This is a draft of the paper I read at the 1996 APA Central Division in
Chicago on Saturday, April 27, (Alan Code commented). Clicking on footnotes will
take you to them; a link at the end of the footnote takes you back to where you
were (approximately) in the text. Alas, the footnotes contain some modest chunks
of Greek in the form of TLG Beta code. I have tagged it all as 'preformatted'
text; if you set your browser's choice of 'fixed width' or 'fixed-pitch' font
to something which works with TLG Beta code (such as
Sgreek Fixed from
Silver Mountain Software,
if you can use TrueType fonts
), you should then see real Greek in the footnotes.
Here's the same paper in
postscript format and in
PDF (Adobe Acrobat) format.
Send complaints to rasmith@tamu.edu
In his 1961 paper "Tithenai ta Phainomena",1
G. E. L. Owen addressed the problem of the relationship
between science as preached in the Analytics and the practice of the
Aristotelian treatises. However, he gave this venerable crux a novel twist by
focusing on a different aspect of the issue. According to the Prior
Analytics, it appears that the first premises of scientific demonstrations
must be obtained from collections (historiai) of facts derived from
empirical observation. However, many of the treatises seem to make little use of
empirical inquiry and instead concern themselves more with 'conceptual
analysis.' This is especially true in the Metaphysics and the ethical
treatises, but it is also very much characteristic of the Physics. How
are these two kinds of inquiry related?
Owen took his cue from Aristotle's customary procedure of beginning his account
of a subject with a diaporia, a survey of the available data and the
views of others (including his predecessors) in which he notes the 'puzzles' (aporiai),
i.e. inconsistencies and paradoxical consequences, that result from them. Owen's
unifying proposal was that both empirical data and the opinions of others can be
described as 'appearances' (phainomena), once we recognize that there is
a crucial ambiguity in this term: not only 'what is apparent' (data of empirical
observation) but also what 'appears to' people (the opinions of people, at least
those with some level of general acceptance or philosophical currency). It is
then possible to describe Aristotle's overall method in philosophical inquiry as
beginning with the appearances and undertaking to resolve the puzzles, while
retaining as many of those appearances as possible. Moreover-and crucially-Owen
was able to assimilate Aristotle's philosophical method to dialectic, on a
certain understanding of that term. If, as he supposed, dialectic is 'argument
from
endoxa,' where endoxa in turn are commonly-held opinions, then
it is to be expected that some form of dialectic will be the source of the first
premises of scientific demonstrations.
This article, like so much of Owen's work, initiated a wealth of further
studies. It has now become a commonplace among many interpreters that
Aristotle's method of inquiry was dialectical, where that in turn means roughly
what Owen took it to mean. Terence Irwin, in particular, has developed a
modified form of Owen's view in extensive detail.2
To confront the problem how argument from 'common beliefs',
which are not necessarily all true, can establish the most secure of principles,
Irwin first distinguishes between 'pure dialectic,' which is simply argument
from the opinions of 'fairly reflective people after some reflexion', and a more
critical type of argument he calls 'strong dialectic.' The latter, according to
Irwin, emerges in the Metaphysics and elsewhere as a way of securing the
otherwise indemonstrable first principles on which scientific demonstrations
rest: Aristotle came to see the inadequacy of the appeal to 'intuition' for the
justification of these first premises which he had made in Posterior
Analytics II.19 and sought its replacement in a reconsidered form of
dialectical proof.
Irwin's defense of this picture rests in part on detailed analyses of the
argumentative procedures of the treatises. To challenge these would require a
comparably detailed study of these analyses. I do not claim to offer anything
like that here. Instead, I want to concentrate on the more direct textual basis
for the view that Aristotle thought dialectic, defined as a technique of arguing
from a special class of premises called endoxa, could provide a form of
justification of the first principles of sciences. Generally speaking, that
evidence can be divided into two parts. First, there are several proof texts for
certain critical theses Second, there is a certain picture of the endoxa,
the 'common beliefs' on which dialectical arguments rest. I would like to raise
some problems for each of these. My claims are that (1) if we take account of
their contexts and what they actually say, the proof texts turn out simply not
to support the claims which have been built on them; (2) the interpretation of
the endoxa of dialectic rests on a serious misunderstanding both of what
dialectical argument is and what the goal of the Topics is. In making my
limited case, I will return to the fons et origo, Owen's article, and in
particular to the proof texts on which it rests.
