What is a Fallacy?
Definition
A fallacy is a general type of appeal (or category of argument) that resembles good reasoning, but that we should not find to be persuasive.
Discussion
We need to be careful in our definition of fallacy. What we include in our study and what we exclude from it will be determined by how we define our subject. In his 1970 book Fallacies, C. L. Hamblin offered some devastating criticism of the way in which fallacies traditionally had been treated. Among his criticisms, he pointed out that many of the fallacies on classical lists fail to qualify as fallacies according to the classical definition. Since then, serious students of fallacy have scrambled to find a revised list more suited to the definition, or a revised definition more suited to the list, or to make judicious revisions in both. This is not the place for a comprehensive review of the issues that have been raised in the wake of Hamblin's criticisms, but let me touch on a few of the considerations that went into my definition.
According to Hamblin, the classical definition of a fallacy is, "an argument that appears to be valid, but is not." (Hamblin 12) One might take issue with this definition on a number of points, but let us consider just three:
(1) whether fallacies should be regarded as types of arguments;
(2) whether fallacies are invalid; and,
(3) whether we can make sense of a subjective claim about how fallacies appear.
Are fallacies arguments?
Strictly and classically speaking, while we may say that an argument is fallacious or commits a fallacy, the term "fallacy" does not refer to an argument, but to an error of some identifiable kind. All of the arguments that are guilty of that error may be said to be instances of that fallacy, so fallacies are strictly and classically considered to be types of arguments.
However, there is a growing movement among modern students of fallacy to consider fallacies, not as errors in a single argument, but as illegitimate moves in the broader context of a dialectical discussion. Aristotle himself seems to support this view, referring to his examples of bad reasoning as "elenchoi" or refutations rather than as arguments. Schopenhauer also clearly falls into this camp. His essay "The Art of Controversy" is not so much a list of fallacies in the classical sense as a compendium of debator's tricks that can be used to put an opponent at a disadvantage. Shopenhauer's list might easily have included such tactics as burping loudly while an opponent is making a particularly cogent point, to break the opponent's (and the audience's) concentration. Such ploys clearly are not arguments, but many of the items that appear on at least some classical lists of fallacies are not much better. Perhaps they, too, would be better understood as rhetorical ploys rather than as failed arguments. For this reason, some modern students of fallacy, including Hamblin, have taken the view that the study of fallacies should break out of the limits imposed by focusing on arguments in a narrow sense.
I shall, however, stay with the more narrow and classical sense of the term "fallacy." While persuasive or dialectical discourse may have a larger structure of its own, and much may be learned by studying disputation in its broad context, persuasive discourse must ultimately be composed of premisses used in support of conclusions. It is the errors that occur at the level of the premiss-conclusion relation that I am interested in examining. I try to limit the fallacies on my list to errors that affect the way in which at least implicit conclusions are drawn--by one party or another--from at least implicit premisses. However, by way of illustrating what I have not included, I have compiled a short list of rhetorical ploys that are treated by some logicians as "fallacies."
Are fallacies invalid?
Validity has a very precise--indeed a mathematically precise--meaning in modern logic. (An argument is said to be valid if and only if the form of the argument is such that any set of true premisses that instantiate that form must necessarily accompany a true conclusion). This definition was unknown to the classical philosophers who first studied fallacies, so it should not be surprising that there is a discrepancy between their intended meaning and the modern meaning. Unfortunately the modern meaning applies only to Deductive arguments. Since I think a treatment of fallacies ought to include consideration of the errors made in Inductive and hypothetical (or Retroductive) arguments as well, the modern conception of validity has little value in this context. Moreover, the modern concept of validity would have been misplaced in any case. It is manifestly obvious that a valid argument can be fallacious. For example:
All members of the National Rifle Association are ignorant yokels. No ignorant yokels are people whose opinions are worth considering. Therefore, no members of the National Rifle Association are people whose opinions are worth considering.
This argument is clearly valid (and provably so, by standard rules of syllogism); but, it is also clearly fallacious: it is an example of the Ad Hominem - Abusive fallacy. So, clearly, valid arguments can be fallacious.
I think logicians have traditionally misunderstood the role that validity plays in the critique of reasoning. We do not use validity to decide whether an argument is good or bad. Rather, I believe that the concept of validity serves merely to distinguish utterances that qualify as arguments from utterances that fail to qualify as arguments. The concept of validity in logic corresponds to the concept of grammaticality in grammar. A sentence that fails to be grammatical is not correctly structured, and so fails to be a sentence at all (technically speaking). Likewise, an argument that fails to be valid is not correctly structured (for its type), and so fails to be an argument at all (technically speaking).
