Philosophical Review

Pragmatism, Categories, and Language

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

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 [email protected].

Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend access to The
Philosophical Review.

http://www.jstor.org

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM, CATEGORIES, AND LANGUAGE

PRAGMATISM is getting respectable again. Some philos-

ophers are still content to think of it as a sort of muddle-

headed first approximation to logical positivism-which they

think of in turn as a prelude to our own enlightened epoch.

But those who have taken a closer look have realized that the

movement of thought involved here is more like a pendulum than

like an arrow.1 This renewed interest in pragmatism has led to

a new interest in Peirce, who somehow seems the most "up-to-

date" of the pragmatists,2 and whose work in logic permits

one to call him muddle-headed only if one is also willing to call

him schizophrenic. But students of Peirce, even the most sympa-

thetic, have had trouble digesting what he called his "Scotistic

realism" and his categories of Firstness, Secondness, and Third-

ness. These are obviously central features of his thought, yet

they do not seem to sit well with his pragmatism. Still, Peirce

insists over and over again that "the validity of the pragmatic

maxim" and "Scotistic Realism" mutually entail each other,

and he suggests that they are both expressions of "the irreduci-

bility of Thirdness."3

My purpose in this paper is to try to show that the point

Peirce is making in this identification is sound and important.

Focusing on this point shows how far Peirce was in advance of

the positivism of his day and how close his views are to the present

trends in philosophy which have arisen in reaction to the more

sophisticated positivism of Wittgenstein's Tractatus and of the

Vienna Circle. I want to suggest that Peirce's thought envisaged,

and repudiated in advance, the stages in the development of

1 Cf. Morton White, Toward Reunion in Philosophy (Cambridge, Mass., I956),

pp. 268 if., and Alan Pasch, Experience and the Analytic (Chicago, 1958).

2 Perhaps because he was neither as concerned with religion and morality

as James, nor as interested in social and political issues as Dewey.

3 Cf. Collected Papers of Charles Sanders Peirce, 8 vols. (Cambridge, Mass., I 93 I -

I958), 5.453, 5.470, 5.527, 5.503, 5.4 (on pragmatism and synechism), 5.469

(on pragmatism as an outcome of the logical derivation of Peirce's categories),

I.26. Hereafter references to the Collected Papers (by volume and paragraph

number) will be inserted in the text.

I97

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTr

empiricism which logical positivism represented, and that it

came to rest in a group of insights and a philosophical mood much

like those we find in the Philosophical Investigations and in the

writings of philosophers influenced by the later Wittgenstein.

A little empiricism, plus a passion for rigor, will make a man a

nominalist. Thinking about the antinomies created by the mutual

repugnance of experience and rigor will drive him, if he thinks

as long and as hard as Peirce and Wittgenstein did, to something

quite different. In trying to show that this "something different"

was pretty much the same for both men, I shall argue for the

following points:

(i) What Peirce called "nominalism" and what present-day

philosophers call "reductionism" are forms of a single error.

(2) The error in both cases goes back to "the Protean meta-

physical urge to transcend language."4

(3) Peirce's attempt to give sense to the notion of universalia

ante rem is not a result of succumbing to this urge, but is rather

his device for repudiating it as strongly as possible.

(4) When Peirce says that "vagueness is real" and when

Wittgenstein points to the difference between causal and logical

determination, the only differences between what they are saying

are verbal (or, to give the cash value of this overworked word,

uninteresting).

(5) The similarity of their insights about language reflects

that fact that the slogans "Don't look for the meaning, look for

the use" and "The meaning of a concept is the sum of its possible

effects upon conduct" reciprocally support each other.

Before proceeding to these points, however, it may be useful

to remark that I am trying to show neither that Peirce saw through

a glass darkly what Wittgenstein saw face to face, nor the reverse.

One can take the first slogan mentioned in (5) as a special case

of the second, or vice versa; which way one sees it depends, and

should depend, on the purposes of one's inquiry at a given

moment. What I am trying to show is that the closer one brings

4 The phrase is taken from D. F. Pears's article "Universals," reprinted

in Logic and Language, Second Series, ed. by A. N. Flew (Oxford, I955).

