ARGUMENTUM : FROM CHAOS TO FORMAL DIALECTIC The method of dialogue-tableaus as a tool in the theory of

E.M. BARTH and J.L. MARTENS

Introduction

Surveying the field o f extant theories o f fallacy, under the guidance of C.L. Ha mb lin , (sent authors. I t became obvious to us that in this field in tu i-

(1) Painstaking h isto rica l investigations o f t h e v ie ws o f fallacies which are to be found in philosophical literature from till the present moment are needed, fo r the systematic suggestions th e y offer and also because con- temporary uses o f language are, as Russell said, ( 2 ( ) q s h o t ma1l d i a l )(2 ) Bertrand RUSSELL, Wis d o m o f t h e Wes t, Mac donald, London, 1959. e cCt i c » . i Ls t .a k He n A f M r B o L m I t N h , i F s a b l o l o a k c . i e s , M e t h u e n , L o n d o n , 1 9 7 0 . T h e e x p r e s s i o n « f o r - ARGUMENTUM AD HOMINEM 77

through wit h the fading hues o f past philosophic theor- ies» (p. 309). The questions that have to be asked there- fore co mp rise t h e fo llo win g : « I n wh a t w a y does (author) use the expressions 'argumentum a d hominem', 'petitio principir, 'non sequitur' ?» No realistic theory con- struction concerning fallacies i s t o b e expected i f th e history o f fallacies is bypassed. A n important contribu- tion t o th is historical wo rk as a p re limin a ry t o theory construction is th e above mentioned wo rk b y Hamblin. (2) Th e re is a lack of analytical tools for the task of analysing and classifying fallacies. That we have not yet got much in the wa y of a theory of rational argumentation is at least partly due to the shortage of instruments suitable for this task. Ou r contention is that in the field o f argumentation generally and mo re especially i n the sphere o f fa lla cy theory construction can profit from certain tools developed in (modern) f o rma l logic, viz, i n th e th e o ry o f the so - called logical constants. O u r wo rkin g hypothesis w i l l be that especially the dialogical set-up o f first-order predic- ate logic that we owe to P. Lorenzen (and in d ire ctly to E. W. Beth) w i l l be ve ry suitable as a starting point fo r further theory construction in the field o f argumentation. In fact the method of dialogue-tableaus is the first success- ful fragment o f a th e o ry o f discussion, since o th e r a t- tempts a t developing such a theory do n o t contain a n y definition of what it means to win or to lose a critical dis- cussion.

In this paper we shall use these ideas as wo rkin g hypotheses in order — first goal — to cla rify what various authors have meant when saying that an argumentum ad hominem has been employed in a certain discussion. The following pages contain:

I. an attempt to map the terminological differences we have found in the uses of the expression 'argumentum ad hominem'. We shall t ry to point out some constants in these uses and in the views that are expressed b y means o f them, an exposition of the main elements in the history of this 78 E. M. BARTH and J. L. MARTENS expression, b y means o f the framework offered in Lorenzen's dialogue-theory of logical constants, a conclusion, and IV. suggestions for further research.

I. I n the literature we have investigated, at least four differ- ent meanings of the expression 'argumentum ad hominem' can be discerned:

I. B y 'argumentum ad hominem' is meant a purely personal and direct attack; this is the wa y in which Beardsley, ( 3ney and Scheer, (4) Kahane, ( )5th eC exaprres-sion. I n the words o f Carney and Scheer (o.c., p. 2) 0):M i c h a l o s ( ' ) a Thne ad dhomin em fallacy (fallacy directed to the man) is P cuommrittetd whi enl thel con clusion of an states that ( a vie w is mistaken, and the reasons given f o r this con- 7 clusion amount to no more than a criticism of the person ) or persons maintaining the view. u s e Copi, ( 8'abusive argumentum ad hominem'. ) R e ( s1953c0. h )(e rNe M4w York, 1964. ( o)( CoJ5mp., Belmont, 1971. 9 n ra)( )woomH6o d Cliffs, 1970. a eo)( 1972. n CswA7 .Cal)( ' ) I rv ing M. Copi, Introduction t o Logic, second edition, Mac millan, Ne w dYork, 1 961. BAreR G ( ERdxi uYo9rk, 1969. ANKCc )( ") S.D. GUTTENPLAN and M. TAMNY, Logic, Basic Books, Ne w York , 1971. t REA.h t N e DYHMa i n SaAIr c LnNCd p h EdHL l o YR,A. l a ,iLP a n PcoOU s a rhgSR R ai,T n E crcI d S tdamL T C iSnpL U a cCdr, E maHCoL R n lEov , LEnig y I oRtni ( n g,egc t 1 iFmYa r 5 cupol o ) ,nouT d Pdrh c u raaRi a c emren l t neyak i l tnRsi o t ithon n h caeng t elti, i o -sonH s L Horga o a afi,r g n lLcPp i lo,re c ,gWer , Eian S ncdtn t g,sid . lMwcR M eaoe a wcr- r oMtH, t oihaN i dlPle n Clulw ' lab,Y s in!Eo P f,.nr r fgk e sl, s ,e -s , N e w ARGUMENTUM AD HOMINEM 79

