Fallacies Not Fallacious: Not!

Author(s): Jonathan E. Adler

Source: Philosophy & Rhetoric, Vol. 30, No. 4 (1997), pp. 333-350
Published by: Penn State University Press
Stable URL: http://www.jstor.org/stable/40237963 .
Accessed: 31/12/2010 08:19

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=psup. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Penn State University Press is collaborating with JSTOR to digitize, preserve and extend access to Philosophy
& Rhetoric.

http://www.jstor.org
Fallacies Not Fallacious: Not!1

Jonathan E. Adler

The currently dominant view of fallacies is that instances of a fallacy

'
type are not always fallacious (the "Fnotf thesis, for short). In a
récent collection on fallacies, the editors observe approvingly that

no simplistic formula can capture the différence between, say, a falla-

cious appeal to authority and one that is not fallacious. . . . One must
trace or illuminate the subtle différences between dialectical or argu-
mentative moves that are fallacious or objectionable and superficially
similar moves that are not objectionable. (Hansen and Pinto 1995, 198)

Toulmin, Rieke, and Janik in their textbook write, "Arguments that

are fallacious in one context may turn out to be sound in another
context. Therefore, we shall not be able to identify any intrinsically
fallacious forms of argument" (1984, 131; for exulting approvai, see
Willard 1995, 158, and 1989, 226).
The Fnotf thesis not only is widely endorsed, but is a starting
point for many extended inquiries. (Test thèse claims by sélection
through the bibliographies in Hansen [1990] and Hansen and Pinto
[1995]). These inquires aim to delineate settings in which instances
of a fallacy are fallacious and those in which they are not. Walton,
the leading researcher on fallacies, writes,

A fallacy is a technique of argumentationthat may in principle be rea-

sonable but that has been misused in a given case in such a way that it
goes strongly against or hinders the goals of dialogue. . . . The problem
then is one of distinguishing between the fallacious and nonfallacious
instances of use. (1992, 18; see also 1996, 270-7 1)2

In this paper, I criticize four reasons thought to support or imply

the Fnotf thesis,3 without defending any traditional list of fallacies.
Indeed, Hamblin's Fallacies (1970), which contains the définition of
fallacy that I broadly follow, is a harsh and sustained critique of the
fallacies tradition.

Philosophy and Rhetoric, Vol. 30, No. 4, 1997. Copyright © 1997 The Pennsylvania
State University, University Park, PA.

333
334 E. ADLER
JONATHAN

The fallacy that I refer to most often is thè ad hominem, which,

for reasons that will emerge, promises thè best case for thè Fnotf
thesis. It is probably the one that those who hold to the Fnotf thesis
most often cite. But even if their attention extends to ail fallacies ad,
the skewing is suspicious. If the thesis is generai, why shouldn't their
examples be spread more evenly to embrace formai, scientific, and
statistical fallacies?
The suspicion festers. A crucial différence among types of falla-
cies is over limits in how recognizable to a sincere argueris the struc-
ture of his or her argument,when it is a fallacy. The différences follow
upon thè assumption that a sincere arguermust believe that the argu-
ment he or she puts forward commits no fallacy. One can no more
recommend "S is a good argument, though a fallacy," than one can
assert, as Moore's Paradoxshows, "/?,but I do not believe it."Roughly,
first, appeals ad are readily recognizable as appeals (though not as
premises for the truth of a claim). Second, other fallacies, especially
slippery slope arguments and arguments from ignorance, are recog-
nizable for their structure. The structure is not believed defective.4
(In the argument of William James's discussed below, section 5, it is
not only an illustration of an argumentfrom ignorance, but a défense
of such arguments.) But, third, formai fallacies are not recognizable
as having their defective structure.
Now if the Fnotf thesis is correct, then the structure of
nonfallacious instances of any of thèse fallacy types should be
unproblematically recognizable by arguers. But it is only in the first
and second category that we plausibly find such récognition. Accord-
ingly, the fallacies I consider are limited to the informai ones, unless
explicitly stated otherwise.
As a working account, I rely upon the following broadening of
Hamblin's présentation:An argumentis a fallacy when it seems to be
good, though this apparentgoodness masks defects in its underlying
form or structure (1970, 12). But I am not satisfìed with this pro-
posai, and I rely on it only for heuristic purposes. For one problem,
what are we to say of instances of a defective form that are not per-
suasive? My inclination is to hold that such arguments are fallacies
because their form gives rise to instances that tend to be persuasive
(contrast with Powers 1995). But matters of définition or classifica-
tion are to be justified pragmatically, by their fruitfulness.
Since this paper's main point is criticai, little is said in support
of any contrary view to the Fnotf thesis, except as asides and by con-
trast. The working définition, though, does suggest the view that I do
 FALLACIES
NOTFALLACIOUS:
NOT! 335

favor,which is thatfallacies should be characterizedstructurally.Since

a fallacy always suffers a structural defect, it follows that to offer
such an argument is always to be arguing fallaciously.
The proper notion of structureor form is much broader than the
notion of logicai structure or form. Whenever we distinguish in an
inference pattern between constant éléments and variables, open to
Substitution, where the inference turns on the pattern of thèse con-
stantéléments, and not the substitutionsfor the variables,we are speci-
fying a structure or form (Brandom 1988). Additionally, the pattern
must yield a rieh set of inferences. On this conception, critieizing
some argumentsfor the falsity of a premise, when it expresses a rieh,
structuralpattern, does constitute the finding of a defect in form.