The fundamental supporting texts for Owen's view are Nicomachean Ethics VII.1, 1145b6-7, and Physics IV.4, 210b32-211a7, which by his account both contain ringing declarations of a dialectical method for establishing principles:
'For if the difficulties are resolved and the endoxa are left standing', as Aristotle says in both the Physics and the Ethics, 'this in itself is a sufficient proof.' (244)
The clear implication of Owen's quoted paraphrase is that the
Physics and the Ethics advocate a common method and that the
method in question consists entirely of 'resolving the difficulties and
preserving the endoxa.' I do not think that either passage actually
supports that interpretation. If we look carefully at them in their contexts, we
will find that the two are making distinct claims, each more modest than what
Owen imputes to them.
Let me begin with the Physics passage. Aristotle says:
We must try to conduct our search so that the essence will be given in such a way as to solve the puzzles, and what appears to be true of place will be true of it, and moreover the cause of resistance3 (duskolia) and puzzles about it will be evident. For this is how anything might be most beautifully shown (houtô gar an kallista deiknuoito hekaston).4 (211a7-11)
Owen presumably is offering us paraphrase, not translation, but still there is a
poor fit between his account and what this passage says. The former leads us to
believe that Aristotle is stating minimum sufficient conditions for establishing
something: 'this in itself is sufficient proof.' However, Aristotle says that he
is giving conditions for the 'finest' or 'most beautiful' way to prove anything.
These are counsels of perfection, not minimal conditions of adequacy; they go
far beyond mere sufficiency to tell us what the best of all possible outcomes
is. Moreover, he does not tell us that this is how we must conduct our search if
it is to issue in proof: he says that this is how we ought to
try to conduct it. It is consistent with this demand that an adequate
proof may fail to achieve some of these desiderata. He does not say that this is
how we may prove, or how we
must prove, but rather that this is how we should try to conduct
an inquiry: if indeed we can not only solve the puzzles and leave standing what
is thought to be true, but also explain what causes the difficulties in the
first place and why people have problems with them, then we will have the finest
proof imaginable. Nothing that Aristotle says implies that these are necessary
conditions for an inquiry, or even that they are attainable in every case. Since
elsewhere (for instance Metaphysics A,
NE X) he clearly thinks that sometimes our pre-philosophical opinions
cannot be retained after philosophical inquiry, we ought to suppose that the
outcome envisaged here will only be possible in some most happy sets of
circumstances.
Aristotle also includes a third requirement: we should try to make evident 'the cause of the resistance and puzzles about it.' A passage from the Eudemian Ethics elaborates this same requirement more fully. Since it has often been seen as providing an epistemological reason for attaching weight to the opinions of 'the many and the wise,' I will quote it in full:
We must try, by argument, to reach a convincing conclusion (zêtein tên pistin) on all these questions, using, as testimony and by way of example, what appears to be the case. For it would be best if everyone should turn out to agree with (phainesthai sunomologountas) what we are going to say; if not that, that they should all agree in a way and will agree after a change of mind (hoper metabibazomenoi poiêsousin); for each man has something of his own to contribute to the finding of the truth (echei gar hekastos oikeion ti pros tên alêtheian); and it is from such <starting-points> that we must demonstrate: beginning with things that are correctly said, but not clearly, as we proceed we shall come to express them clearly, with what is more perspicuous at each stage superseding what is customarily expressed in a confused fashion.5 (1216b26-35)
I have cited Woods's translation6
here, though I shall quarrel with it in a moment. Now, there is a clear
resemblance between this passage and Physics 211a7-11 on several points.
Again, Aristotle says that we should
try to achieve a certain result, not that we must achieve it. Next, he
indicates different degrees of success at which we may aim: best of all if
everyone 'turns out' to agree with our conclusions, second best if they agree
'in a way.' However, he now adds an explanation why we should aim at this:
everyone has 'something of his own in relation to the truth' (oikeion ti
pros tên alêtheian), and it is from this that demonstrations
must start. Thus, as it seems, Aristotle holds that everyone has a
certain built-in grasp of a little of the truth, and we should therefore treat
the opinions of the many, as well as those of the wise, with respect, in need of
correction and clarification rather than refutation and rejection. Aristotle is
not simply advocating a dialectical method in philosophy but giving us a reason
for doing so.