More will be heard of this article in Section II below.

I98

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

pragmatism to the writings of the later Wittgenstein and of those

influenced by him, the more light they shed on each other.5

Peirce liked to refer to any doctrine he disagreed with as

"nominalistic." One of the dozens of different ways in which he

tried to formulate the common error of all nominalists was by

calling nominalism the doctrine that vagueness is not real.6

Nominalists thought, that is, that whatever was real had sharp

edges (like a sense datum or an atomic fact), and that whatever

did not have sharp edges could be "reduced" to things that did.

Most of Peirce's work was devoted to showing that this reduction

could not be performed. Among the vague things which, he

thought, nominalists could not reduce (and hence could not

account for consistently with their assumptions) were Intelligence,

Intention, Signs, Continuity, Potentiality, Meaning, Rules, and

Habits. All these he blithely baptized-to the perpetual delight of

neo-Pythagorean hedgehogs among his readers and the confusion

of all foxes-"Thirds." The point of the baptism was his claim

that phenomena which exhibit features referred to by some or

all of these capitalized terms have in common a certain peculiarity:

their adequate characterization requires a language which

contains, as primitive predicates, the names of triadic relations.7

5 In particular, Peirce and Wittgenstein complement each other especially

well; one presents you with a bewildering and wonderfully abstract apparatus

of categories; the other shoves you into very particular puzzles. Peirce's odd

numerological categories, just because they are so abstract and so far from the

cliches of the history of philosophy, are perhaps the best handles for grasping

what one learns from Wittgenstein. Conversely, Wittgenstein's riddles and

aphorisms, just because they are so fresh and fragmentary, let one see the point

of some of Peirce's darker sayings.

6 On the reality of vagueness as the thesis of scholastic realism, cf. 5.453. On

the opposition between nominalism and scholastic realism, cf. I . I 5 ff. On nomi-

nalism as the thesis that all vagueness is due to a defect of cognition, cf. 4.344.

7 Thus the translation of the metaphysical thesis that "Thirds are real" into

the "formal mode of speech" is: "No language will be adequate to reconstruct

the meaning of sentences referring to 'Intelligence,' 'Signs,' etc. unless it

contains as primitives the names of n-adic relations with n >3." Peirce claimed

that tetradic, pentadic, etc. relations could all be analyzed into triadic ones,

but that no triadic relation could be built up out of monadic and dyadic

relations. Cf. I.345 ff., I.363.

I 99

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD R OR T-

This claim involves two theses: (i) that triadic relations cannot

be built up out of monadic or dyadic ones and still retain their

original significance, and (2) that the phenomena in question

cannot be made intelligible in language which lacks names for

triadic relationships. I shall not be concerned with deciding

whether he was right about either of these points. This is because

I do not think (and here I differ from Peirce, who believed that

there were decision-procedures for these theses8) that either

(i) or (2) is the kind of thesis that can be decided. They cannot

be decided because both of them have built-in escape clauses

which permit one to reply to the offer of a counterexample:

"but that does not retain the original significance" or "but that

does not make it intelligible." The presence of such escape clauses

is not a defect in Peirce's thinking, but a characteristic of all

utterances which are intended simply to point to similarities and

yet find themselves forced into sentences about which the un-

answerable riddle "analytic or synthetic?" can be raised. To free

either thesis from these escape clauses, one would have to find a

criterion for knowing when the "same" significance is preserved.9

But even though it would be useless to try to prove either

thesis, it is useful to show how Peirce gave them plausibility. His

clearest example, perhaps, was the act of giving.10 If I give you

a book, can you describe my action "adequately" in terms which

avoid the prima-facie triadic character of the situation? Can you

replace the three-place predicate "giving" with a set of two-

place or one-place predicates? The obvious move is to try some

8 He would have said that (I) was a matter of formal logic (cf. 5.469,

I.345) and (2) of empirical fact (I.345). I should want to argue that in attempt-

ing a priori deductions of the categories (cf. 4.2 ff.) Peirce was unfaithful to

his own better insight when (in such passages as 5.36) he says that logic must

be founded upon aesthetics.