II

Primarily, four reasons encourage, and appear to support, the Fnotf

thesis, though the first and second are usually conflated. The first is
that certain traditional ways of characterizing a fallacy allow subcat-
egories that are not defective. The second reason, with which the first
is confused, is that there are instances of fallacies in which the pre-
mises do support the conclusion.
An example that illustrâtes the first reason is the abusive and the
circumstantial ad hominems. The former is "the fallacy of rejeeting
the Claimsa person advances simply on the basis of derogatory facts
(real or alleged) about the person making the claim" (Toulmin, Rieke,
and Janik 1984, 144; original in italics). The latter, in Locke's words,
is "to press a Man with Conséquences drawnfrom his own Principles,
or Concessions" (1975, 686; for good discussion of this thème, see
Walton 1985).Whateversimilarities there are between these two kinds
of arguments, their structuraldifférences are sharp. The former rep-
resents inferences from personal characteristics or properties of an
arguer to the quality of his or her argument. The latter is an attempt
to show an internai flaw, especially inconsistency, in the joint set of
an arguer's commitments. That instances of the former are typically
bad, while the latter typically good, does not imply that the same
structureis good and bad.
Another case is standardly exemplified by affirming the consé-
quent. Such arguments often are held to be, though deductively in-
valid, inductively strong. Consider the following example from van
Eemeren and Grootendorst:
336 E. ADLER
JONATHAN

If it snows, then the roofs turn white.

The roofs are white.
Therefore it has snowed. (1992, 181)
Biro and Siegel comment on this example that "good inductive argu-
ments often involve affirming the conséquent" (forthcoming, 24).
All that the example reveals, however, is that the same words can
instantiate two différent forms of argument.Affirming the conséquent
has two premises, and it purports to be a valid deductive argument.
The suggested good argument is inductive (or probabilistic). It re-
quires additional premises, such as that the roofs turning white calls
for explanation and that no other explanation of why the roofs are
white is acceptable.5
So, there is not one argument hère of two forms. Rather, there
are two arguments in the same words, analogous to, though less ap-
parent than, when we hâve an ambiguous sentence such as "Flying
planes can be dangerous."
So this first reason I grant, though only as distinguished from the
second. It does not involve categorizing, within a fallacy type, falla-
cious and nonfallacious instances. Rather, it justifies a disambigua-
tion of an overly loose characterizationof the fallacy type in question.
This first reason is, then, simply a matter of getting it right.
Admittedly, the criticism of this first reason is not neutral. It as-
sumes a structuralview. In such a view, différent structure,différent
argument.Those who maintain the Fnotf thesis characterize fallacies
in looser terms. They can then trace the same argumentthrough dif-
férent contexts, even when thèse contexts generate arguments with
différent numbers of essential premises. In one field, such as law, an
appeal to précèdent may back an argument while in another field,
such as science, it may not. Structurally,however, the différence in
backing marksout différent arguments,even if expressible in the same
or similar words.
Although the criticism is not neutral, it is, nevertheless, capable
of supporting a structuralview. For by taking that view, distinctions
are illuminated that are otherwise overlooked, blunted, or conflated.
Consider again the first example of the distinction between abusive
and circumstantial ad hominems. It may be claimed that this is a mis-
leading example, since it is too easy. Defenders of the Fnotf thesis
accept it. Still, how sharply can they distinguish thèse? The two types
of argument are both instances of arguments directed to a person;
some of either kind tend to hinder, while others do not, the goals of
dialogue.
 FALLACIES
NOTFALLACIOUS:
NOT! 337

III
The second and most important reason thought to favor thè Fnotf
view is that some fallacious argumentsdo provide reasons to believe
their conclusions. The most well-known such cases are (noncircum-
stantial) ad hominem arguments.
Ad hominem arguments are held not to constitute fallacies be-
cause it is often legitimate to attack thè credibility of arguers and
sometimes, as in cross-examination in courts, it is virtually a manda-
tory strategy.Hamblin quotes Joseph: "A barristerwho meets thè tes-
timony of a hostile witness by proving that thè witness is a notorious
thief, though he does less well than if he could disprove his évidence
directly, may reasonably be considered to hâve shaken it; for a man's
character bears on his credibility" (Hamblin 1970, 42, quoting Jo-
seph 1916, 591). In one of many works building on this point, Hinman
Claimsthat questioning the reliability of an arguer, as a basis to cast
doubt upon the truth of his premises, is relevant, for "it does affect
our assessment of the soundness of a deductive argument or of the
strengthof an inductive one" (1982, 339). Similarly, Brinton, defend-
ing Aristotle's Claims in thè Rhetoric,6takes ad hominem arguments
to be typified by the following: "Candidate Jones has no right to
moralize about the family, since he cheats on his wife" (1995, 215).
Restricted to X's-arguing-for-/?,ratherthan the truth of itself, they
are legitimate: "The ordinary ad hominem may in generai be under-
stood to be an assault on the rhetoricalethos of a speakeror writer.. . .
It is a legitimate form of argumentand is logically acceptable in many,
perhaps most, of its actual occurrences" (222).
Works that argue for or assume the Fnotf thesis in regard to a
spécifie fallacy standardlytake the form just suggested. Fallacy type
A is not fallacious in such-and-such cases because the premises do
render the conclusion more crédible or worthy of belief. Therefore,
fallacy A cannot generally be fallacious and so, if this patterncan be
repeated for any type of fallacy, fallacies are not generally fallacious
(the Fnotf thesis).
This standard argument turns on endorsement, usually tacit, of
an epistemic analysis of arguments. An argument is nonfallacious if
its premises increase the degree of belief in the conclusion. A good
argument will justify füll belief in the conclusion. In the most ex-
tended défense of this position, Feldman writes, "Good arguments
are arguments that provide people with reason to believe their con-
clusions" (1994, 165).
338 JONATHAN
E. ADLER

