Are most logical fallacies defective? Below, I will argue that the answer is ‘yes’. That is, I shall argue that a great many logical fallacies do not themselves provide even prime facie grounds for rational doubt, even when applied in the standard appropriate context.
Skeptics in the ancient world referred to the “Agrippan trilemma”, which is a set of three fallacies — infinite regress, arbitrary stipulation, and circularity — that are supposedly shared all (or virtually all) arguments. However, the state of the Agrippan trilemma isn’t what it used to be. A complaint about infinite regress is wholly uninteresting to the defender of infinitism; complaints about arbitrary assumptions are of no consequence to the foundationalist; the complain about circularity has no traction for the coherentist. None of these views are absurd (though some are late bloomers). (Even the idea that all contradictions are false is now suspect, if you’re a dialetheist, though this is far more controversial.) It is striking to note that a war of attrition is being fought against skepticism, and that skepticism appears to be on the losing side.
Let’s consider some other examples.
The “strawperson fallacy” is hard to take seriously when many quite good articles in philosophy engage in a refutation of ideal-types of a cluster or syndrome of related arguments in a corpus. On one very plausible reading of the concept of intuition, the “appeal to incredulity” is simply an appeal to intuition under a guise, which has a non-trivial (though limited) role in legitimate inquiry. Since intuitions are an intellectually complex form of feeling, “appeals to emotion” must also be valid on occasion: in particular, when pointing to the difference between inferences and mere associations.
The “slippery slope fallacy” is hard to reconcile with a standard worry issued in critical theory, which is that inquiry has to take into consideration the consequences of the thing being posited. If my conception of “racism” or “sexism” has pernicious consequences, then that would seem to count as a reason against that conception, irrespective of its empirical plausibility. The reason is *not* because we think justice trumps truth, but because we acknowledge that social groups are interactive kinds.
If Kuhn is right, then “special pleading” is routine in the natural sciences. When you are confronted with a surprising and seemingly unnatural result, the right heuristic is to assume you did the experiment wrong. Potential falsifiers show up all the time, and nobody cares, because these would-be falsifications are probably just mistakes. See, e.g., cold fusion.
If we are to have any respect at all for the dignity of other groups to define their own self-conceptions, then we end up having to concede that “ad hominems” are legitimate when they are levelled against speakers who have crossed epistemological jurisdictions, and the assertion of what counts as a “true Scotsman” is legitimate when asserted within the scope of those jurisdictions. “Bandwagons” and the “genetic fallacy” are legitimate under the same conditions.
On the face of it, “appeal to authority” would make nonsense of legal positivism (and, in my opinion, the entirety of moral discourse), which if true would be pretty good reason to think it is a hasty accusation. Also, the accusation latent in the “tu quoque fallacy” seems to undermine a vital presupposition of moral claims, which is that the person who asserts a moral claim has some kind of shared access to the conditions that make the rational authoritativeness of the claim. Hence if I say “stealing is wrong,” and I am a thief, then not only can you accuse me of hypocricy — you can also infer that I am no justification for believing that stealing is wrong. Since the burden of proof is on me to provide that justification, then all other things equal, you can forbear from deferring to what I have said: i.e., that stealing is wrong. But maybe, when it comes to some subjects, the burden of proof does not lie in the one who asserts, but instead in any interested party. In that case, “tu quoque” remains a fallacy, though the idea of “burden of proof” looks like it has some holes in it.
They say that “the plural of anecdote is not data”. (Taken literally, this is nonsense: if anecdotes were not data, they would be so fully uninstructive as to be unintelligible.) What people mean is that anecdotes are not evidence — that is, it is not on the face of it public reasons for belief in the truth of some proposition. But while anecdotes are not public reasons for belief, they surely are private reasons for belief insofar as the stories we tell ourselves about our experiences are involved in the production and reproduction of accurate memories. The plural of “anecdote” is not “evidence”, but rather, “narrative”: and I am partly made up of my honest narrative, so that had better count for something.
In the above, I presented a litany of arguments against many fallacies. If I am right, then they are defective: either they apply in a narrower ranger of contexts than are advertised, or they apply across those contexts with null force. In either case, it seems like a pedagogically important point to make.
After all, that list is incomplete. I have ignored some other fallacies, which I do not really have occasion to doubt: the gambler’s fallacy, false dichotomy, loaded question, begging the question, false cause, appeal to nature, composition/division, Texas sharpshooter, and the middle ground. (I could probably appeal to ordinary usage in nitpicking some of these fallacies, too: e.g., I might say that current citation practices in philosophy are less about rigorous meta-analysis and more about “Texas sharpshootin'”. But good taste forbids doing such a thing.)