9 That one cannot get such a criterion is a corollary of Peirce's claim that

vagueness (Thirdness) is irreducible. If one notes that rules are Thirds, it can

also be seen as a corollary of Wittgenstein's remark that "The use of the word

'rule' and the use of the word 'same' are interwoven." (Philosophical Investiga-

tions, New York, I953, No. 225.) All future references to Wittgenstein will

be to Part I of the Philosophical Investigations, will be made by paragraph

number, and will usually be inserted in the text.

10 Cf. I.345, and compare Wittgenstein, No. 268, where he discusses why

your right hand cannot give something to your left hand.

200

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

such pair of dyads as "You shoved it toward me and I picked

it up." But something is missing. What? Well, roughly the same

sort of thing that is missing when I analyze "I lifted my arm"

into "First I had kinaesthetic sense datum a, and then I had

visual sense datum b, and so forth." It is probably also the same

sort of thing that is missing when one substitutes causes for motives

or tries to reduce "ascribing" to "describing."11 What is missing

is, in short, the kind of thing people mean when they talk of the

"meaning" of the action or of the "intention" behind it. Now

Peirce's way of describing the loss is that "Seconds" have been

substituted for "Thirds." To put it loosely, if something passes

from my hand to yours we are, in so far forth, just two things

bumping into one another in a somewhat complicated way. The

situation thus does not differ in any essential way from the

collision of two billiard balls (which is one of Peirce's examples

of pure Secondness). Putting it another way, the action can be

described in the same "language-stratum"12 as can the billiard balls.

This example perhaps makes clearer what Peirce meant by

describing nominalists as people who try to reduce Thirds to

Seconds. It should also suggest that these "nominalists"-who for

Peirce included just about everyone from Descartes to J. S. Mill,

with the possible exception of Kant-are the intellectual ancestors

of the "reductionists" whose downfall Mr. Urmson takes to be the

prelude to "the beginnings of contemporary philosophy." 13

These latter philosophers, best exemplified perhaps by the

Aufbau phase of logical positivism, did most of their reducing

in two main areas. One was the cluster of notions which center

around "intention." We have just seen how Peirce's apparatus

of categories is applied to a member of this cluster. The other was

11 Cf. H. L. A. Hart, "The Ascription of Responsibility and Rights,"

reprinted in Logic and Language, First Series (Oxford, I95 ), pp. I45 ff. Peirce

would have noted with delight the triadic character of ascription (I ascribe x

to you) and its difference from a simple description (I sawy). Notice that

"I saw you do that" is a true triad for Peirce only if it means "I sawyou ('you

responsible person' or 'you swine') do that"; if, on the other hand, it is replace-

able by "I saw the following sense data . . ." then it is a pseudo-triad.

12 I shall be using this term in the sense given it by Dr. Waismann in

"Language-Strata," Logic and Language, Second Series, pp. I i ff.

13 J. 0. Urmson, Philosophical Analysis (Oxford, I956), ch. Io.

20I

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTY

the epistemology of perception, where they restated phenomenal-

ism as the doctrine that anything sayable in the language of

material objects could be said equally well in the language of

sense data. Here too Peirce is antireductionist, despite the prima-

facie phenomenalist character of pragmatism.14 His explanation

of the failure of phenomenalism is, once again, that Thirds (in

this case material objects15) cannot be built up out of Firsts

(unsensed sensibilia) or Seconds (acts of sensing). Material objects

are permanent possibilities of sensation, and, as such, they have

the character of laws (I.487). Looked at from the side of the

knower, this point is made in the doctrines (i) that a percept is

always "excessively vague" (4.539) and therefore requires a

"logical" as well as an "emotional" (First) and an "energetic"