I cannot hère address this conception in détail, nor offer an alter-

native analysis of good argument. Aside from reasons of space, my
focus is on fallacies.
However, my criticai remarks will indicate an alternative. The
key assumptions implied by thè standardepistemic analysis subject
to my criticisms are thèse: First, supporting or justifying or render-
ing more reasonable to believe a conclusion is nonfactive. Unlike
knowledge, X's being justified in believing that/? does not entail that
is true. Second, if thè premises render/? more probable, then those
premises constitute reasons that help establish thè truth of thè con-
clusion. Third, being a good argument is relative, depending upon
whether premises continue to yield reasonable belief in thè conclu-
sion.
Each of these assumptions (nonfacticity, probabilism, relativism)
has a legitimate place. But that piace is not as an analysis of good
argument. As a slogan, a good argument aims at knowledge; in thè
standardepistemic analysis, it aims at reasonable belief.
In such a revised epistemic analysis, justification must be factive
(i.e., that thè premises justify thè conclusion implies thè truth of thè
conclusion) and nonrelative (implied by facticity) and they must draw
a distinction between reasons that render thè conclusion more prob-
able and good reasons (or évidence) that bear a further (epistemic)
relation to that conclusion.7 By élimination, I indicate thè reasons
favoring a revised epistemic approach, although my criticisms stand
on their own.
The criticisms of thè standard epistemic analysis are three, thè
latter two being most directly relevant to thè Fnotf thesis. First, thè
claim that good argument is relative is counterintuitive. For a long
time premises of thè following kind were reasonable to believe and
transmittedthat reasonable belief to their conclusion:
It was an empirical discovery that thè Morning Star and thè
Evening Star was Venus.
So, it is a contingent truth that thè Morning Star is thè Evening
Star.
One of Kripke's (1980) important fìndings was that this argu-
ment is not cogent. This finding led us to view thè argument as hav-
ing a missing premise that what is known a posteriori is contingent,
an assumption that he successfully questioned.8
In this analysis, thè argument was once good, but now is bad.
Nevertheless, I would hold (and assume that this is thè intuitive ver-
 NOTFALLACIOUS:
FALLACIES NOT! 339

dict) that if Kripke is correct, this argument is not good and never
was. It involves a crucial false assumption. The naturai way to ex-
press this verdict is that earlier we had reason to believe the argu-
ment good, though it wasn't.
The second criticism is spécifie to epistemic analyses of falla-
cies. Biro and Siegel write, "The crucial question is whether the pre-
mises of the argumentin question provide warrantfor the conclusion.
What an epistemic account maintains is that their failing to do so is
both a necessary and a sufficient condition of fallaciousness" (forth-
coming, 7; see also 1992).
But the analysis fails to distinguish between a fallacious argu-
ment and a merely weak one. A fallacious argument implies one that
cannot be built upon. An argumentis built upon, ratherthan replaced,
only if the resulting argumentretains the original structure.No mere
supplementationof an argumentthat affirms the conséquent can, trivi-
ally, both render it valid and retain its form as affirming the consé-
quent.
Not so a weak argument. If I hâve examined too small a portion
of ravens to warrant the belief that ail ravens are black, then I am
directed to supplément, ratherthan abandon, my premises. I improve
the argument simply by finding more black ravens. (The "fallacy of
hasty generalization" is not then truly so called, when it is solely due
to having gathered too few instances.)9
In reducing the notion of fallacious argumentmerely to one of a
failure of warrant,epistemic analyses rob the accusation of a fallacy
of its preemptive rôle (compared, say, to assessments of soundness or
the truth of the premises) in criticism. If the structure of the argu-
ment is no good, the direction or form of reasoning is to be rejected,
not supplemented (while if it is only weak, supplementation alone is
required). This crucial advantage justifies the directive to look first
for a defect in the form of an argument. Only if the form is cogent
should one proceed to assess the truth of the premises.
The third criticism of epistemic analyses is directed against their
probabilism. Propositions that increase rational subjective probabil-
ity (or partial belief) are not thereby reasons to establish the truth of
the conclusion. This thirdcriticism mainly focuses on how probabilism
fails to capture the distinction between judgments of credibility and
judgments of probative force, since each bears on partial belief. In
the example from Joseph, the challenge to the credibility of the ar-
guer has no bearing on the probative question, as to, if the claim is
accepted, how strongly it supports the conclusion. (The credibility-
340 JONATHAN
E. ADLER