However, appearances cannot always be trusted. There are clues in the language of the passage that he has something quite different in mind. One such clue is the word metabibazomenoi: 'after a change of mind,' in Woods's version. This is not a very common word. Aristotle uses it only here and in two places in the Topics, each of which sheds important light on our present passage. First, in Topics VIII, he is discussing the use of false as well as true premises in arguing dialectically. For Aristotle, one thing which distinguishes dialectical arguments from 'contentious' or 'sophistical' ones in that they must be valid. However, their premises may sometimes be only apparently true. Indeed, in some cases they must be false: if the task at hand is to argue for a false conclusion, then false premises must be used (otherwise the argument would be invalid and therefore not dialectical). Aristotle goes on to note that it can also be dialectically appropriate to use false premises in establishing a truth or refuting a falsehood:
Sometimes, even if a falsehood has been supposed, it should be refuted by means of falsehoods. For nothing prevents things which are not so seeming more so to some individual than what is true, so that if the argument arises from what seems so to that person, he will be more effectively persuaded or benefited. And whoever changes minds (metabibazonta) well must change them dialectically, not contentiously (just as the geometer must do so geometrically), no matter whether the conclusion drawn is false or true.7(161a30-36)
The point is that dialectical arguments are always directed at someone and rely
on that person's opinions. If my goal is to persuade you, it will do me no good
to use true premises which you do not believe; I would be better off, in fact,
using false ones you did believe, so long as they led to the result I wanted. To
'change minds' is to lead people to have different beliefs, and that can only be
accomplished rationally by beginning with beliefs they actually do have.
Aristotle says much the same thing about 'changing minds' and dialectical argument in his remaining use of metabibazein, near the beginning of the Topics:
[Our dialectical method is useful] in connection with encounters, because if we have reckoned up the opinions of the many, we will speak to them not from foreign opinions but from their own, changing their minds about anything they do not seem to us to have said well.8 (101a30-34)
I take 'encounters' here to mean simply any occasion for argument with the
public. A compilation of the opinions of others, as we shall see below, is one
of the components of the dialectical method it is the purpose of the Topics
to present. Again, Aristotle's point is that we can only change people's minds
argumentatively from opinions they actually accept.
These passages should lead us to see the EE passage in a different
light. Getting everyone to agree with our view after a 'change of mind' means
leading each person, from premises he accepts, to accept our view. This
is intrinsic to the persuasive function of argument. In each of our individual
collections of opinions there is bound to be something true-'something of our
own', oikeion ti, with some relation to the truth-and this is the
starting-point from which others can persuade us to believe the truth. (Woods's
translation obscures this by paraphrasing
oikeion ti pros tên alêtheian as 'something of his own to
contribute to the finding of the truth': Aristotle says nothing about
finding or contributing to a search for the truth.) In short, then, what
Aristotle is talking about in the EE passage-and in the Physics
as well-is not discovering the truth but persuading others to believe it.
EN VII.1 comes much closer to presenting a general methodology for establishing first principles. Irwin9 translates it as follows:
As in the other cases we must set out (tithentas) the appearances, and first of all go through the puzzles. In this way we must prove the common beliefs about these ways of being affected-ideally, all the common beliefs, but if not all, then most of them, and the most important. For if the objections (ta duscherê) are solved, and the common beliefs are left, it will be an adequate proof.10 (1145b3-8)
In contrast to the Physics passage, Aristotle is clearly talking about
what is sufficient as proof here. Even so, we do not find anything quite so
strong as Owen's 'this in itself is sufficient proof'. In fact, it is difficult
to be certain that Aristotle is stating anything like a general
condition of adequacy for all proofs. If instead he is pointing to something
specific about the case at hand, then his intent may be quite the opposite: he
may be saying 'if we can accomplish this much, it will at any rate be enough.'
The discussion above of the Physics and Eudemian Ethics passages
gives this possibility more substance. To begin with, note where this passage
occurs. It is not at the beginning of a treatise but rather prefaces a section
devoted to an unusual topic: the possibility of weakness of will. One unusual
element is that the conclusion Aristotle will ultimately defend in this case
includes a rejection of a commonly held belief: like Socrates, Aristotle holds
that strictly speaking no one can possibly act incontinently. It is striking
that this passage, which has come to be a sort of locus classicus for
finding a dialectical methodology in Aristotle, is the prelude to the rejection
of a very widely held belief. However, if instead Aristotle is concerned with
'changing minds,' then its occurrence here is especially apt. It is prima
facie implausible that there is no such thing as weakness of will; if we are
ever to persuade others of this, we must begin from their own views. However, if
we can eliminate the difficulties (duscherê) that stand in the way
of their accepting it, then we will have shown them adequately (dedeigmenon
an eiê hikanôs) that it is so.