(Second) interpretant (5475 if.); and (2) that the logical inter-

pretant will, if it is simply some determinate image or other

mental state (or set of states), always require further inter-

pretation, and that therefore it must be something as indeter-

minate in its application as a law, namely, a habit (5.486).16 In

plainer language, one might explain what makes a batch of

sense data a cat by saying either that it means a cat to somebody,

or that somebody intends to take it to be a cat, or that somebody

follows a rule in terms of which it represents a cat, or that some-

14 Cf. Peirce's contrast between reductionist Humian phenomenalism and

Kantian or pragmatist phenomenalism (8.I5). The latter, which Peirce calls

"phenomenalism aufgehoben" (8.i86) turns out to be the sort of perspectival

realism which Roderick Firth puts forward in "Sense-Data and the Percept

Theory, Part II," Mind, LIX (I950), 34-55. Cf. especially pp. 48 if. of

this article on the "sign function" of the ostensible physical object, a topic

which Peirce developed at great length and which is at the heart of his episte-

mology. The relations between pragmatism and perspectival realism are

brilliantly exhibited by Pasch (op. cit., esp. chs. iv, vi).

15 On matter as Third, and as such opposed to mere quality or mere action,

cf. I.420.

16 The doctrine of the "ultimate" logical interpretant, and Peirce's semiotic

generally, contain many puzzles which we cannot touch on here. The dif-

ference between an infinite series of determinates and an infinitely determin-

able indetermination will be discussed in Section III below. Peirce's use of

this distinction is well treated by George Gentry in "Habit and the Logical

Interpretant" (Studies in the Philosophy of Charles Sanders Peirce, Cambridge,

Mass., I952, pp. 75 ff.; this collection of essays will be referred to hereafter

as Studies.)

202

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

body has a habit of saying "cat" when he encounters it, or that

somebody expects the usual laws describing the behavior of such

sense data to hold. Peirce's point is that all these italicized terms

are names for Thirdness, and that consequently any of them

may be analyzed in terms of another, but that none of them can

be reduced either to the sense data themselves (Firsts) or to the

merely dyadic relations which hold among sense data (for

example, such Seconds as spatiotemporal nextness and sheer

similarity17). Any "reduction" of cats to patches will, therefore,

miss the reference to a logical interpretant which makes the cat a

cat. It will lose the same kind of thing that gets lost when we

"reduce" giving to handing over and taking.

In applying the name "Thirdness" to all the things which

reductionists mislay, Peirce is trying to do in a wholesale way what

current antireductionist writers have been doing case by case.

The most fashionable antireductionist argument at the moment

runs as follows: reductionism represents a confusion of the meaning

of something (for example, a word) with the reasons which we

give for applying it in a given case. The cash value of this argument

is: certain statements which the unreduced item entails or other-

wise licenses are not entailed or otherwise licensed by the reduced

form of this item. Now the obvious come-back for the reductionist

is: tell me just what these statements are, and I shall fix my

reduction up to take care of them. And the clinching reply to

this is: we cannot tell you what they are, because there are an

indefinite number of them.18 Now this clincher is, as we shall

see in more detail later on, just what Peirce is insisting on when

he says that "there is no exception to the law that every thought-

sign is translated or interpreted in a subsequent one" (5.284) or

that "no collection of facts can constitute a law" (1.420) or that

"there is no absolute third, for the third is of its own nature

17 Not similarity in some given respect, for this would be triadic ("x resembles

y in being a z"). If sheer similarity seems unintelligible, Peirce would rejoin

that this unintelligibility just shows you that it is Second, and therefore brute

and unmediated.

18 Cf. Urmson on the difficulties of analyzing propositions such as "Britain

declared war on Germany" and his conclusion (op. cit., p. i6i) that "the

ancient doctrine of British empiricism that all non-simple concepts must be

reduced to complexes of simple concepts must finally go."

203

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTY

relative" (I.362).19 All these dicta amount to saying to the

reductionist: language is incurably vague, but perfectly real and

utterly inescapable.