probative force distinction parallels thè fundamental distinction be-

tween soundness and cogency.)
For another case: if I am sent a letter of recommendation for a
candidate for a position in my department, my single judgment is
based on two distinguishable éléments- the strength of the recom-
mendation and the credibility (e.g., réputation) of the recommender.
Although both must be integrated to yield an overall rating of the
candidate, their respective contributions are very différent. One bears
on the (degree of) admissibility or worth of a relevant datum, the
other on its strength of praise.
The conclusions can be expressed in terms of an idiosyncratic
version of the distinction between the context of discovery and the
context of justification. The stage of justification (or proof) aims to
réfute or establish a claim; the stage of discovery aims to determine
whether it is worthwhile to examine a claim (or argument), or to find
relevant reasons. The distinction as so drawn is epistemically, not
pragmatically, based. Reasons are being directed toward différent
judgments (truth,worthiness of investigation). The latter stage is nec-
essary, as it is in science, because our time and resources are very
limited and the number of arguments (hypothèses) for claims (phe-
nomena) we are interested in is staggering.
Argument as a mode of inquiry, as contrasted to persuasion or
reasonable belief or practical judgment, aims to be a contribution to
the stage of justification. Credibility bears not on the stage of justifi-
cation, but only on discovery.
The position Brinton advocates suffers for the absence of this
distinction. Conclusions as to the ethos of an arguercan bear only on
discovery, not on justification. If the arguerlacks credibility, then we
hâve reason not to admit his argument for serious justificatory argu-
ment (or proof), particularlyif our time or resources are very limited.
However, once an arguer(or argument)is admitted, such ad hom-
inem arguments are ruled out. First, they are irrelevant to establish-
ing or refuting the arguer's conclusion. Second, they would distract
focus from the argument onto the arguer.Third, they would threaten
autonomy, undermining individuate' (Citizens') rights to défend and
criticize positions in public forums. Although the basis for the exclu-
sion is ethical, it, too, originates in an epistemic considération, paral-
lel to that in science. The process of inquiry undermines its claim to
objectivity if the range of views is constricted. If so, we can be less
assured that judgments stem from fair assessments ratherthan bias.
 FALLACIES
NOTFALLACIOUS:
NOT! 34 1

Common advice in the ethics of argument is to avoid (abusive)

ad hominem appeals. It is the kind of advice that defenders of the
Fnotf thesis tend to reject, and that they must reject, insofar as they
endorse this second reason.
It is good, if imperfect, advice. Although the stages of discovery
and justification are analytically distinguishable, in ongoing argu-
ment the distinction is obscured. Consequently, even good ad
hominems would infect the stage of justification and damage it in the
three ways just noted. In particular, we are likely to treat negative
results of discovery, as to thè dubious worth of hearing an argument,
as réfutations in justification. So it is good advice, though
overridable,10not to engage in ad hominem arguments,but to respond
to the content of an argument.
The overall conclusion of this section is the following. Let us
grant that a set of premises increases reasonable belief or subjective
probability that an individuali argument is likely to be distorted or
biased or otherwise fail. Still, the error in this second reason is to
conclude that, in the stage of justification, the argument is refuted
(lacks cogency, its conclusion false). (Similarly, for the converse case,
to increase reasonable belief that an individuars argument is likely
to be good is never adequate to establish that argumentasconclusion.)
The more generai point is this: probabilistic relevance is not suffi-
cient for evidential relevance. I use evidential relevance as shorthand
for the relation that any statement that could serve as a reason or
premise in a (minimal) argumentto establish a conclusion^ truthbears
to that conclusion. As a final illustration, I would express thè usuai
reading of the Euthyphroas follows: if an act is dear to the gods, then
the rational subjective probability that it is right is increased, even
becoming, let us concede, maximal. Nevertheless, that the act is dear
to the gods cannot serve as a premise in a proof that the act is right.

IV

The third reason is that in many cases fallacy-looking contributions

are not arguments, and so cannot be fallacies. This reason offers little
support for the Fnotf thesis, and it does a disservice to the rôle of
fallacies in diagnosing faults in reasoning.
The wrong turn Startswith Hamblin. Soon after his initial défini-
tion of a fallacious argument, he writes, "A fallacy, we must repeat,
is an invalid argument" (1970, 39; italics in original; see also pp. 1,
12). At that point, he dismisses the "many questions fallacy" because
342 JONATHAN
E. ADLER

