This is a list of some of the more frequently used informal fallacies. If you are studying logic from a textbook, you may notice both similarities and differences from your text. I'm organizing them in an order similar to the way Robert Paul Churchill organizes them in his Logic: An Introduction (St. Martin's Press), a textbook superb in its treatment of fallacies.
(We are covering only informal fallacies here. Formal fallacies, based on failure to follow rules of well-structured arguments, also exist -- e.g. "undistributed middle," "affirming the antecedent" -- but it isn't necessary to cover these here for our purposes.)
(Some enterprising people manage to commit more than one fallacy at a time! And sometimes, it's not obvious which category of fallacy is being committed. That just goes with the territory. )
1. FAILURES OF ARGUMENT:
Begging the Question,
2. Non Sequitur FALLACIES OF RELEVANCE
3. Non Sequitur FALLACIES OF WEAK INDUCTION
4. Non Sequitur FALLACIES OF PRESUMPTION
5. Non Sequitur FALLACIES OF AMBIGUITY, VAGUENESS, etc.
6. Nasty persuasive rhetoric, a.k.a. SOPHISTRIES
7. OTHER FALLACIES
8. A few notes on Statistical and hypothetical reasoning
1. FAILURES OF ARGUMENT: Inconsistency, Begging the Question, Disappearing Hedge, Non Sequitur
An argument (in the philosophical sense, i.e. a usually very pleasant discussion) consists of:
1. Premise or Premises that are already accepted by both the presenter and the listener,
And which give good grounds for accepting the:
2. Conclusion that follows from the premise(s), already accepted by the presenter.
In other words, the presenter is convincing the listener that since she already accepts the premises,
she should also accept the conclusion as well.
An argument can fail in a variety of ways.
Some of the biggest howlers include:
- Fallacy of Inconsistency: in which either the premises are inconsistent with each other, or the conclusion is inconsistent with a premise.
(Here's an injunction committing this fallacy: "Stock up and save! Limit: one per customer."
Or: "Human life has absolute value. That's why the only punishment severe enough for a murderer is execution.")
- Fallacy of Begging the Question ("Circular Reasoning"): assuming the thing-to-be-proven as a (concealed) premise.
But the listener doesn't accept the conclusion yet, so it's not fair to sneak it in to "prove" itself!
(Sometimes referred to as a "loaded question.")
(E.g.: "God exists b/c the scriptures say so, & we can trust them, because they are divinely inspired.")
- Fallacy of Disappearing Hedge: making a conditional claim ("If x, then y." -- the "if" is the "hedge" or qualification), and then conveniently forgetting the "if" later on, so that the "hedge" has discreetly disappeared.
(E.g.: "If God chooses to manifest himself in the visions reportedly taking place at city such& such, then we should be grateful for receiving them. Given the current state of the world, and . . . Let us then give thanks for these authentic visions!")
(E.g., found by a former student: "If ever capital punishment were to be justified, this would be the case." Implied: "Therefore it's justified in this case.")
- Fallacy of "I'm Just Sayin' . . . ": in which the speaker is trying to "have it both ways": to get others to give weight and consideration to his statement (usually for rhetorical punch), without having that premise or conclusion abide by the usual standards of being relevant and plausible (often because it's not relevant or not plausible).
(But if the speaker really is just saying whatever, not offering it for rational consideration and not willing to have it discussed, then we can treat it as if he were just whistling or humming. I.e. we're just disregardin' . . .)
- Fallacy of just plain Non Sequitur: "doesn't follow." The claim/conclusion doesn't follow from, isn't proven by, the evidence/premises/reasons given.
The premises may be simply irrelevant, or may be insufficient to prove the claim due to weak induction, or unwarranted presumptions.
The gap in a Non Sequitur argument may be concealed by vagueness and ambiguity.
There are so many different kinds of non sequitur fallacies that the next four sections are devoted to them!
2. Non Sequitur FALLACIES OF RELEVANCE
(Here, the premise or premises are irrelevant for the conclusion.)
Appeal to Force: self-explanatory; why Stalin always won the office squabbles. (Also referred to as "scare tactics.")