2. B y 'argumentum a d hominem i s meant a n indirect o r direct attack on the person: Copi, ( a8nd Tamny. ("). E.g. (Copi, oc., pp. 54f): ) R e s c h e r ( D We may designate this fallacy on the first interpretation ) as tahe «anbusivde» va riety. I t is committed when, instead G uof trtyintg teo dnisprpovel thae tru th o f what is asserted, one n attacks the man who made the assertion The other interpretation of the fallacy of argumentum ad hominem, the «circumstantial» variety, pertains to the re- lationship between a person's beliefs a n d h is circu m- stances. Wh e re t wo men are disputing, one ma y ignore the question of whether his own contention is true or false and seek instead to prove that his opponent ought to ac- cept it because o f his opponents special circumstances.

3. A ve ry wid e meaning is assigned t o th is expression i f one says (in the words o f Whately, (") p. 200) th a t an argu- mentum a d hominem « is addressed t o th e peculiar circu m- stances, character, A V O WE D OPINIONS, o r past conduct o f the individual». To this definition corresponds the language used b y Whately himself (oc.), Hyslop ( 1 4. B y 'argumentum ad hominem' is meant an argument e x 1co ncaesnsis:d cp . JSchoophennhausert, ("o) Lnoceke (") and Perelman and (Olbre"chts-)Tytec. a (1. Locke says (o.c., Bk. 4, Ch. 17, §§ 19-21):

(II )( YoR1rk, 1905. i2(") Henry W . JOHNSTONE J r., Philos ophy a n d Argument, Penns y lv ania Stac)t e UP., 1959. hJ(") A rt h u r SCHOPENHAUER, 'Eristische Dialek t ik : I n: De r Hands c hriftlic he Naachlass, vol. III, Waldemar Kramer, Frank furt 1 9 7 0 . rm( 1963. de1 Ws5(1 ts) Ch . PERELMAN a n d L . OLBRECHTS-TYTECA, Th e N e w Rhet oric — A TreH) atise on Argumentation, Univ ers ity of Not re Dame Press, Not re Dame, 196AYJ9. Trans lation of: Trait é de l'Argumentation, Presses Univ . de France, Paris, 1958. TSo ELh LOn YPL ,O ETC lhK eE mE, elE nes tms sea ony fto Lsn oH gfu iLm coa ,gn BiU .cn F,d eC Bhr oas wrt ela sen ,sd LSi ocn nrg di, obS nc ,ei 1re 8'n 2st 9Si .oa nV se ,r Nl ea wg , A a l e n , 80 E. M. BARTH and J. L. MARTENS

Before we quit this subject, it may be worth our wh ile a little to reflect o n fo u r sorts o f th at men, i n their reasonings with others, do ordinarily make use of to prevail on their assent; o r at least so to awe them, as to silence their opposition Thirdly, a third wa y is to press a man with consequences drawn from his own principles o r concessions. This is al- ready kn o wn under the name o f argumentum ad homin- em.

If we use the numbers in the above numbering as names fo r the extensions (i.e., fo r the classes) o f the concepts described in 1 to 4 inclusive, then we can say:

i.e., 3 contains the other classes as proper 1c2c314 3 sub-classes,

1 () 4 — 0 i . e . , 1 and 4 are completely disjunct. In other words, 1 and 4 are completely distinct uses o f the expression 'argumentum ad hominem'.