a question is not an argument. Again, at thè start of chapter 7, "The

Concept of Argument,"Hamblin introduces the problem of nailing a
fallacy, following upon his assertion that "A fallacy is a fallacious
argument":"In many cases of supposée fallacy it is possible for the
alleged perpetratorto protest, with an innocent face, that he cannot
be convicted because he has not been arguing at ail" (224). n His ex-
ample is the argumentumad hominem:
PersonA makesstatementS: person says 'It was C whotold you
that,andI happento knowthathis mother-in-lawis living in sin witha
Russian':A objects, 'The falsity of S does not follow from any facts
aboutthe moralsof C's mother-in-law: thatis an argumentumad hom-
inem': may reply did not claim thatit followed. I simplymadea
remarkaboutincidentalsof the statement'shistory.Drawwhatconclu-
sionyou like ... If the capfits . . . ' This wouldbe disingenuous,butthe
pointremainsthat cannotbe convictedof fallacyuntilhe can hâvean
argumentpinnedon him. (224-25)
Although he takes his two latterdéfinitions as merely verbal varia-
tions on the opening one, there is a telling différence. The opening
définition is of afallacious argument, thèse later ones are définitions
of fallacy itself. So thèse définitions, unlike the opening one, commit
Hamblin to the claim that ail fallacies are arguments.
Being an argument implies being an actual linguistic contribu-
tion. But the argumentis only the public or visible aspect of what the
accusation of a fallacy is about. It is about a failure in reasoning,
which refers to the steps and inferences thinkers actually go through.
In so focusing on the public or visible (or verifiable) facet of
argument, Hamblin obscures the potential for fallacy in the actual
reasoning. The problem is manifest in Hamblin's very problem of
nailing the ad hominem, for what is this problem but the finding of
grounds sufficient for attributing the fallacy to that agent.12Without
the grounds, Hamblin holds, there is no fallacy. The switch from de-
fining a fallacious argumentto defining fallacies as argumentsfacili-
tâtes the displacement of the problem of the nature of fallacies for
the différent, epistemic problem of justifying the attribution.
Arguments ad characteristically work upon reasoning or belief,
bypassing verifiable argument. It is easy to do so because, first, we
cannot fully control the influences on belief - and much of the con-
nections in real argument are left tacit, for hearers to fili in- and,
second, reasons and reasoning serve as unifiers or connectors of dis-
course, and so as a main way to satisfy conversational expectations
of truthfulness and relevance.
 FALLACIES
NOTFALLACIOUS:
NOT! 343

So, appeals ad readily influence argument, without constituting

an argument. It is precisely B's point, in Hamblin's example, to have
the ad hominem comment influence thè judgment of his audience
without hearing responsibility for it. In this case, as with popularity
or force, thèse appeals are not counted as reasons for the conclusion
being argued. But they are, nevertheless, (unacknowledged) causes
affecting, and so distorting, the reasons we treatas relevantand sound.
The support the Fnotf thesis dérives from appeals (ad) not being ar-
guments is at the cost of ignoring their powerful role in affecting
reasoning.
Once the distinction between the actual nature of the reasoning
and the justification of its attribution is kept in mind, it is clear that
our failure to nail the fallacy does not imply that that very fallacy is
not realized in the speaker's reasoning (see Adler 1994) Consequently,
the Fnotf thesis can, at most, derive support from this third reason in
respect of our attribution of a fallacy, not for the matter of whether
the reasoning going on is fallacious.
This third reason reveals a bad bias resulting from the dominant
concern with argument over reasoning. Reflecting on how blatantly
fallacious fallacies ad are as arguments, it appears implausible that
they could "trick anybody,"as it appears to Woods for the ad bacu-
lum (Woods 1995, 241). But it is decidedly less implausible if we
take a participantperspective, not the misleading omniscient perspec-
tive, as reflectors upon the fallacy. In normal argument, fallacies are
interwoven with much eise, and so may be effectively hidden by it.
Moreover, even when the fallacy is fairly stark when presented as
argument it can still be, as previously mentioned, an influence on
reasoning. Argument requires a set of articulated assertions, and so it
is much more explicit in its assumptions, premises, and inferences
than reasoning itself. But it is when largely implicit that appeals ad,
as well as other fallacies, are most effective.

The fourth reason is that often appeals within an argument, which is

an instance of a fallacy type, are "relevant,""appropriate,""accept-
able," "reasonable,"or "justified."My criticism is that such relevance
is not sufficient for refuting an attribution of a fallacy. The brazen
and exploitable vagueness of these terms is a major source of unclar-
ity that encourages the Fnotf thesis. (This is not an objection to the
importance of these terms for understanding argument, let alone an
344 JONATHAN
E. ADLER

objection to vagueness itself. At the very least, however, these terms

must be acknowledged as purpose relative, for example, is relevant
for purpose R.)
In the example above from Brinton and those many other ex-
amples of "legitimate" fallacies, especially the ones ad, the Claims
that these arguments are fallacious seem inconsistent. The seeming
inconsistency involves the assumption that
an argument cannot be fallacious and acceptable, appropriateto
offer, or relevant.
What sustains this fourth reason is the evaluative classification
of fallacies ad as fallacies of irrelevance. The classification is too
broad, an unsurprising fault given the indefiniteness of "relevance."
One kind of irrelevance (cali it "discourse irrelevance") is irrel-
evance within an argumentative piece of discourse. It is rare for a
significant portion of a cohérent piece of argumentativediscourse to
be discourse irrelevant. Discourse relevance can be secured merely
by being a helpful illustration. (However, there is a narrowernotion
of discourse relevance, "judgment relevance," that is, relevance to
audiences' or reasoners' judging an argument or position appropri-
ately. Sound appeals to pity or popularity are judgment relevant.)
Second, there is probabilistic irrelevance. In rejecting (section 3,
above) a probabilistic account of good reasons in the stage of justifi-
cation, I have implicitly rejected this notion of irrelevance as unsatis-
factory for picking out the defects in fallacies ad.
Third, there is evidential irrelevance. This is the spécifie failing
of arguments ad (appeals to . . .). Ad hominems show that evidential
irrelevance need not be irrelevance in the previous two ways. Evi-
dential relevance entails, but is not entailed by, probabilisticrelevance.
The Fnotf thesis seems to gain support according to this fourth rea-
son because the first two kinds of irrelevance, especially the first,
have been confused with evidential relevance. Popularity, personal
characteristics, pity, or authoritycan ali provide reasons that rendera
conclusion or argument more reasonable to believe. But they do not
thereby serve as évidence that it is.
However, clarification of what counts as genuine appeals or ar-
guments ad is necessary. These do not turn on a premise claiming
reliability. If one appeals to an authority by citing that authority's
reliability on the topic, there is evidential relevance. This is not argu-
ment by authority,but only straightforwardprobabilistic or induetive
argument.
 FALLACIES
NOTFALLACIOUS:
NOT! 345