Appeal to Pity: self-explanatory and similar to Special Pleading
("I'm special; the usual rules don't apply to me. I'm superior/delicate//important/unhappy/misguided/a celebrity. . .")
Related fallacies involve Appeal to Anger and Appeal to Desire, a.k.a. "wishful thinking."
(Appeal to desire e.g.: "Which would you rather claim as an ancestor: Adam or some monkey?")
Appeal ad Populum: to popularity, or directly to mob mentality, or indirectly to bandwagon or snob mentality, etc.
(Sometimes referred to as : "appeal to belief," "common practice," "peer pressure")
(E.g.: "Most people in the U.S. believe that God [the Judeo-Christian god, specifically] exists. Therefore God must exist."
Or "Five million people can't be wrong!" To which one can always respond: "Why not?")
Abusive Ad Hominem and Circumstantial Ad Hominem: attacking the arguer personally, rather than his/her argument. (The Rush Limbaugh specialty.)
(E.g.: "He smokes pot! Can't trust what he says.")
(Connected to: "Guilt by Association": self-explanatory.)
Poisoning the Well is a kind of preemptive Ad Hominem: undercutting one's opponent by setting things up so that anything he or she says will be dismissed as untrue or deceptive.
(E.g.: Nietzsche: "Those who disagree with me when I say that humanity is corrupt prove that they are already corrupted.")
Tu Quoque ("you, too"): accusing one's accuser (& of hypocrisy), rather than answering the accusation.
(Connected to: "Two Wrongs make a Right": tolerating a wrong b/c a similar wrong has already been tolerated.)
Straw Man: Attacking a weaker or distorted misrepresentation of someone's argument, not the actual argument. (It's much easier to win a fight with a straw man than with a real human!) (Another Limbaugh specialty.)
Fallacy of Accident: Misapplying a general rule to a case not intended to be covered.
(E.g. "But my right to free speech was violated! They wouldn't let me yell 'fire' in the crowded theatre!")
Missing the point: Drawing a different conclusion from the premises than is warranted.
Red Herring: A distraction. Also referred to as a "smokescreen."
3. Non Sequitur FALLACIES OF WEAK INDUCTION
Improper Appeal to Authority: accepting a conclusion only because a dubious "authority" does.
Here, the "authority" may be unqualified, or be a genuine authority but in another field, or may be simply unidentified.
(E.g.: "But Shirley McLaine believes in reincarnation, too.")
Argument from Ignorance (or "Negative Proof"): accepting a conclusion only because it hasn't been proven false yet.
(E.g.: "Dangerous and polluting dragons exist, because no one has proven they don't.")
Hasty Inductive Generalization: reaching conclusions prematurely, on too little evidence.
Fallacy of the Small Sample: self-explanatory.
Fallacy of the Biased Statistics (a non-representative sample)
Fallacy of Data of Different Quality: (E.g.: Has child abuse increased, or been reported more?)
- Post Hoc Ergo Propter Hoc: Assuming that b/c X precedes Y, X must have caused Y.
- Confusing Correlation & Cause: assuming that a correlation must be causal.
(E.g.: "Whenever I wash my car, it rains. The Hopis in Arizona have a drought? I'll just drive down there and wash my car!")
- (Causal) Slippery Slope: attacking a position by claiming w/o proof that it will cause a series leading to a bad outcome.
(E.g.: "If we let Viet Nam fall to the communists, Mexico is next, and then us.")
(Cf. Slippery Precedent: arguing against a case only b/c it might set a precedent for the acceptance of bad cases.
(E.g.: "I acknowledge that there were strong extenuating circumstances, but if we let him off easy, then the next guy we catch will probably get off unfairly!")
& Slippery Assimilation: arguing that the existence of any borderline cases means that no distinction can be drawn. This is similar to Moore and Parker's "Line drawing fallacy")
(E.g.: "If I just eat one cookie, that'll just be 3 grams of fat. My, that was good! Gee, if I had just another cookie, that would just be 3 more grams of fat, and 3 grams of fat isn't much difference. Yum! Well, if I had just one more cookie, that would only be an additional 3 grams of fat, and 3 grams isn't much. )
(E.g.: a disputable analogy: is a fetus like a helpless patient, or an appendix, or neither?)