We shall now t ry to sketch the main stages in the history of the expression 'argumentum a d hominem'. I n doing th is we may confine ourselves to a part of the sources quoted, with - out any loss. The f irst t o use th is expression as a technical t e rm wa s Locke. He explains what he understands by it in a single sent- ence (see quotation sub I above). Hamblin maintains (o.c., p. 161) th a t Locke g o t the idea o f introducing th is expression from Aristotle ( 1 7 ) «But when not harping, you have the power to harp: and ( 1 7th7erebfo)re: you could harp when not harping.» «No: he has not the power to harp-while-not-harping; merely when he

(II) Aris totle. D e Sophisticis Elenchis, Th e Wo rk s o f Aris totle, v o l. I , Oxford UP., Oxford, 1937. ARGUMENTUM AD HOMINEM 81

is not doing it, he has the power to do it.» Some people solve this last refutation in another way as well. For, they say, if he has granted that he can do anything in the wa y he can, still it does not follow that he can harp when not harping: fo r it has not been granted that he will do any- thing in every way in which he can; and it is not the same thing «to do a thing in the wa y he can» and «to do it in every wa y in wh ich he can)). But evidently they do not solve it properly: fo r of arguments that depend upon the same point the solution is the same, whereas this will not fit a ll cases o f the kin d n o r ye t a ll ways o f putting the question: it is valid against the questioner, but not against his argument.

In several medieval treatises we find the distinction, influen- ced b y Aristotle, between 'a d hominem a n d 'a d orationem' (cp. Hamblin, o.c., p. 161). I n the treatise Summa Elencorum Sophisticorum (") we find (cp. oc., p. 430):

Sed tamen prius sciendum est quod solutionum alia est ve- ra, alia falsa. Quam quidem divisionem ostendit Aristoteles aliis verbis dicens alia esse ad orationem, alias ad homin- em (our italics — I l i a solutio est ad °ratio- nem que ostendit e t so lvit vit iu m ipsius orationis; q u e quidam solutio dicitur vera. Ilia vero est ad hominem que non vitiu m orationis solvit, sed hominem impedit.

The works o f Whately, Schopenhauer, Hyslop, Perelman and Olbrechts-Tyteca, and Johnstone (sub 3. and 4. above) co n - tain extensive discussions o f ad hominem that go into much more detail; we shall lo o k into these further on. The group of modern introductions to lo g ic (Copi, Caste11, Carney and Scheer, Rescher, Guttenplan and Tamny) and of popular scien- tific works in the field o f (still embryonic) argumentation the- ory (Beardsley, Michalos, Purtill, Kahane) does not offer many new points of view. The second group (Whately etc.) is much

( 1 8 ) L . M . D E R U E , L o g i c a M o d e r n o t u m I , V a n G o r c u m , A s s e n , 1 9 6 2 . 82 E. M. BARTH and J. L. MARTENS more interesting for our purpose: I. Whately's vie w is that what he calls an argumentum ad hominem can be permissible (oc., pp. 201f):

It appears then (to speak rather mo re technically) th a t in the «argumentum ad hominem» the conclusion wh ich actually is established, is n o t the absolute and general one in question, but relative and particular; viz, not that «such and such is the fact,» b u t that «this man is bound to admit it, in conformity to his principles o f Reasoning, or in consistency with his own conduct, situation, & c.» A ll this, as we have said, is perfectly fair, provided it be done plainly, and avowedly, but if you attempt to substi- tute this partial and relative Conclusion for a more gener- al one — if you triumph as having established yo u r pro- position absolutely a n d universally, f ro m h a vin g esta- blished it, in reality, o n ly as fa r as it relates to yo u r op- ponent, then you are guilty of a Fallacy of the kind which we are now treating of.

In order to analyse what Whately says here it is convenient to e mp lo y the terminology introduced b y Paul Lorenzen i v his dialogical set-up o f formal(ized) lo g ic ("). Lorenzen fo rm- ulates the definitions of the logical constants of first-order pre- dicate logic (connectives and quantifiers) as rules f o r verbal behavior i n crit ica l debates. Th e se a re t wo -p a rt y ve rb a l «feuds» where the issue is a thesis T, wh ich one p a rty (the proponent, P) wants to defend and the other party (the oppon- ent, 0 ) wants to criticize. The opponent may o r may not con- cede th e tru th o f certain propositions a t th e outset o f th e discussion, in the sense that the proponent ma y make use o f these concessions — b y attacking (criticizing, challenging)

('°) Fo r expositions in English of the dialogic al tableau-method, see a.o.: Paul Lorenzen, Normat iv e Logic and Ethics, BA. Hochschultaschenbticher, Mannheim, 1969; Kuno Lorenz, 'Rules versus theorems — A Ne w Approac h for Mediat ion Between int uit ionis t ic a n d Two-v alued Logic ', i n J ournal of Philosophical Logic 2 (1973), pp. 393-415. ARGUMENTUM AD HOMINEM 83

them — and the opponent ma y not. Th e proponent argues for his thesis, we may say, ex concessis (i.e. on the ground of, or: b y means of, th e opponents concessions). Th e class o f the opponent's concessions ma y, o f course, also b e empty, provided the proponent agrees to enter such a discussion.