II

Here, then, we see the way in which the two Peircian definitions

of "nominalism" coincide: to assert that Thirds can be reduced

to Seconds and Firsts is to deny that vagueness is real. Further,

we see how the denial of the reality of vagueness leads to reduction-

ism. Having now suggested that Peirce and "postpositivistic"

analytic philosophy have common enemies, I want in this section

to compare and contrast their approaches to a particular prob-

lem: that of naming. In doing so, we shall see what Peirce meant

by "Scotistic Realism." Further, we shall see how Peirce's old-

fashioned solutions of philosophic problems with the help of an

array of ontological categories resemble, when looked at closely,

contemporary "dissolutions" of these problems. As an example

of the modern dissolution of the problem of naming, I shall use

D. F. Pears's article, "Universals," whose concluding remark I

partially quoted above: "The desire to go on explaining naming

... is the result of the Protean metaphysical urge to transcend

language. 20

One achievement of Pears's masterly discussion is to show

that the same Sehnsucht impels the Platonist and the nominalist.

The one thinks "Nature but a spume that plays / Upon a ghostly

paradigm of things." The other thinks of language as a haze

drifting among sharp-edged sense data, or neural sparkings, or

Democritean atoms; if it does not actually hamper our vision

of these divinely actual and determinate realities, it at least needs

to be crystallized into equally sharp-edged units before it can

represent them properly. Both look toward a day when thought

19 One may find these phrases reminiscent of some key idealist arguments.

But Peirce is no idealist; for him, its error lies in ignoring Secondness and

Firstness (cf. the description of "Hegelianism of all shades" in 5.77n.). In

other words, idealists reasoned illegitimately from "there is a sign (Third)

behind every sign" to "there are nothing bqt signs (Thirds)."

20 Loc. cit., p. 64. Hereafter, references to Pears's article will be inserted in

the text.

204

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

will, in Pears's terms, "exit from the maze of words" (p. 53)

and will know just where and why it exited, and just what it

found. In this self-conscious exiting, man will differ from the

beasts of the field (and from the user of C. I. Lewis's "expressive

language") in that he will somehow be able not only to exit but

to return and drag others up (if he is a Platonist) or down (if he

is a nominalist) to the same exit. Men will thus "transcend lan-

guage" either by finding the color-patches behind the use of

"cat" or by finding "Die schlnen regionen / Wo die reinen Formen

wohnen" beyond both patches and cats. In either case, as

Pears says, we could "combine the concreteness of ostensive

definition with the clarity of verbal definition" (p. 63). As Hegel

saw, when thought strives for self-consciousness it strives for a

concrete universal.

By taking "naming" as the datum which theories about univer-

sals are intended to explain, Pears is able to put his critique of

all such theories-nominalist, conceptualist, and realist-in the

following two theses: (i) "any comprehensive explanation of

naming is necessarily circular" and (2) "all other processes [to

which naming might be analogized by philosophers] either already

contain the very feature of naming which was puzzling, or else

are too natural or too artificial really to be analogous" (p. 53).

"The possible analogies," Pears says, "can be mapped in this

simple way": either the many things related to some single thing

(in the way in which instances of a name's application are related

to the name) are related to this single thing naturally (independ-

ently of what we do about it) or they are related just because we

choose that they shall be related (purely artificially) or else we

choose to relate them because of some feature which makes it

"convenient but not necessary" so to relate them.21 Now only the

latter, obviously, will do if we are to have a satisfactory analogy.

But, Pears says, "This compromise between the two extremes

introduces into the analogy the very feature which it was intended

to explain. For just how something works in influencing usage was

what was to be explained. Nor is there a fourth alternative"

(pp. 6o-6 i).

21 This is a paraphrase of Pears's "map" as given on p. 6o.

205

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTT

Taken positively, what Pears's article shows us is roughly this:

(i) we now know as well as we ever will why it is "convenient

but not necessary" to give certain batches of things a common

name; (2) the hunt for an insight into batchiness will either fail

or else simply send us back to this original knowledge. "Naming

cannot be explained by anything which goes byond a reasoned

choice of usage" (p. 62). So the puzzling thing is not so much

naming as the existence of theories about naming, the dialectic

of which Pears exhibits as follows:

Thus moderate nominalists maintain that similarity is a better expla-

nation of the unity of a class than the presence of a universal. (But

why should people not just recognize the presence of universals?)

And moderate realists retort that this admits the existence of at least

one universal, similarity. (But why should the presence of a universal

explain the recognition of similarity if it cannot explain the recognition

of anything else? Why should we not just recognize similarity?)