My repeated complaint about thè Fnotf thesis is that it obscures

valuable distinctions. But with fallacies ad we can expose an opposed
failure to unify. The fallacies ad fall under thè unifying description
of reasons that influence or cause one's judgment of thè truth of a
conclusion without constituting reasons or évidence for that conclu-
sion. Locke, in rejecting three arguments ad, présupposes this kind
of unity when he says of thèse appeals that they "may dispose me,
perhaps, for thè reception of Truth,but helps me not to it; That must
come from Proofs, and Arguments, and Light arising from the nature
of things themselves, and not from Shamefacedness, Ignorance, or
Errour"(1975, 687).
The fault in the assumption that being a fallacy is incompatible
with being acceptable to argumentativediscourse is just that fallacy
refers to a defect in the reasoning. The latter embrace all the func-
tions of such discourse. Just as, after speech act theory, it is a banal-
ity that intelligible utterances can serve more functions than
describing, so it is a banality that a contribution, containing an argu-
ment, will hâve other purposes besides arguing. It will aim to be per-
suasive, engaging, interesting, informative, and so on. Scant few will
read an argumentativeessay that has its arguments in the form of a
reconstruction. In a reconstruction, there is an ordered and numbered
list of premises and terminology is homogenized, répétitive, and shorn
of any nonargumentativelyrelevant features. It is followed by a con-
clusion, with the connections intended between premises and con-
clusion stated explicitly (. . . deductively implies ).
Since one must satisfy many purposes any contribution, even
in
an overtly argumentative one, there will be many features of it that
are relevant, appropriate,acceptable, and so on to only some of thèse
purposes. They may be relevant and so on to nonargumentativepur-
poses but as reasons to establish the conclusion weak, irrelevant, or
false.
Take the argumentfrom ignorance. Exemplifying a typical fault
in the Fnotf thesis, three arguments hâve been located under this la-
bel. Only one of thèse is, however, fallacious, and clearly of a dis-
tinct structure. (For récent writings that unite the three, see Walton
1996 and Krabbe 1995.) Usually it is represented as involving one
premise and a conclusion: not-proven that not-p/So, p. Actually, how-
ever, the form that I find most pervasive in contemporarypublic dis-
course is this: There is (can be) no proof that is false./So, it is possible
that is true. /So, it is reasonable (permissible) to believe that p. For
example, there is (can be) no proof that creationism is false, so it is
346 E. ADLER
JONATHAN

possible that creationism is correct. So it is reasonable (permissible)

to believe (teach) creationism.
We all know of an instance of this fallacy, which must be accept-
able, appropriate,relevant, and so on for most purposes of communi-
cation and argument,since it has been engaged seriously and fruitfully
as an argument.This is James's (195 1) famed argumentthat it is some-
times permissible to will to believe where évidence cannot decide an
issue. (James also requires that not to decide is effectively to decide
and that the matter for décision is "living" and "momentous." But
thèse qualifications do not affect our discussion.)
Most of James's reasoning is badly sketchy, but it seems to in-
volve at least the following: Since the belief that/? (e.g., God exists)
is not subject to evidential proof or réfutation, one cannot be com-
mitting an epistemic wrong by believing that p. If one would not be
committing an epistemic wrong by believing thatp, then it is permis-
sible to believe that p.
The argument suffers three defects characteristic of arguments
from ignorance (of this kind). First, its presupposition of evidential
indeterminacy is dubious. So, too, then the corollary that the truthof
is possible. Second, James takes the proper standardsfor the legiti-
macy of some beliefs to be the absence of decisive (counter-) évi-
dence. However, when one believes a statementtrue, one must at least
believe it more probable than any competitor, which is not secured
by the absence of decisive évidence against. Third, one's décision
cannot itself be a basis for the truth of what is believed. One's "will-
ingness to believe" that, for example, "God exists" does not bear on
its truth-conditions, and so cannot serve as adequate évidence. But it
is precisely such évidence that is necessary for one's coming to be-
lieve that in füll awareness.
If James's argument is an argumentumad ignorantiam, it shows
that a fallacy can be committed while it advances "the goals of the
dialogue" (contrast Walton 1992, 18). James's argument, however
strongly criticized, is still regarded as an important contribution to
arguments concerning the rationality of religious (and ethical) be-
lief. It is discourse relevant, perhaps even judgment relevant. But it is
still neither probabilistically nor, especially, evidentially relevant.
James's argument is appropriate to offer, despite committing a
fallacy. It commits a prima facie wrong, since it is a defective argu-
ment. I understand rimafacie along the lines of Ross's analysis of
primafacie duties or obligations. A primafacie duty is a real duty, so
 FALLACIES
NOTFALLACIOUS:
NOT! 347