I do not want to claim that these considerations absolutely rule out Owen's
interpretation of EN 1145b3-8. In isolation, the passage does strongly
appear to be advocating a general method of inquiry. However, as we have seen,
elsewhere Aristotle uses similar language when his real concern is with
persuasion rather than proof. I conclude, therefore, that there is no necessity
for Owen's interpretation, and good reasons to reject it.
None of this denies that Aristotle regularly makes use of surveys of the puzzles inherent in the views of his predecessors and of people in general. Indeed, he uses these surveys as a guide to the development of his own position: the puzzles set questions to be answered, and good answers are those which account for all the puzzles. What is wanted, however, is some evidence that this type of resolution of puzzles constitutes a proof.
Underlying these attempts to find a dialectical method of proof in Aristotle is
a conception of what dialectic is; and here, in my opinion, is where the real
trouble lies. Interpreters have generally accepted an account of dialectic which
places all its emphasis on the curious compilation of opinions called endoxa-the
views of 'the many or the wise,' to give it its briefest of several disjunctive
formulations. Precisely why Aristotle should single out this collection of
opinions for attention is as vexing a question as how he thinks the principles
of sciences are established. Following Owen, many interpreters have tried to
link the two, not only finding in dialectic a source for the principles but also
finding in this function of dialectic a partial explanation of the attention
Aristotle pays to the endoxa. Irwin has developed a particularly rich
interpretation along these lines. Since the picture of dialectic it includes has
been influential, I will turn my attention to it.
Irwin finds a definition of dialectic in the first sentence of the Topics:
Dialectic is 'a method from which we will be able to syllogize from common beliefs (endoxa) about every topic proposed to us, and will say nothing conflicting when we give an account ourselves.' (36)
But in 100a18-20, Aristotle does not say that he is giving us a definition of
dialectic at all. Instead, the words Irwin quotes are prefaced by 'The goal of
our treatise is to find a method …' Is Aristotle then claiming that
in the Topics he means to discover dialectic? The answer, as I have
argued elsewhere, is that there is a difference between dialectical argument
and a dialectical method or art. Dialectical argument existed
long before Aristotle; he himself credits Zeno of Elea with its discovery. What
he offers in the Topics is an
art of dialectic, to stand in the same relationship to dialectical
argument as does the art of rhetoric to orations.
Failure to recognize this distinction is, I think, the major cause of
misunderstanding Aristotle's remarks about endoxa. In the Topics,
he defines a dialectical deduction (sullogismos) as one with premises
which are endoxa, as opposed to the 'true and primary' premises
characteristic of demonstrations. If we concentrate only on this point, we may
wonder exactly what this class of endoxa is. However, in other places
Aristotle notes another equally important difference between dialectical
arguments and demonstrations: the premises of dialectical arguments are questions
(An. Pr. 24a25, Top. 104a8). What differentiates dialectical
arguments from demonstrations is that there are two parties to a dialectical
argument, one of whom presents the argument to the other as a series of
questions held out for acceptance or rejection.11
In demonstration, one chooses as premises the true and primary propositions
which underlie all further truths about the subject matter at hand; no audience
is necessary, and it is the task of the learner not to question but to accept.
However, in a dialectical argument, the questioner can only develop an argument
on the basis of an answerer's responses. Consequently, the questioner must take
account of the opinions of that answerer and whether the premises needed are
acceptable (endoxos) to the answerer.
In the Topics, it is Aristotle's purpose to develop a method that can be
used in any such dialectical context. He pays particular attention, especially
in Book VIII, to a stylized form of debate that apparently was practiced in the
Academy (he calls this 'gymnastic'), but he does not lose sight of the wider
applicability of his method to any context in which arguments must be developed
that depend on the opinions (or at least responses) of others. This, for
instance, is the reason why he tells us in the Rhetoric that the
rhetorical art is a kind of hybrid of the dialectical art (dialektikê)
and ethics: orations are directed at individuals, and we must understand the
opinions they have in order to persuade them.