Really these are not two arguments but two bare assertions of superior-

ity. . . . Yet these theories do seem to be striving toward something.

And they are. Their goal is the unattainable completely satisfactory

explanation of naming. And, as so often happens in metaphysics,

progress is measured by distance from the starting-point and not by

proximity to the goal whose unattainability each uses against its rivals

without allowing it to deter itself [p. 62].

Peirce's solution to the problem of naming, realigned and

restated so as to form a commentary on Pears's dissolution,

involves the following points:

(i) What is wrong with most theories about naming (nominal-

ism and conceptualism, and also realism as it is usually stated)

is not that they strive after an unattainable goal. (This, roughly

speaking, is not a specific vice of metaphysics, but a generic one

of all inquiry.) Rather, what is wrong is that they block the way

of inquiry by appealing to one form or another of "just seeing."

(2) Properly understood, realism is not the postulation of a

new breed of entity combined with the claim that we somehow

intuit these entities but is, on the contrary, a way of asserting that

no such postulation will do.

(3) Such an assertion is contained in the doctrine that Third-

ness is real, for this doctrine has a corollary that the "convenience

206

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

without absolute necessity" involved in naming is not collapsible

into absolute necessity nor into sheer arbitrariness.

(4) Saying that "vagueness" (or any of the other sobriquets

of Thirdness) is real is preferable to saying that "naming is

sui generis"-and thus closing off the possibility of "explaining"

naming in any sense-because it points the way to an indefinitely

long series of things which are like naming. To say that these

analogies do not explain is true only if one means by "explana-

tion" something like a Cartesian "clear and distinct apprehen-

sion."

Thus when Peirce says that naming is possible because Thirds

are real he is saying something like this: "It is true that nothing

which is not a Third is sufficiently like naming to be used as an

explanation of naming, and also true that all other Thirds are,

as such, like naming merely in being triadic. But by exhibiting

more and more of the ways in which Thirdness appears we gain

understanding of any given Third." Peirce can agree when

Pears says that no theory of naming "goes deep enough to satisfy

the true metaphysician who is in all of us, since though they take

us to the bottom of naming, we were in a simpler way already

there, and they do not succeed in showing us how naming is

founded on something else which lies even deeper" (p. 62).

But Peirce points out that explanations can succeed by being

broad as well as by being deep. Any analogy, no matter how

illuminating at first glance, can be made to look ridiculous by

asking "and in just what respect are these purported analogues

analogous ?" 22 But the illumination one gets from an analogy

may be nonetheless real, even though it is gained not by going

deeper but by looking about and noticing that the object of one's

previous puzzlement is no more and no less puzzling than a lot

of other things. After all, we have to take our illumination where

we find it, and we cannot know that we are no closer to our

goal of complete illumination unless we somehow know the

criteria for such completeness in advance. The naming-theorist's

myth of a complete noncircular explanation of naming is, in

22 Think how easily one could "expose" the Philosophical Investigations if one

insisted on pressing this question, and how pointless it would be to do so.

207

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD R OR TY

Peirce's eyes, simply turned upside down to produce Pears's

myth of a simple understanding of naming possessed prior to

all analogies and explanations. For Peirce, both are myths

because both "admit something to be absolutely inexplicable"

(5.3I8), thereby succumbing to the Cartesian yearning for the

intuitive. To say that we "just know what naming is" is no more

of an advance, and is no less immodest a claim, than to say that

we "just recognize" similarities or Similarity or Catness.

But what illumination does Peirce's Thirdness offer us? And

why must he call his doctrine of the reality of Thirdness "Sco-

tistic Realism"? The answer to these questions must begin by

distinguishing the intent behind realism from what Peirce

thinks of as a parody of realism- the notion that universals are

"things." The parody consists in holding that realists believe

in two independent sets of sharp-edged, fully determinate entities

particular universals and particular particulars. This redupli-

cation of the world, like all parodies of philosophic theories, is

easily refuted; in this case, the job is done by one variant or

another of the "third man" argument. But for Peirce, this notion

of what realism is could only have occurred to a mind so imbued

with nominalism as to give a nominalistic twist to anything it

encounters.23 The intent of realism is the opposite: rather than

adding new determinate entities to the world, it was intended

precisely to get rid of "the Ockhamistic prejudice ... that in

thought, in being, and in development the indefinite is due to a

degeneration from a primary state of perfect definiteness. ...