Opponent Proponent

Initial concessions

thesis

Logical truth can n o w be defined as follows: a thesis T is a logical truth iff a proponent of T has a winning strategy against any opponent wit h respect to T (whether he/she makes a n y concessions in advance o r not). A thesis T follows lo g ica lly from a set of concessions if f a proponent of T has a winning strategy against any opponent wit h respect to T wh o makes at least these concessions. Lorenzen does n o t use the expression 'e x concessis'. W e have found it a. o. in the wo rk o f Perelman and Olbrechts- Tyteca in their discussion of arguments ad hominem and think it is eminently suited as a technical te rm also in Lorenzen's dialogue-logic (as we shall call it, fo r short). Whenever the opponent has a choice as to what he/she may do next we shall speak of different lines of attack (or lines of criticism). Then we can also say: a thesis T is a logical truth iff a proponent of T has a winning strategy against every pos- sible line of attack (according to the rules o f the verbal

If P has defended his/her thesis T successfully, against O s criticism, b y arguments ex concessis, then it is not yet settled whether T is true, not to speak of: whether T is logically true (valid). Wh a t has been settled is me re ly tha t anyone wh o concedes what 0 concedes (namely O s avowed opinions) can- not maintain this criticism of T. Notice that we do not say that such a person cannot suc- cessfully ca rry out a n y critica l attack o n T. Fo r it is quite possible that, fo r a ll we know, the same o r another opponent making the same concessions (and n o other ones) ca n take up another line of attack such that P cannot maintain T against that attack. In Whately's text this is not ve ry clearly put, wh ich is un- derstandable, since h e d id n o t possess th e analytic to o l o r instrument fo r structuring his thoughts wh ich we have used here: the method of dialogue-tableaus, borrowed from element- ary logic. In fact, Wh a te ly mixes up the following points:

a. the insight that T is n o t necessarily defendable against a person who makes the same concessions b u t wh o chooses another line of attack; b. the insight that even if T is defendable against any op- ponent 0 wh o makes these concessions, irrespective o f the line o f attack 0 ma y choose, th is does n o t guarantee th e truth of T; fo r T to be true the truth o f those o f O s conces- sions which are needed for the defence (relative to some line of attack) is also required.

2. The argument in Schopenhauer's Eristische Dia le ktik is not clear: a. I n Schopenhauer's vie w the a im o f scientific dialectic is to analyse qjene Kunstgriffe der Unredlichkeit im Disputieren» (o.c., p. 676). i.e. the various artificial tricks o f an unreason- able kin d wh ich are used in discussions. His negative v ie w with respect to these Kunst griffe is understandable from what he says in a note (oc., p. 671):

Meine A n sich t a lso ist , d ie Dia le kt ik v o n d e r L o g ik ARGUMENTUM AD HOMINEM 85

scharfer zu sondern ais Aristoteles gethan hat, der Logik die objektive Wahrheit, so we it sie formell ist, zu lassen: und d ie Dia le kt ik a u f das [ n ich t imme r o b je ktive Rechtbehalten zu beschrânken

b. A fte r having formulated this a-logical starting point he announces that he will describe the various Kunstgrifte with - out relating them to the question o f who is objectively rig h t or wrong. He seems to hold, therefore, that it is possible to use them without being unreasonable. Nevertheless the examples he offers o f such Kunstgrille ca rry a strong negative flavor. c. I t is interesting to observe that Schopenhauer first men- tions ad hominem as one of the modi o f discussion but later lists argumentum ad hominem as one of the Kunstgrille (cp. 3. below). Our conclusion is that Schopenhauer fails to g ive a clear exposition of what he considers to be admissible moves in a rational discussion; i f he had done that first, h e could have analysed the Kunst griffe by means of this system of rules fo r admissible moves (see sub I V below).

3. Ne w is what Hyslop says about the possibility of leading ones opponent into contradiction b y means o f an argumentum ad hominem (o.c., p. 251f):

They [i.e. the argumenta non ad res — E.M.B./J.L.M.] are invalid only as proofs o r disproofs of a matter in discus- sion, but they are not invalid as means o f establishing a contradiction between two propositions H e n c e the se- veral argumenta non ad res, in merely proving a contra- diction somewhere, a re fallacies o f ignoratio elenchi, i n the relation of assuming that they prove anything. But we must distinguish between th is a n d t h e ir va lid use f o r establishing a contradiction.