Now, Aristotle makes it clear that a dialectical argument, as he understands it,
must be a valid argument (sullogismos), that is, its conclusion must
actually be entailed by its premises. This sets one task for an art of
dialectic: we need a method for discovering premises that imply a given
conclusion. But since dialectical arguments must be constructed from the
concessions of a respondent, premises will be useful to us only if they are
accepted by our interlocutor. This sets a second task for the dialectical art:
we need a method for determining, or at least predicting, which premises our
respondent will accept. The first task can be solved if we have something like a
theory of validity, or at least some general rules relating conclusions to
premises. The second task would be solved if we had a systematic classification
of premises according to the type of person who will find them acceptable. This,
in my opinion, is exactly what Aristotle means by the endoxa. The
dialectician is to collect views of types of person-the views of everyone, of
the many, of the wise, of various celebrated wise men-and use them to gauge the
acceptability of premises to a particular opponent.
The clearest statement of this project comes, not from the Topics, but from the Rhetoric:
For since what is persuasive is persuasive to someone (and sometimes is directly persuasive and convincing through itself, sometimes because of appearing to be proved through such things), but no art investigates the particular (e.g., medicine does not investigate what is healthful for Socrates or Callias, but rather what is so for this type or these types of person-for this is artful, but the individual is infinite (apeiron) and not knowable (epistêton)), then neither will rhetoric study what is individually acceptable (to kath' hekaston endoxon), e.g., to Socrates or to Hippias, but rather what is so <sc. endoxon> to such-and-such people (tois toioisdi), just like dialectic.12(1356b28-35).
This is a familiar Aristotelian thesis: art and science are of the universal, not the individual. It is only accidentally that medicine studies what is healthful for Socrates. Likewise, dialectic, if it is to be an art which studies what is apparent, must study what is apparent to types of person, not to this or that individual except incidentally. Against this background, consider the statement often quoted as giving a definition of endoxa:
The endoxa are what seems so to everyone, or to most people, or to the wise (and of them, to all, or to most, or to the most famous and best accepted).13 (100b21-23)
The great majority of interpreters, regarding this as defining a single class of endoxa, then are much exercised to understand what that class might be and why Aristotle assigns it such importance. But compare the definition of 'dialectical premise' Aristotle gives a little further on in Top. I.10:
A dialectical premise is a question that is acceptable (endoxos) to everyone, or to most people, or to the wise (and to all of them, or most of them, or to the most famous)14 (104a8-10)
Here, Aristotle qualifies the term endoxos with almost the very same words used in 100b21-23. If that earlier passage is a definition of endoxos, then this one is, in Aristotle's terms, an example of dis tauto legein. It seems much more likely that the first passage is not a definition but a clarifying enumeration. That is, Aristotle says: the premises of a dialectical argument must be acceptable; by 'acceptable' I mean what seems so to everyone, or what seems so to most people, or what seems so to the wise … We find some confirmation of this in the continuation of the passage:
… so long as it is not paradoxical: for someone will concede (theiê)15 what seems so to the wise, if it is not contrary to the opinions of the many.16 (104a10-12)
In other words, Aristotle gives as a reason for including the opinions of the
wise among endoxa the fact that people will usually accept these (with
an appropriate citation of authority) if they are not 'paradoxical'.
Once we realize that dialectical argument and the dialectical art are distinct,
the mystery about endoxa completely disappears. Aristotle's entire
purpose is to spell out an art for arguing successfully with other people on the
basis of their opinions. Part of what that art must include is a study of the
opinions of various types of person. The endoxa of the dialectical art
are simply lists of opinions, categorized in this way.
This interpretation is at variance with the understanding Irwin and others have of the endoxa. Supposing that Aristotle is trying (for unclear reasons) to work out a specialized kind of argumentative method, relying on aspecial class of opinions, Irwin seeks an account of just what those opinions are. He concludes that in fact, Aristotle wants to restrict the endoxa to the opinions of 'fairly reflective people after some reflexion' (38). In defense of this, he adduces two passages. The first is from On Sophistical Refutations: the common beliefs of dialectic must be apparent,
but apparent not to just anyone, but to people of a certain sort (tois toioisde); for it is an indefinitely long task to examine the things that make something apparent to just anyone.17 (170b6-8)
The phrase 'people of a certain sort' strongly suggests that the point here is
to restrict the relevant opinions to the opinions of a certain class of people.