that to violate it involves a wrong, but it is a wrong that may be over-

ridden (Ross 1980; for problems, see Searle 1978).
In committing a fallacy, an argument is, thereby, involved in a
wrong (i.e., a defective argument), though, for a number of further
reasons, the argument may still be acceptable or admissible. (Com-
pareto lying. There are circumstancesin which a lie is justified. None-
theless, it does not follow that no wrong has been committed.)
To illustrate further, consider appeals to pity. Deigh makes an
importantpoint about why such appeals are often appropriate:
Contextoften makesthe différencebetweencogency and subterfuge
whenit cornesto assessingan argumentthatis basedon appealto émo-
tion.An argumentcallingfor individuaior collectiveactionin the face
of truly wretchedor alarmingcircumstancescould not be faultedfor
prémissesdesignedto conveypity or fear,if it were otherwisesound.
(1995, 114)
But the appropriateness,even necessity, of the appeal is compatible
with holding that it is likely to induce fallacious or distortedreasoning.
Assume that unless some vivid remindersare offered of the avoid-
able human suffering involved in a lack of adequate health care, the
public, most of whom do have adequate coverage, will not assess
properly the value of a National Health Plan. Assume also that the
appeal, while improving judgment compared to when the émotions
are unengaged, is likely to overshoot. It will bias us towardjudging
more favorably arguments for a National Health Plan than is strictly
evidentially warranted. (Socrates in the Apology [34c-d] observes
that the appeal to pity may be both common in a setting and even
one's best hope, and yet there is the option of resisting it.)
The Fnotf thesis draws support from the seeming incompatibil-
ity of holding that an argument involves a fallacy and yet can be ac-
ceptable and appropriate.The reasoning here is mired in the fuzziness
of these terms and related ones such as relevance. But on even a par-
tial understandingof these terms, there is no incompatibility. An ar-
gument that involves a fallacy can be acceptable for serving a number
of argument-relatedpurposes. So, too, it can be appropriateto offer,
facilitating good judgment.

VI

Defenders and followers of the Fnotf thesis view themselves as offer-

ing pragmatic (or contextual) analyses of fallacies and argument.
While, as indicated, I have différences with a number of those who
348 JONATHAN E. ADLER

accept thè Fnotf thesis in how to analyze or characterize some falla-

cies, I have had little to say about thè pragmatics of arguing (or argu-
mentation) itself.
Broadly, my criticisms are directed at treating thè pragmatics of
arguing or an argument as analyses. Characteristically, adhérents of
the Fnotf thesis cite the work of Grice (1989) on conversational rea-
soning to support their approach. But they betray his message.
Grice erected his powerful account of the pragmatics of meaning
to clear the way for a serious, formai semantics. Specifically, he ex-
plicitly attempted to show that the seeming déviation of thè logicai
constants in ordinaryEnglish from their analyses in formal logic was
due to thè intrusive, however naturai, imposition of conversational
assumptions and expectations. A similar Gricean approachto fallacy
and argument would distinguish pragmatics from analysis, with the
former clearing the path for the latter.

Department of Philosophy
Brooklyn College and thè Graduate School, C.U.N.Y.

Notes
1. Thanks to Julia Driver, Eugene Garver, Ralph Johnson, and Harvey Siegel for
helpful comments. m also grateful to the editors for granting me extra space to
respond to a referee's comments.
2. Walton is influenced by van Eemeren and Grootendorst (1987 and 1984, chaps.
7 and 8, among other works), though for them a fallacy is a violation of a rule of
dialogue.
3. In a 1996 article, I criticize the implications of the Fnotf thesis for teaching.
4. I am doubtful that Powers's (1995) clever theory can cover thèse cases. The
assumption he makes is that equivocation must explain why fallacies appear to be
good. But why cannot an argumentjust (falsely) appear to be good?
5. This example also illustrâtes the strategy of adding, without independent sup-
port, crucial premises to render a fallacious argument nonfallacious, which is central
to arguments that proper (charitable) interprétations of alleged fallacies reveal them
not to be fallacies. Against this strategy, see Adler (1994 and 1996).
6. For a searching examination of this thème in Aristotle, see Garver (1994, chap.
6, and also his forthcoming article).
7. In différent ways, my positive view is influenced by Cohen (1977) and Fogelin
(1994).
8. Hère is an example in which, in the view I favor, thè falsity of a (missing)
premise counts as a defect in form or structure.
9. Independently,Driver,Johnson, and Siegel each raised an objection on this count.
10. For example, an arguer cites statistical studies to further his argument, which
he represents as authoritative. A participant knows that the arguer has a forceful per-
sonal Investment in persuasion, but the audience is not able to check on thèse reports.
1 1. Concerning some fallacies ad, Finocchiaro writes, "Being non-arguments, they
cannot be logically incorrect arguments" (1981, 17). Hamblin (1970, 251-52) con-
fuses the matter of nailing the fallacy (of showing that there really is an argument)
with the différent problem of whether a good argument has to be accepted by ail.
12. Walton devotes chapter 9 of Arguer's Position (1985) to this problem.
 FALLACIES
NOTFALLACIOUS:
NOT! 349