Hyslop here uses some seemingly opaque expressions, viz. o(in)valid as proofs o r disproofs ...» and «(in)valid as means of establishing a contradiction ...». What he means is however 86 E. M. BARTH and J. L. MARTENS clear and can be put differently; in fact what he says corresp- onds fa irly we ll wit h what we are acquainted wit h fro m Lo- renzen's dialogue-method in formal logic (a negation is attack- ed b y conceding the negated proposition):

Opponent Proponent

UI Initial concessions Uk

i = n U, (attack on — U,)

Ipse dixit ! (or: Ipse concedit !)

Unless 0 can embark o n another lin e o f attack, he/she has lost th e discussion; th e lin e o f attack o r criticism depicted above is in any case lost. But, as Hyslop says, P has not there- by proved his point (cp. our discussion of Whately, sub II, 1). Hyslop does n o t seem t o realize, however, t h a t f o r some theses and concessions P can prove his/her point b y refuting every possible line o f attack in this manner. Precisely wh a t i n O s ve rb a l behavior turned o u t t o b e «fatal» (to this line of criticism) ? The answer is clear: 0 has questioned, o r challenged, a proposition (U,) wh ich he himself has conceded not to question: one of his own initial o r subse- quent concessions. We n o w come to a point in Lorenzen's model fo r rational discussions wh ich is o f considerable importance and which, in our experience, puzzles many students of dialogue tableaus. It concerns what may be called the basic asymmetry between the debaters, which at first is felt by many to be «unjust» to op- ponents, i.e. to grant a special favor to proponents: wh ile the proponent ma y attack any of the opponents statements, a n y ARGUMENTUM AD HOMINEM 87 number of times, the opponent may, a t any stage o f the dis- cussion, o n ly attack P s last statement (utterance o f a propo- sition). That there is no question of an injustice o r a privilege here follows fro m (1) and (2):

(1)Already at the outset o f the discussion the relation bet- ween the debaters is not a symmetric one. Fo r in Lorenzen's conception o f a rational discussion there is o n ly one in itia l P-concession: the thesis T. (I n order to understand that the rules o f the discussion are not only reasonable but completely natural it is also necessary to keep f irmly in mind that T may be a conjunction of any (finite) number of sentences: T may be a wh o le philosophical system T h i s w i l l become important again when we analyse Johnstone's views below.)

(2) I f we allow one and the same individual opponent 0 to pursue various lines o f attack, one after the other, P has not won the discussion with 0 u n til 0 cannot think of any other possible lin e o f attack and has lo st a ll those he/she co u ld think of. (In this wa y one individual critic can in principle explore a complete philosophical system, lo o kin g f o r lo g ica l we a k- nesses b y paying attention to the logical constants occurring in the propositions o f which the system consists.) So, if, and only if, 0 is granted this right o f multiple lines of attack can he/she reasonably be required to attack any one statement o n ly once with in the same line o f attack; and this may be formulated as a requirement to always attack P's last statement, if an attack is performed at a ll (rather than a de- fense o f an attacked o wn statement). Now observe that P can compel the opponent into making the «fatal» step o f doubting his o r her own concessions o n ly because 0 is only allowed to attack P's last statement. In the above schema this was U, one of O s concessions. Observe also that P cannot be so compelled. In the fo llo w- ing situation: 88 E. M. BARTH and J. L. MARTENS

Opponent Proponent

v P will lose this iline of attack if he/she is so unwise as to chal- lenge Os last statement, V,:

Ipse dixit ? (V (opponenti wins) ) But P, unlike 0 , does not have to do that, fo r he o r she ma y also a tta ck o n e o f O s e a rlie r statements,and s o perhaps avoid losing this line of attack. This shows that the basic fo rm o f verbal inconsistency is not to utter two contradictory propositions, one after the other, but to utter and to challenge the same proposition with in one and the same line of attack from ones adversary. (The oppon- ent has always a winning strategy, however, if P utters a con- junction o f t wo contradictory propositions, V and V , b u t we need not go into that here.) I f inconsistency is defined in this way, as basic inconsistency, then consistency is required of both parties in a rational discussion: a n y participant in a rational discussion wh o shows verbal inconsistency with in a certain line of attack loses that line of attack. In order to interpret and to make more precise b y means o f dialogical tableaus such discussions of arguments ad hominem as that of e.g. Hyslop, one obviously has to have a f irm g rip of the rules for negation in dialogue situations.

