X is likely to be biased in his choice of examples, and that we need to take this into account. We are receiving messages all the time from people with an agenda to pursue; advertisers, politicians, representatives of particular interest groups. Indeed, there is one important situation - the court case - where it is the duty of the advocates to be biased. If we simply ignored their arguments for that reason, we would indeed be guilty of the ad hominem fallacy, but it would be just as stupid if, for fear of committing such a fallacy, we failed to take their bias into account. Again, as BPP urge, invoking the name of the fallacy does not relieve us from the hard work of weighing the evidence. And, far from misleading us, ad hominem reasoning can usefully remind us that the evidence itself might be weighted.
Next, BPP discuss argumentum ad ignorantiam, the alleged fallacy of taking absence of evidence for evidence of absence. They consider two cases. On the one hand, a sceptical response to the waves of witchcraft hysteria that periodically sweep the United States, and led talk-show host Geraldo Rivera to say of satanic child abuse "The odds are that this is happening in your town." The sceptic would point to the complete lack of evidence in support of such allegations, apart from dubious and highly fantastical "recovered memories". BPP then consider the case of evolution, and the absence that in 1859 Darwin deplored of the intermediate fossil forms that evolution implies. Darwin's response was to admit that this was indeed a weakness, attributing it to the rarity of fossil formation and the fact that so little of the Earth's surface had been surveyed for fossils, and to hope that the weakness would be cured in time, as it has been. We quite correctly use an argumentum ad ignorantiam to discount Geraldo, while objecting to its use against Darwin. The difference is, that if Geraldo's claims had been based on fact, we would have expected abundant material evidence for them, so their absence gives us good reason to reject them, whereas absence of the necessary evidence in 1859 was not really that surprising even if (as we now know) Darwin was essentially correct.
The argument from absence now strongly confirms the evolutionary sequence. As the pioneer geneticist JBS Haldane is said to have remarked, oover fifty years ago, there are no rabbits in the Precambrian. Once again, the issue turns, not on the logical structure of the argument, but on the background plausibility expectations that we bring to it.
The genetic fallacy, of which the ad hominem fallacy is a special case, is when we reject an argument because of its origins; "genetic", of course, is being used here in its older sense of "related to origin". We might be accused of committing this fallacy, as BPP point out, if we dismiss religion on the grounds that we see it as an evolutionary adaptation, or if we were to oppose smoking bans on the grounds that such bans had been instituted by the Nazis. But once again, whether or not a fallacy has been committed depends on background information, not the logical structure of the argument. If we could explain religion in evolutionary terms, that would not refute religion, but it would nonetheless be a valid counter to Calvin's doctrine that the propensity to believe is evidence for a Creator. One argument (it increases fitness) really does undermine another argument (it must be God-given), and we have all the work of sifting the evidence still to do. And few of us would be so foolish as to advocate meat-eating on the grounds that Hitler was a vegetarian.
In all these cases,to label the argument from the outset as a fallacy is to assume what is meant to be proven. This really is a fallacy.
There remain two further fallacies that, unlike BPP, I do regard as well-defined and always fallacious. Firstly, there is the argument from nature. BPP think it legitimate to argue that it is wrong to feed a cat on vegetables, because it's not natural. But it is equally unnatural to neuter the cat, or even to feed it on factory-prepared cat food. The reason it is wrong to feed the cat on vegetables is that the cat can't digest them. There is no need to appeal to nature in a broader sense at all, and such appeals, which reach across the usual left-right political spectrum, are dangerous and damaging. I have already mentioned how they feed into the antivaxx movement. They are used to justify opposition to GMOs (as in the slogan "Frankenfood"), and also lie behind restrictive sexual moralities, such as the Catholic ban on contraception, which do great harm
My other example is the strawman fallacy. I think you really are committing a fallacy if you claim to have demolished an opponent's argument, but the argument demolished was different from the one being put forward. For example, A might accuse B of denying Israel's right to exist, when B really did no more than criticise Israel's administration of the Occupied Territories. B may then accuse C of anti-Muslim bigotry  if C criticises Muslim inheritance law. Given sufficient goodwill on both sides, further discussion could make this clear, and leave A, B, and C with a more nuanced understanding of each other's true positions, although in these cases I doubt that this would happen, and indeed the fallacies may well have been committed deliberately, as a way of sabotaging discussion.
Finally, I return to my own earlier critique of the traditional example of sound argument: all men are mortal, Socrates is a man, therefore Socrates is mortal. Of course the argument is formally valid; if the premises are true, the conclusion cannot possibly be false. But the argument is useless, because we had already implicitly accepted the conclusion (that Socrates is mortal) before we assented to the initial premise. Indeed, the conclusion is easier to establish than the premise; we will accept it even if we believe that Elijah, Ganymede, or the Virgin Mary were taken up into heaven while still alive.
BPP describe the entire activity of searching out fallacies, and indeed of arguing by reference to formal structures, as uncharitable. I agree. There are, moreover, much more serious errors of reasoning, such as confirmation bias, cherrypicking, respect or even reverence for the ideas that define one's own group, and focusing on the recent, the spectacular, and the unusual (terrorist drivers as a cause of death, for example, rather than drivers distracted by cellphones). These, however, must be addressed, not in terms of the formal logic of the arguments, but rather of their psychological underpinning. We are the flawed products of our evolutionary history, and no amount of appeal to formal rules can rescue us from our fallibility.
1] The conclusion may nonetheless be true, for other reasons, and the erroneous view that it is bound to be false is sometimes known as the fallacy fallacy.
2] Our museums are now overflowing with intermediate fossils, but that does not stop evolution deniers from perpetuating the argument, focusing attention on particular cases where the record is thin, such as the early Cambrian, and quote-mining Darwin in support of their case.
3] This is what is properly referred to as begging the question, although the meaning of this expression has lately become blurred.
4] I avoid the term "Islamophobia" because it so strongly invites precisely this kind of confusion.
Posted by Paul Braterman at 12:35 AM | Permalink