Analogies have the following structure:
Premise: "The primary case X has the following similarities to case Y."
Conclusion: " Therefore, Y will also share X's characteristic of whatever."
In evaluating analogies, we can ask the following questions:
Are the similarities cited in the premise central ones or merely incidental?
Are the similarities relevant to the characteristic under discussion, the one in the conclusion?
Are there any relevant differences that undermine the analogy?
4. Non Sequitur FALLACIES OF PRESUMPTION
Fallacy of the Complex Question: presupposing a certain answer to an unasked question.
(Sometimes called a "loaded question.")
(E.g.: Prosecuting attorney to innocent defendant:
"So when did you begin to plan the fiendish crime you carried out?")
False Dichotomy Fallacy: prematurely limiting the alternatives to two.
(Also known as "false alternatives," since one could misleadingly cover only, say, three of four alternatives.)
(E.g.: "Either people are self-interested or they're altruistic. They are not altruistic, therefore they are utterly self interested."
Or e.g. "Check the box that applies to you: African American, or European American." )
("Perfectionist fallacy": "if X isn't perfect, we should reject it.")
Suppressed evidence: important evidence supporting a different conclusion is ignored.
(We want the whole truth, as well as nothing but the truth.)
(Suppressing evidence is very, very bad. Except for monkeys.)
5. Non Sequitur FALLACIES OF AMBIGUITY, VAGUENESS, GRAMMATICAL ANALOGY
Vagueness: using terms without enough precision to be able to determine a claim's truth or falsity.
(Including using relative terms without specifying what they are in relation to.)
(E.g. a phone call heard on talk radio, January, 1998:
"I think we [the U.S.] should just come down hard on Saddam [Hussein of Iraq]."
Host asks policy analyst who is the day's guest: "Do you agree with that?"
Policy analyst: "Well, what do you mean 'come down hard'?" Caller: "We should take him out!"
Policy analyst -- I'm not making this up!!! : "Well, I agree with that."
Was the policy analyst recommending assassination? One certainly hopes not, but can't be sure.)
[Note: I wrote this in 1998, back when this was surprising, rather than merely depressing.]
Equivocation: a term used ambiguously, and the conclusion turning on the different meanings.
(Puns are equivocations. "Only man is rational. No woman is a man. Therefore no woman is rational." )
Amphiboly: ambiguity of grammatical construction.
(E.g. Groucho: "I shot an elephant in my pajamas. How he squeezed into my pajamas I'll never know.")
Ambiguous Comparison: attributing a relative characteristic (e.g. "better") to one thing or event
without stating what it is relative to.
(So better than what? Often the comparison is simply left incomplete. "The quicker picker upper.")
(E.g.: "Our employees shouldn't complain. We're paying them more than we did last year!" More than what?
"Than their take home pay last year. Even though we cut their wages.
After all, we have mandatory overtime, and most of them are working 60 hours a week as opposed to 45 last year!")
Phantom Distinction: appearing to make a distinction that doesn't really exist. (E.g. "I'm not making a threat. I'm just letting you know my plans.")
Conflating: ignoring a distinction that does exist.
(It's like hammering two different things together that don't belong together.) (E.g. "Aw, open space, farming, new subdivisions, it's all the same thing.")
Fallacy of Accent: improper emphasis; often involves quoting things out of context.
(E.g.: Critic X: "If this was the best show in town, I still couldn't recommend it."
Ad quoting critic: "A real blockbuster! Critic X raves: ' . . . the best show in town . . .' ! ")
Fallacy of Composition: falsely attributing a feature of a part to its whole.
(E.g.: Each cookie is small. Therefore the whole bag isn't very high in calories!"
And its companion: Fallacy of Division: falsely attributing a feature of a whole to one of its parts.
(E.g.: The virtues of a state must come from the virtues of its members (Plato, Republic, 435e).)
6. Nasty persuasive rhetoric, a.k.a. SOPHISTRIES
(From B.N. Moore and R. Parker's Critical Thinking, 5th edition (Mayfield Publishing Co.)