4. Perelman and Olbrechts-Tyleca make the following dis- tinctions (o.c., p. 110f): The possibilities fo r argumentation depend o n what each participant is ready to concede, o n the values he recog- ARGUMENTUM AD HOMINEM 89

nizes, on the facts on wh ich he indicates his agreement: for this reason, any argument is an argument ad hominem or e x concessis. The frequent opposition o f argument ad hominem to argument ad rem, the first relating to opinion, the second to the truth o r the thing itself, is due to the fact that people forget that the truth in question has to be accepted. In terms of our theory, argument ad rem cor- responds to an argument that is claimed to be valid for all reasonable beings, that is, ad humanitatem. Argument ad humanitatem wo u ld be a special, b u t important, case o f argument a d hominem. A r g u m e n t a d hominem mu st not be confused with argument ad personam, wh ich ma y be defined as a personal attack on the opponent and which aims essentially a t disqualifying h im.

Using our terminology from the dialogical set-up of the theory of logical constants we can sharpen these distinctions. Assume that P is the proponent of a thesis T and 0 an opponent of P in respect o f T; then the distinctions made b y Perelman and Olbrechts Tyteca can, we think, be formulated as follows: — the authors ca ll P's argumentation 'ad rem' if and o n ly if P pretends to be able to win a ll possible lines of attack; — P's argumentation is called 'ad hominem' if and o n ly if P conducts his argumentation me re ly wit h the cla im that he can win one o r more lines o f attack fro m this opponent, 0 , owing to concessions made by 0 ; — P's argumentation is called 'ad personam' if and o n ly if P attacks the person o f 0 although a description o f the per- sonal traits ascribed to 0 are not a part o f P's thesis, n o r o f O's concessions. Arguments o f the first two kinds can be permissible, those of the third kind never are.

5. Johnstone gives a definition of 'argumentum ad hominem' which coincides wit h Whately's. Nevertheless, i t is hard fo r him to distinguish between the meanings o f 'argumentum ad rem' and 'argumentum ad hominem' (o.c., p. 3f): 90 E. M. BARTH and J. L. MARTENS

the abyss that separates conflicting philosophical sys- tems precludes a n y use o f argumentum a d re m; f o r t o appeal t o evidence i n attacking a position th a t cla ims to include all evidence is to beg the question. Thus every valid philosophical argument is ad hominem

What fo llo ws is a n attempt t o cla rif y Johnstone's position, again by means of distinctions and terminology taken from the dialogue-theory of logical constants. In a «formal» dialogue (cp. Lorenzen ( en2ts arguments are ex concessis, b y definition o f 'formal dia- logue';)) a hlenl c e thoey fare argtumehnta ead h ominem in the termino- logpy orf Peorelmpano andn Olb-rechts-Tyteca. A «material» dialogue is, again b y definition, one in wh ich arguments e x concessis are permitted, but in addition also «ostensive» arguments, in defence o f attacked atomic propositions. Notice t h a t a discussion about mathematical t ru t h i s a «formal» dialogue, i.e. a dialogue in wh ich a ll arguments are ad hominem in the sense o f e x concessis. I n discussions i n other fields o f science the situation is generally otherwise; there ostensive arguments can and must be used: «But lo o k here I» and the like, unless the debaters are concerned wit h whether one statement with empirical content follows logical- ly fro m other statements wit h empirical content. Now Johnstone claims tha t i n a philosophical discussion, too, a ll arguments a re a d hominem. He probably means t o say, among other things, that with the exception of reference to relevant literature (wh ich ma y be grouped wit h the con- cessions) ostensive arguments usually cannot be applied. But he means more than that: «... to appeal to evidence in attack- ing a position that claims t o include a ll evidence is t o beg the question». I.e., he offers a reason wh y ostensive argum- ents cannot be used. Observe that Johnstone is concerned wit h what an oppon- ent may do. He obviously does not consider it reasonable that the opponent demands a preliminary discussion as to whether or not the system to be defended by P really includes all evid- ence I Rather he presumes that the parties take it fo r granted ARGUMENTUM AD HOMINEM 91 that T is a complete philosophical system and that P is unable to add another word to it. I f that is so, then indeed 0 cannot expect a n y material justification f ro m P (although 0 is o f course free to accept ostensive, o r material, arguments if they are offered). However, 0 is entitled to expect some justifica- tion of T from P (otherwise there would be no discussion o f T). There now seem to be the following possibilities: either (a) P defends the atomic propositions in T which 0 has attacked with «Ipse dixit !» — i.e., defends the system by show- ing that someone who makes O s concessions cannot maintain O's criticism of it; or else (b) P defends the attacked atomic propositions by show- ing that they are not incompatible with O s concessions; or else (c) P defends the system as a whole merely b y show- ing (which may be hard enough) that it does not contain any inconsistencies (viz. Hilbert's position in respect of the ju sti- fication o f mathematical theories); or else (d) P defends the system as a whole by challenging 0 to point o u t an inconsistency in T. The kin d o f defence described sub (a) is cle a rly stronger than that sub (b), which is stronger than that sub (c), and the one sub (d) is the weakest o f these kinds o f justification, o r ways to meet criticism. Johnstone is so unclear because he does n o t sharply dis- tinguish between the roles of opponent and proponent in res- pect o f a thesis T. I t is impossible to impose our distinctions (a) ( d ) upon h is text, wh ich brings t o min d the frequently heard expression «immanent criticism». This expression, too, is much too vague to allow for a clarification in terms of one of the alternatives (a) - (d). We rather th in k that a ll the pos- sibilities (a) - (d) a re running through the head o f those wh o demand that th e ir systems be criticized o n ly «immanently». Observe, finally, that no matter h o w we interpret h im John- stone cannot possibly mean the same by 'argumentum ad rem' as Perelman and Olbrechts-Tyteca do. Let us now take a close look at one of Johnstone's examples of an admissible argumentum ad hominem (o.c., p. 64): 92 E. M. BARTH and J. L. MARTENS