However, Aristotle's own explanation of this limitation already undercuts
Irwin's interpretation. If his purpose were to say that dialectical arguments
rest on the common opinions of a specific class of persons, we would expect him
to say something like 'for there is no point bothering with the opinions of
certain types of people.' What he actually says, however, is that the task of
determining 'the things that make something apparent to just anyone' is too
indeterminate to pursue-'an indefinitely long task,' in Irwin's rendering. And
as we have just seen, almost exactly the same phrase (tois toioisdi) is
found in Rhetoric 1356b35, where it clearly is intended to indicate
types of person as opposed to individuals.
To find an explicit declaration that 'not everyone's opinions count equally,' Irwin turns instead to Eudemian Ethics 1214b28-1215a2, where Aristotle says that we should not waste our time examining the opinions of children, the ill, and the mad about happiness, nor indeed should we give any special weight to the opinions of 'the many,' since 'they speak haphazardly about practically everything, and especially about happiness.'
Now, this is already a remarkable passage from which to seek support for a view of dialectical argument as resting on, and giving initial weight to, the opinions of 'the many or the wise,' since it amounts to a global dismissal of the opinions of the many concerning 'practically anything.' Furthermore, its entire weight, with respect to Irwin's thesis, is negative, showing only that Aristotle thought it 'a waste of time' (periergon) to examine what most people think. If the passage is to provide evidence that Aristotle thinks there is a certain class of persons whose opinions do count, we should expect it to say who they are. Instead, Aristotle says:
But, as each inquiry has its own problems, so, evidently, does that concerning the best and highest life. It is these opinions, then, that it is right18(kalôs echei) for us to investigate (exetazein); for the refutation of those who dispute a certain position is a demonstration of the opposing view.19 (1215a3-7, Woods's translation)
Aristotle has said earlier that argument about ethical matters is not
what is wanted in the case of the immature, the unsound, and the immoral: what
they need instead is experience, therapy, or punishment, as appropriate. His
point here is quite precise. Argument serves to change the opinions of others by
taking their own opinions as its starting point and showing them that these
opinions entail other views; if those further views are repugnant to them, they
are thereby motivated to change their opinions. Those whose opinions are subject
to rational modification in this light are candidates for argumentative
persuasion, and to that end it is useful to study their opinions and the
arguments that can be constructed from them. However, children, the insane, and
the wicked lack the opinions from which rational persuasion might begin; it is
pointless to consider how to refute their views, since what causes people of
these classes to have their opinions is not argument but some form of pathology.
Aristotle takes part of this to be obvious: we cannot make a child an adult by
argument, nor can we heal the sick (or convert the wicked, for that matter). The
rhetorical strategy of the passage is a comparison of these cases with 'the
many': we cannot change their views by argument because they do not hold their
views for reasons in the first place but only 'speak haphazardly' (eikêi
legousi) about pretty much everything. Therefore, there is no more point in
trying to argue with them than with children or the insane, and we may forgo
discussion of their views.20
In summary, the EE passage does indeed say that only
the views of certain people count, but that passage is not making anything
approaching a general claim about either philosophical or dialectical argument.
Aristotle is giving a narrowly-focused reason for ignoring the views of the mass
of humanity about what life is best: they do not have reasons for their
opinions, so we need not worry about refuting them. Nothing about the context
licenses the elevation of this into a general methodological principle. There is
no warrant for using it to explicate SE 170b6-8.