For Deductive arguments, the modern concept of validity tells us whether an utterance is a well-formed Deductive argument, not whether we should be persuaded by it. A Deductive argument is well-formed when it is impossible for the premisses to be true while the conclusion is false. A doctrine of well-formedness can also be developed for Inductive and Retroductive arguments. My 2016 paper, "Deductively valid, inductively valid, and retroductively valid syllogisms," Transactions of the Charles S. Peirce Society, vol. 54, no. 4, makes a first step toward developing such a doctrine. In the extended sense, it would be true to say that all arguments, both fallacious and non-fallacious, are valid (for their type), simply because any utterance that failed to be valid would fail to qualify as an argument of the specified type.
Having dispensed with the concept of validity as a criterion of good reasoning, it remains to consider what criteria should be used instead. In the above definition I have simply used the non-committal phrase "good reasoning," in the hope that it will stand in for whatever criteria we ultimately decide to adopt.
Good reasoning is reasoning that tends, in the long run, to produce true conclusions. In the end, the measure of good reasoning is that it tends to move us closer to the truth. However, a fallacy is not just any type of reasoning that might lead to a false conclusion. Even perfectly legitimate patterns of reasoning might lead to a false conclusion from time to time, simply because uncertainty is a necessary feature of the logical landscape. Whenever we generalize from a sample (Inductive reasoning) we run the risk that our sample--however carefully we draw it--might not accurately represent the population from which it was drawn. Induction is notoriously unreliable, and Retroduction is worse! Even a Deduction guarantees a true conclusion only when its premisses are true. However, for all their faults, Deduction, Induction and Retroduction, used with appropriate care, can lead us to the truth in the long run. Fallacies occur when something undermines or subverts this general tendency.
How can bad reasoning appear to be good?
An argument is generally considered to be fallacious not merely because it commits an error, but because there is some risk that someone might be taken in by the error. A fallacy is not just bad reasoning, but bad reasoning that appears to be good. This is an idea that has its origin with Aristotle. In the Sophistical Refutations Aristotle spends quite a bit of time explaining that sophistical reasoning mimics good reasoning, i.e. that fallacies are a counterfeit of legitimate reasoning. In his essay, "The Blaze of her Splendors: Suggestions about Revitalizing Fallacy Theory," (which appears in Fallacies: Classical and Contemporary Readings) Ralph H. Johnson argues that we should drop this aspect of the definition on the grounds that it renders the concept of fallacy wholly subjective. An argument is fallacious only if it appears (to someone) to be good reasoning. Yet, appearance is subjective, and we surely want to be able to identify fallacies by some objective criteria rather than by subjective reactions.
I find the idea that fallacies are a counterfeit of legitimate reasoning to be highly revealing, and I am reluctant to give the notion up, even for Johnson, whose views are otherwise similar to my own. Fortunately, I don't believe it is necessary to do so in order to make our criteria objective. Johnson does not see a problem in identifying arguments that fail to be good reasoning; he even offers some specific criteria distinguishing bad reasoning from good reasoning. The only problem, then, is in determining when a bad argument resembles or appears to be a good argument. Johnson is right that this should not be left as a merely subjective judgment. However, I don't believe we need to get lost in an intricate theory to render the notion of resemblance more objective. The solution is really quite simple: we can know that a bad argument appears to be good (to someone) if we can cite instances of people having been fooled by it, or at least instances of people having offered the argument in the hope that someone would be fooled. (Whether a speaker believes his own argument is a matter that we cannot decide, but we can surely understand when he hopes that someone else will believe it, for why else would he offer the argument?) To identify an argument as fallacious, then, we need to show that it fails to be good reasoning; but, beyond that, we only need to document that the argument was actually used. A fallacy will be in common use only if some of the people are fooled by it at least some of the time (or someone else hopes they will be), and they will be fooled only if it appears (to them) to be good reasoning. Johnson himself proposes something like this criterion when he suggests that a fallacy should occur "with sufficient frequency in discourse to warrant being baptized."
In addition to documenting the currency of fallacies in actual practice, I think it is also worth considering what sort of legitimate appeals may lend themselves to being mimicked. Contrasting fallacious reasoning with the legitimate reasoning that it counterfeits serves two useful purposes. On the one hand, it gives us a real insight into the nature of the fallacy, including why it is an example of bad reasoning. On the other hand, the contrast constantly reminds us that there is such a thing as good reasoning. Students who get caught up in the study of fallacies often get so good at recognizing fallacies that they see fallacies everywhere. The can even come to believe that ALL arguments are fallacious. This is an important error. Fallacies mimic good reasoning in the same way that a counterfeit bill mimics real money. If there were no real money, there would be no counterfeit money either, since there would be nothing to mimic. Similarly, if there were no truth, or no possiblity of arriving at the truth, then there would be no "fallacies," for there would be no path to the truth for a fallacy to undermine. Above all else, the study of fallacies should remind us that there is such a thing as truth, and that the truth is worth finding.
WELCOME EXPLANATION OF PRINCIPLES TABLE OF FALLACIES EXERCISES INDEX