One of the arguments used b y Eudoxus in the attempt to show that pleasure is the chief good was that «any good thing — e.g., just o r temperate conduct — is made more desirable b y the addition of pleasure.» But Aristotle called attention to the fact that an argument of exactly the same type can be constructed to show that the chief good is not pleasure. For, as Plato had already argued, ‹

Schematically:

Aristotle (0) Eudoxus (P)

Pleasure is the chief good (T) F? I Argument A: all good things are made more desirable by the addition of pleasure Counterargument B: all good things are made more desira- ble by the addition of wis- dom

Johnstone analyses this discussion thus: Aristotle does n o t attack T d ire ctly (hence w e have p u t the first question ma rk in parentheses), but attacks argument A b y wh ich Eudoxus supports the thesis, T. Johnstone asks us to notice that Aristotle cannot now advance as an argum- ent a case of something that does not become more desirable by th e addition o f pleasure, f o r that wo u ld be (compare our first quotation from Johnstone above) ! Aristotle's argument B is, and rig h tly so, in Johnstone's o pi- nion, ad hominem, in the sense that he attacks Eudoxus' thesis with «his own» principles. S imila rly Eudoxus cannot answer argument B b y showing, e.g., that pleasure does not become more desirable b y the addition o f wisdom; to Johnstone this ARGLIMENTUM AD HOMINEM 93 would mean begging the question, too. Fo r this reason A ris- totle's criticism is devastating, in Johnstone's opinion. We do not find this analysis ve ry satisfactory. I n o u r opi- nion the force of Aristotle's argument can only be the follow- ing: i f counterargument B is accepted, then Eudoxus cannot any longer support T by means of argument A. Aristotle must, furthermore, be allowed to attack A directly (by instantiation, i.e. by means of directly a counter-example) and Eudoxus must be allowed to attack B directly. In our opinion neither of the two gentlemen has as ye t lost the discussion. What Johnstone wants to say in his discussion of this clas- sical example seems to be the following: some people (p ro - bably including Johnstone himself) expect a n opponent t o confine h im- o r herself to principles of argumentation and re- futation that are accepted by the proponent — i.e. to the pro- ponent's logic and perhaps also o f other first principles. This, too, seems to be a component of the demand which is so often encountered, especially in connection with idealistic philosophy, th a t criticism be «immanent criticism». O n one very important condition this requirement, as fa r as lo g ic is concerned, seems a ve ry reasonable one; fo r if P and 0 use different logics and consequently different (if any) ru le s o f discussion, a discussion between them is unlikely to take place, and if it does, it cannot be considered a rational one. P's de- mand that 0 must use P's logic in attacking P's philosophy seems reasonable, however, o n ly o n th e fo llo win g two fo ld condition:

(1) P mu st ha ve expressed t o 0 his/her readiness t o enter into a discussion about his or her own principles of argumentation a n d refutation a s w e l l a s about those favored b y O. Th is discussion should be con- ducted in accordance with the following rules: (i) each party is obliged to answer all questions rais- ed by the other party about the logic favored b y the former, (ii) each party is obliged to listen carefully to an explanation of, and to learn, the principles o f the logic favored b y the other party. 94 E. M. BARTH and J. L. MARTENS

(2) 0 must either have refused to enter into this discus- sion or the discussion must already have taken place with loss fo r 0 , o r without any of the parties having clearly «won».