Moreover, if we look carefully at the context of the SE passage, it becomes clear that Irwin's translation fails to capture an important implication of the term 'indefinite'. Aristotle's purpose in the entire passage is to explain which refutations his treatise can legitimately study and which it cannot. He begins by stating a distinction, familiar from other works (including the Rhetoric as well as the Topics), between arguments that fall under the purview of dialectic and arguments proper to one of the sciences. Refutations are arguments (sullogismoi); therefore, if we know what kinds of things (par' hoposa) arguments arise from-i.e. what kinds of premises-then we will thereby know what refutations arise from. But arguments are classified according to their premises: some are 'in accordance with a particular art', i.e. rest on premises peculiar to that art, whereas others are general. It is only the latter which fall under dialectic. As a result, says Aristotle, to undertake a completely general study of how all refutations come about would require having a science of everything, which is not the task of any single science:
For the sciences are likely infinite in number (apeiroi), and consequently so are demonstrations. But these are refutations, and true ones: for whenever something can be demonstrated, it is also possible to refute one who accepts the denial of this truth. For instance, if someone accepts that the diameter is commensurate, someone could refute him with a demonstration that it is incommensurable. Thus, we will have to be scientists about everything.21
If every refutation corresponded to a demonstration, and if
every demonstration were proper to some science, then there could be no such
thing as the study of refutations, except in a Pickwickian sense: studying
refutations in general would require scientific omniscience. However, Aristotle
holds that there are some refutations that depend only on certain 'common'
premises which are not peculiar to any given. He sees these as being the special
province of dialectic and its cousin, rhetoric.22
Aristotle continues:
Clearly, then, it is not the topoi of all refutations that are to be grasped, only of those that arise from dialectic: for these are common to every art and faculty. Moreover, the study of a refutation in accordance with a specific science is for the person who possesses that science, both as to whether it appears to be one but is not and, if it is one, why it is. But a refutation from common <premises>, which fall under no science, is for the dialectical to study. For if we have what <topoi> the accepted deductions (endoxoi sullogismoi) about something are from, then we also have what <topoi> the refutations are from: for a refutation is a deduction of the contradictory.23
The point Aristotle makes here is crucial to his
understanding of dialectical argument. In the logical works and the Rhetoric,
(1358a2-35) he differentiates the arguments, premises, and refutations proper to
the individual sciences from the 'common' ones applicable to all sciences. The
latter, precisely because they are common, cannot serve as the basis of any kind
of scientific proof: each science is autonomous with respect to demonstrations
about its subject matter, and there is no single overarching science embracing
them all. Here, Aristotle draws from this the corollary that, since a refutation
is just a kind of deduction, the theory of refutations that concern any subject
falls under the science of that subject, not a general science of refutations.
Having made all these distinctions, Aristotle then reaches the conclusion he wants: the general study of refutations applicable to all subject matters is part of the dialectician's art. In the course of stating this conclusion, he makes in passing the remark Irwin takes to be restricting the starting points of dialectic:
Thus, we possess the kinds of <premises> which such refutations are from. And if we have that, then we also have their solutions: for the objections to these are solutions. And we also have what apparent refutations are from (but apparent not to just anyone, but to people of a certain sort: for they would be indefinite if someone were to inquire from how many <kinds> they appear <to refute> to just anyone). Thus, it is evident that it is for the dialectician to be able to grasp from what sorts of premises either a real or an apparent refutation arises through the common <topoi>,…24
It should now be obvious that this phrase actually has a
completely different purpose from the one Irwin ascribes to it. Aristotle is not
talking about apparent propositions but about apparent deductions,
since his overall goal is to argue that there is a class of refutations (real
and apparent) which fall under the scope of dialectic rather than any special
science. At the very least, it is a distortion of emphasis to wrest this
qualification out of its context and see in it a defining characteristic of
dialectical argument itself.
It appears, then, that neither of these passages will bear the interpretive
weight Irwin requires of it. The SE passage simply does not mean what he
takes it to, and the EE passage is not about dialectic. Neither provides
any evidence that the starting points of dialectical arguments are 'the opinions
of fairly reflective people after some reflexion'
Let me now turn to what is perhaps the most crucial text linking dialectic with the indemonstrable first premises of scientific demonstrations. In Topics I.2, Aristotle gives three venues in which dialectic will be useful: 'gymnastic' exercises, 'encounters,' and 'the philosophical sciences.' I have alluded to the first of these briefly above and discussed his remarks about the second. The third appears to be of much more moment. Here is how Irwin translates it (36):
It is useful for the philosophical sciences, because if we fully examine the puzzles on each side (diaporêsai), we will more easily see what is true or false. And it is also useful for <finding> the first principles of each science. For we cannot say anything about them from the proper first principles of the science in question, since the first principles are prior to everything else. Hence it is necessary to discuss them through the common beliefs on each subject. And this is proper to dialectic alone, or to it more than to anything else; for since it examines, it has a road towards the first principles of all disciplines.25
Owen takes this to be an explicit declaration that dialectic
'establishes' the first principles of the sciences (244). Irwin, more
circumspectly, repeats Aristotle's phrase that dialectic 'has a road' to the
first principles. In either case, it appears that something much more is being
imputed to dialectic than could possibly be accomplished by the art as I have
described it.
However, context once again turns out to be important. Is Aristotle talking
about dialectical argument itself, or is he talking about the dialectical art?