Remember: We want to hear, and to put forward, the truth, the whole truth, and nothing but the truth.
Thus we want to avoid using these sophistries, and we want to identify them if anyone else is using them.
"Downplayers" /"hyperbole" or exaggeration
"Persuasive" (i.e. emotionally charged) definitions and comparisons
Innuendo (implications that the nasty sophist making them isn't willing to take responsibility for!
These include "weaselers," or qualifications that function to mislead.)
"Inappropriate burden of proof" (i.e. placing the burden of proof on one side against common sense)
(None of these fallacies applies to big orange cats. Outrageous complaints about them are usually perfectly accurate.)
7. OTHER FALLACIES
Some people have been raised to think that it is always impolite to disagree with others. In order to remain what they think is polite, these unfortunates have rendered themselves powerless to defend themselves and powerless to defend others, and even powerless to plan and dream for the future! This bizarre state of affairs has led to some equally bizarre consequences when people try to live both politely and with integrity. One of these is "the subjectivist fallacy." According to this fallacy, one can maintain that, say, "God exists" is "true for me," and still "is not true for you." In other words, God (one assumes the Judeo-Christian god is intended?) both does exist and does not exist!
And this circumlocution is utterly unnecessary. We can politely say to each other, "Well, I don't share your belief/lack of belief. But perhaps one day, we'll be able to come to an agreement." Indeed, reflective, considerate disagreement is a sign of respect!
- The Naturalistic Fallacy (from G.E. Moore, ultimately from David Hume):
Hume: values (evaluated), & facts (merely described), are completely different sorts of things, & can't be mixed.
(A Hegelian scholar once suggested to me that Hume held this position only because his epistemology allowed for knowledge only at the level of "sensation.")
Moore: one can't derive an "ought" from an "is." (Vs. Aristotle's practical syllogism.)
If one defines the word "right" (or "just") in naturalistic (non-moral, factual) terms, this happens:
E.g. utilitarians define the word "right" for any X as "X maximizes happiness."
But we can still intelligibly ask: "Is X really right?"
And if "right" were truly defined in this way, we couldn't intelligibly ask that, b/c it would be "tautologous," i.e. analogous to saying:
"X is 4. But is X really 4?" (which is unintelligible, i.e. it doesn't make sense).
- A related fallacy: that of mistaking a statistical norm, i.e. an average, with an evaluative norm, i.e. an ideal.
E.g.: "Most families in the U.S. have 2.3 children. Therefore it is good to have at least 2& no more than 3 children."
Or e.g. "It's normal to have an IQ of 100. Thus you are over-cranial if you have an IQ of 120."
Mistaking a thing's origins with its import, justification.
(E.g. sociologists who want to reduce philosophy to its societal causes.
Or a possible attack on vegetarianism:
"People who become vegetarians usually do so b/c they were neurotic, b/c they didn't have good enough ego boundaries when they were kids. So we don't have to take vegetarianism seriously."
But even if this dubious claim were true, it wouldn't be relevant to the moral question.)
- The Fallacy of Linearity:
Assuming that if two things differ in any respect, one must be superior; i.e. assuming that a comparison can always be made (even between apples & oranges).
(Could be described as refusing to acknowledge qualitative, "incommensurable" differences, and insisting on collapsing all differences into quantitative ones, i.e measurable on a single scale.)(from Abraham Kaplan, "Human Relations& Human Rights in Judaism.")
- The Fallacy of Misplaced Concreteness:
Mistaking our own abstractions (e.g. "substance" & "quality") for concrete things.
Similar to mistaking processes for things (hypostasis), & to ignoring relations & interactions which
may influence their participants.
This led, A.N. Whitehead charges, to the distinction of primary & secondary "qualities" in Locke,
& ultimately to a harmful dualism & a view of nature as meaningless in itself, & of values as merely subjective.
(From A.N. Whitehead's Science& the Modern World, Chap. 3.)
- The Intentional Fallacy (in art theory):
An author's intention, though it is the origin of the artwork, shouldn't be taken as a standard in interpreting, analyzing, or evaluating the work.