If these conditions are fulfilled, then the demand o f «imman- ent criticism)) seems to be very appropriate. If, however, these conditions are not satisfied, there seems to be no reason wh y P should have the rig h t to prescribe his o r her o wn logical principles t o e ve ry candidate f o r the ro le o f opponent, and still expect to be considered a «reasonable philosopher» b y other minds.

Conclusion

We believe that we have demonstrated the value of Lorenzen's dialogical set-up o f the theory of logical constants as a con- ceptual tool and as a source fo r adequate terminology in the theory of fallacies. The concepts — the thesis under discussion, — the proponent of the thesis, — the opponent o f (th e proponent w i t h respect t o ) t h e thesis, — the concessions of the opponent, and furthermore the concepts, n e w to the present discussion, — a line of criticism (line of attack), — a line of defence (not employed above), — a formal argument, — an ostensive argument, and finally the definitions of

— P has won this line of attack against 0 in the discussion about T, — P has won the discussion about T against this opponent, 0 , ARGUMENTUM AD HOMINEM 95

— P has a winning strategy with respect to T against every opponent wh o makes the same concessions as 0 , — P has a winning strategy with respect to T against every opponent, n o ma t t e r wh a t concessions t h e opponent makes, have helped us to make more precise a number of views and opinions expressed b y present and past authors. Stephen Toulmin's vie w o n the relationship between «for- mal» and «informal» logic, defended in his Th e Uses o f A r- gument, is thereby refuted. There is no sharp distinction bet- ween «formal» and «informal» logic. Conversely, i t i s o n ly reasonable t o expect some feed- back fro m the further study o f fallacy (irrational argumenta- tion) and o f rational argumentation on the model fo r critica l discussions drawn up b y Lorenzen, and on the rules fo r pro- ducing rational arguments. Take as an example the situation that the opponent in the course o f the discussion withdraws one o r more of his concessions as a consequence of the pro- ponent's questions and other remarks. Such behavior is g e - nerally considered as eminently rational. Yet Lorenzen's mo - del in its present form does not accommodate this situation and should therefore be amended so as to account for the develop- ment and change o f opinions and even o f concepts (revision of definitions) i n the course o f a discussion o f some length.

IV. Suggestions f o r further research: fro m a list o f fallacies to a «complete» set o f production rules f o r rational argum- ents. (°)

We conclude wit h some general re ma rks about possible kinds o f systems o f rules fo r the evaluation o f arguments as fallacies. The relationship between inadmissible, me re ly rhe- torical Kunstgriffe (as Schopenhauer called them) and a the-

( tic)2 i n Hamblin's Fallacies, although o u r point o f v ie w deviates f rom his on 0several basic points, ) W e h a v e d r a w n c o n s i d e r a b l e i n s p i r a t i o n f r o m C h . 8 ( F o r m a l D i a l e c - 96 E. M. BARTH and J. L. MARTENS

ory o f wh a t constitutes rational argumentation can b e con- strued in two different ways. (1) The first presupposes that the theory o f rational argu- mentation is formulated as a set of necessary conditions which an argument must satisfy in order to pass as rational. Rhetor- ical Kunstgriffe can then be analysed and unmasked as moves which do n o t satisfy these conditions, e.g. a s moves wh ich transgress certain prohibitions. (2) A second and more contemporary approach consists in formulating a fin ite se t o f production ru le s f o r generating rational arguments. Lorenzen's dialogical rules wh ich consti- tute definitions o f the connectives and quantifiers are in fact rules of this kind. Such rules are sufficient conditions for call- ing th e produced argument rational. O n ly th e condition — here to be called 'the restrictive condition' — that the argum- ent can be generated b y one o r more o f these rules is a ne- cessary one. Schopenhauer's Kunst griffe and o th e r fallacies can then be unmasked as arguments wh ich cannot be gener- ated b y the production rules. Th e y do transgress one rule, viz, the restrictive condition that prohibits the use o f argum- ents that cannot b e generated b y means o f th e production rules. In the pursuit o f a theory o f rational and irrational argu- mentation and of fallacy it seems wise to aim at such produc- tion rules for the generation of rational arguments as the ideal kind of theory, although rules o f the first kin d are not to be despised entirely, a t least as long as the theory o f rational argumentation is in its infancy.

University of Utrecht E . M. Barth and J. L. Martens