The opening of the section (101a25-26) leaves no doubt that he means the latter:
he refers expressly to the
pragmateia, i.e., either to the treatise he is writing or to its
contents. These are uses, then of the dialectical art. Since that art, in turn,
includes a study of logical consequence, it is evident at once how it will be
useful in connection with 'examining the puzzles' concerning any issue: skill in
deducing the consequences of a position is a natural concomitant of skill in
deducing conclusions from an opponent's opinions. In any event, Aristotle does
not say that this will provide us with a proof of anything but only says
that it will help us to see what is true or false. It is consistent with this
claim that other means are also required for discovering the principles, and it
is consistent with the business of the Topics that Aristotle should
refrain from discussing them here.
Aristotle proceeds to give a second use of dialectic in connection with
scientific principles, one which looks at first much more promising-especially
in Irwin's translation, which glosses Aristotle's simple 'in connection with' (pros)
with 'finding' (though as he notes this is not in Aristotle). We seem to find
here an allusion to the doctrine of the Posterior Analytics that no
demonstrative science can demonstrate its own principles. So construed, this
passage promises to be of great value, since it would be speaking to the same
point as the enigmatic final chapter of that work. However, Aristotle simply
does not say here that dialectic establishes the principles: he says
that since a science cannot 'say anything about' its own principles, we must
'discuss' them (dielthein: Aristotle often uses the word of his
preliminary discussions of the views on a subject) by means of endoxa.
But there is a considerable distance between 'discussing' first principles and
'establishing' them.
If that distance is to be bridged, the weight must fall on the last sentence of
the section, which Irwin and others take to say that dialectic provides us with
a road whereby we reach the firsts principles. Now, in what sense might we be
said to arrive at the first principles? Aristotle's own view on this is clear.
The first principles of the sciences are truths which have a certain objective
priority to the conclusions that can be demonstrated from them. This priority
does not depend on our own epistemic reactions in any way: it is not only
possible, but usually true, that we all find these first principles unconvincing
or even absurd before we have acquired scientific understanding. On Aristotle's
view, it is by transforming ourselves, so that the objective first principles
seem to us to be primary, that we come to have scientific understanding. There
are, consequently, two components to reaching the first principles: (1) finding
out what they are, and (2) coming to see them as the principles.
There is considerable evidence in the Analytics that Aristotle thought
(1) could be accomplished in a way that sounds quite strange to us. He thought
that he had discovered a theory of inference (the syllogistic) which covered
all valid arguments whatsoever, and thus all demonstrations. He also
thought-and understandably so, given the properties of his theory of
inference-that there are some true propositions with the unusual property that
they cannot be deduced from any other true propositions, though they may
serve as premises from which to derive others. Indeed, he appears to have held
the even stronger claim that if we collect together all the truths about any
subject, we will find that there are certain truths among then which cannot be
deduced from any combination of the others but from which all the others can be
deduced. These propositions must be the principles for the simple reason that
they cannot be anything else: they cannot be demonstrated. On the other hand, if
they are included among the principles, then we need no other principles, for
all others follow from them. The upshot is that Aristotle believed there was a
way to specify the principles without appealing to epistemic status.26
According to the Analytics, then, dialectic is not
the means of accomplishing (1). What of (2)? Here, I think, we do discover an
important role, but it is not the one Owen supposes it to be. The property of
dialectic which Aristotle appeals to here is that it 'examines' (exetastikê
gar ousa). The word used here for 'examine' is closely connected with
refutation, in particular refuting someone else's views by showing that they
lead to contradictions (Socrates used it of his customary style of questioning
people about their opinions). A process of refutation is not a very likely
candidate for establishing the first principles.27
What it might do, however, is bring about a considerable change in our own
epistemic situation. The process of exploring the contradictions implicit in our
naïve opinions eliminates the air of certainty that attaches to them and
puts us in that unpleasant state of 'wonder' which, according to Aristotle, is
the beginning of philosophy. A continual process of exploring what follows from
what could plausibly be essential to the kind of epistemic conversion required
if we are to become scientific. Aristotle tells us in Met. Z that
scientific education resembles moral education: we strive in each case to change
our untutored reactions, making what is in itself good (or familiar) come to
seem good (or familiar) to us. In ethics, the agent of this transformation is
habit, born of repeated action. Repeated examinations of opinions and their
consequences, and of the principles and their consequences, could be the agent
of a similar epistemic transformation.28