One shouldn't go to "private" evidence, external to the artwork (e.g. diaries), but should restrict one's evidence to the "public" evidence actually within the work.
(This has been challenged as involving an undesirable "internalist" picture of intention.) (from the art theory of Wimsatt & Beardsley.)
- The Fallacy of Imitative Form (also in art theory):
If one is trying to satirize, say, vapid society chitchat, one can't simply write down vapid society chitchat, i.e. imitating it perfectly. There must be some distinction to set one's satire apart as satire.
8. A few notes on statistical and hypothetical reasoning:
- What is to be measured must be defined precisely, to avoid ambiguity or equivocation. Otherwise, we may have a test which fails on validity, i.e. the degree to which a test measures what it is claimed to measure. (Suppose we were trying to measure the overall health of a random group of people by how well they played basketball. We would have an invalid test. All the non-basketball players and short people would test inaccurately low, and all the experienced players and tall people might test inaccurately high.)
- This is distinct from a test's reliability, i.e. the degree to which a test consistently measures what it measures. (We decide to test people's overall health by asking them, verbally, how they are feeling that day. Such a test might be valid, but still have low reliability. Some people might keep stoic and not complain, even if feeling poorly; others might exaggerate.)
- If one is looking for correlations or possible causal relations between two factors or variables, one must try to control for any other possible variables that might affect the outcome of the test or experiment. (I heard once of an odd correlation that had been found: longevity was correlated with less sleep than average. In other words, people who sleep less than average tend to live longer than average. Can we conclude that we all need to cut back on our 8 hours? No! A third factor, good health, probably causes both the longevity and the diminished need for sleep.)
- Sampling: One aims for a sample that is representative of, generalizable to, a larger target population. One's sample should be:
- randomly selected, and
- large enough in relation to the size of the total population.
(If I decided to carry out an informal poll, and asked each of my online students or readers which TV/radio etc. show were your favorites, I couldn't generalize the results to the U.S. population as a whole.)
- What does "average" mean? It can mean different things.
The mean is the arithmetical average (i.e. from dividing the sum by the number of subjects).
The median is the centrally located datum (if the data is arranged in order, it's halfway down the list)
The mode is the number that occurs most frequently.
- Dispersion: To represent the data accurately, these numbers may need to be accompanied by information about:
- range (the space between the largest and smallest number) and
- standard deviation (roughly speaking, the average distance of the data from the mean).
- "Standard error of measurement": the expected frequency of errors. (A small s.e.m. indicates greater reliability.)
- "correlation coefficient": a number between 0 and 1.0, or between 0 and -1.0, which indicates the degree to which two variables are correlated.
- Some hypotheses are more "empirical" than others, which are more "theoretical."
- The "Empirical" hypotheses involve events or objects or processes that can be observed more or less directly, so that the observation is minimally theory-laden. These can be regarded as for all practical purposes confirmable (or more correctly disconfirmable). (E.g. the discovery of a new element.)
- The more "theoretical" hypotheses involve ways of conceptualizing objects or events or processes. These involve less direct observations, that are more theory-dependent. These can't attain the high degree of confirmation of the more "empirical" hypotheses. (E.g. is time a unidirectional, asymmetrical "arrow," or a symmetrical "fourth dimension of space"?)
- Criteria for evaluating hypotheses include:
Adequacy: the degree to which a hypothesis can accommodate all the data to be addressed, so that a given object or event can be interpreted in terms of the hypothesis.
Internal coherence, i.e. simplicity: the degree to which the hypothesis organizes and unifies the data and their interconnections.
External consistency: the degree to which it is consistent with external, or other, well-confirmed hypotheses.
Fruitfulness: the degree to which it generates ideas for future developments, analyses, confirmations, etc..
If a hypothesis isn't fully adequate or coherent or consistent with what we already know, its supporters may tack on little revisions designed to meet these objections, but which don't cohere with the hypothesis. These are called ad hoc (aimed at this thing) additions. They function as excuses for the hypothesis's failure. Thus, if there are many ad hoc elements in a hypothesis, the hypothesis won't be as plausible.
10. Always keep an eye out for the unexpected! That's what makes science so exciting.