Works cited
Adler, Jonathan. 1994. "Fallacies and Alternative Interprétations."Australasian Jour-
nal ofPhilosophy74: 271-82.
. 1996a. "Charity,Interpretation, Fallacy." Philosophy and Rhetoric 29: 329-
43.
. 1996b. "Teaching and the StructuralApproach to Fallacies." Inquiry: Criticai
Thinking across the Disciplines 15: 94-106.
Biro, John, and Harvey Siegel. 1992. "Normativity, Argumentation, and Epistemic
Theory of Fallacies." In Argumentation Illuminateci, ed. F. H. van Eemeren, R.
Grootendorst, J. A. Blair, and C. A. Willard, 85-103. Amsterdam: SICSAT.
. forthcoming. "Epistemic Normativity, Argumentation, and Fallacies." Argu-
mentation.
Brandom,Robert. 1988. "Inference,Expression, and Induction." Philosophical Studies
54: 257-85.
Brinton, Alan. 1995. "TheAd Hominem."In Fallacies, 213-22. See Hansen and Pinto
1995.
Cohen, L. Jonathan. 1977. The Probable and the Provable. Oxford: Oxford Univer-
sitv Press.
Deigh, John. 1995. "Criticai Study: Douglas Walton The Place of Emotions in Argu-
ments." Informal Logic 17: 113-21.
Feldman, Richard. 1994. "Good Arguments." In Socializing Epistemology, ed.
Frederick F. Schmitt, 159-88. Lanham, Md.: Rowman & Littlefield.
Finocchiaro, Maurice A. 1981. "Fallacies and the Evaluation of Reasoning." Ameri-
can Philosophical Quarterly 18: 13-22.
Fogelin, Robert. 1994. Pyrrhonian Reflections on Knowledge and Justification. Ox-
ford: Oxford University Press.
Garver,Eugene. 1994. Aristotle 's Rhetoric: An Art of Character Chicago: University
of Chicago Press.
. Forthcoming. "The Ethical Criticism of Argument."Philosophy and Rhetoric.
Grice, Paul. 1989. Studies in the WayofWords. Cambridge: HarvardUniversity Press.
Hamblin, C. L. 1970. Fallacies. London: Methuen.
Hansen, Hans V. 1990. "An InformalLogic Bibliography."InformalLogic 12: 155-84.
Hansen, Hans V., and Robert C. Pinto, eds. 1995. Fallacies. University Park, Pa.: The
Pennsylvania State University Press.
Hinman, Lawrence M. 1982. "The Case for Ad Hominem Arguments."Australasian
Journal of Philosophy 60: 338-45.
James, William. 1951. "TheWill to Believe." In his Essays on Pragmatism, ed. Albury
Castell, 88-109. New York: Hafner.
Joseph, H. W. B. 1916. An Introduction to Logic. 2d revised ed. Oxford: Clarendon
Press.
Krabbe, Erik C. W. 1995. "Appeal to Ignorance." In Fallacies, 251-64. See Hansen
and Pinto 1995.
Kripke, Saul. 1980. Naming and Necessity. Cambridge: HarvardUniversity Press.
Locke, John. 1975. An Essay concerning Human Understanding. Ed. P. H. Nidditch.
Oxford: Oxford University Press.
Piato. Apology. In Five Dialogues. See Piato 1981.
. Euthyphro. In Five Dialogues. See Piato 1981.
. 1981. Five Dialogues. Trans. G. M. A. Grube. Indianapolis: Hackett.
Powers, Lawrence H. 1995. "Equivocation."In Fallacies, 287-301. See Hansen and
Pinto 1995.
Ross, W. D. 1988. The Right and the Good. Indianapolis: Hackett.
Searle, John. 1978. "Prima Facie Obligations." In Practical Reasoning, éd. J. Raz,
81-90. Oxford: Oxford University Press.
Toulmin, Steven, Richard Rieke, and Allan Janik. 1984. An Introduction to Reason-
ing. 2d éd. New York: Macmillan.
van Eemeren, Frans H., and Rob Grootendorst. 1984. Speech Acts in Argumentative
Discussions. Dordrecht-Holland: Foris.
. 1987. "Fallaciesin Pragma-DialecticalPerspective."Argumentation 1: 283-301.
350 JONATHAN E. ADLER

. 1992. Argumentation, Communication, and Fallacies: A Pragma-Dialectical

Perspective. New Jersey: L. ErlbaumAssociates.
Walton, Douglas N. 1985. Arguer's Position. Westport, Conn.: Greenwood Press.
. 1992. The Place of Emotion in Argument. University Park, Pa.: The Pennsyl-
vania State University Press.
. 1996. Argumentsfrom Ignorance. University Park,Pa.: The Pennsylvania State
University Press.
Willard, Charles A. 1989. A Theory of Argumentation. Tuscaloosa: University of Ala-
bama Press.
. 1995. "Failures of Relevance: A Rhetorical View." In Fallacies, 144-58. See
Hansen and Pinto 1995.
Woods, John. 1995. "Appeal to Force." In Fallacies, 240-50. See Hansen and Pinto
1995.