The truth is rather on the side of the scholastic realists that the

unsettled is the primal state, and that definiteness and determi-

nateness . .. are, in the large, approximations, developmentally,

epistemologically, and metaphysically" (6.348).

Now whether or not Peirce was right in construing medieval

23 This is why "the doctrine of Platonic ideas has been held by the extremest

nominalists" (5.470), and is why Leibniz, with his utterly determinate possibilia,

is "the modern nominalist par excellence" (5.62). For an amusing exhibition of

the central difficulty of Leibniz's theory of possible worlds-a difficulty Peirce

would say was involved, mutatis mutandis, i4 all nonpragmatic theories-see

N.P. Stallknecht, "Decision and Existence," Review of Metaphysics VI, (I952),

3' if..

208

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

realists to mean what he meant,24 his own doctrine of the reality

of Thirdness is no more and no less than this insistence on the

irreducibility of the indefinite and the indeterminate.25 For

Peirce, it is the nominalist and the reductionist who succumb to

belief in metaphysical figments-namely the belief that beneath all

the evident fuzziness, vagueness, and generality which we encoun-

ter in language (and, therefore, in all thought26) there are

nonfuzzy, particular, clearly intuitable reals. (Compare 6.492,

5.3i2.) To repeat an earlier point, Peirce's realism is simply the

phrasing in metaphysical language of the unrestricted form of

the doctrine that language cannot be transcended. And the

illumination given by this way of phrasing the point is simply

that such phrasing lights up, and points to similarities between,

the Protean attempts at such transcendence which have occurred

in the history of thought. In particular, Peirce uses it to expose

the parallelism between the movements of reductionist explana-

tion charted by means of Pears's map of analogies about naming

and the movements of reductionist explanation of such topics as

induction, mind, intention, value, and freedom. In the various

phenomena grouped under these headings we encounter indefi-

niteness becoming definite or indetermination calling on us to

determine it; just in so far as we are reductionists or nominalists,

we assume that the progress toward definiteness and determinacy

is not only completable but that the fully definite and determinate

is somehow there already; just in so far as we are "realists" in

Peirce's sense, we do not assume this.

To show how this general map of philosophical issues works

in all these areas, or even to give the details of Peirce's application

of it to naming and the use of language, is beyond the scope

24 Cf. I.27n.: "It must not be imagined that any notable realist of the I3th

or I4th century took the ground that any 'universal' was what we in English

should call a 'thing' ... ." For a discussion of the historical question, cf.

Charles K. McKeon, "Peirce's Scotistic Realism" in Studies, pp. 238 if.

25 The difference between these two notions is the difference between

vagueness (indefiniteness) and generality (indeterminacy) -a distinction which

is of great importance in Peirce's system but which space prevents being

employed here. Cf. 5.447 if. for the distinction between the two, and 5.450

and 5.506 for their formal identity (as Thirds). In what follows, I shall continue

to use the terms interchangeably.

26 Cf. 5.250 if., and vol. 5, bk. II, ch. i of the Collected Papers, passim.

209

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTY

of this paper. But we can sketch its use in regard to naming by

focusing on two of the most important sobriquets of Thirdness:

"Sign" and "Habit." The reality of Thirdness is sometimes put

by Peirce as the claim that "The entire universe . .. is perfused

with signs" (5.448n.). Combined with the doctrine that no sign

is fully determinate (3.93, 5.5o6), 2 and that "everything in-

determinate is in the nature of a sign" (5.448n.), this gives us

the corollary notions of a man's mind being itself a sign (5.3I3)

and "nature" (the sum of objects of knowledge) being an utterer

of signs which we interpret. But no sign-neither a thought nor

a natural event-is completely indeterminate. If it were, it would

be unrecognizable and thus could not be this sign rather than

that one. A sign is thus always a determinate indetermination.

The act of signifying, of meaning something, is thus analogous

to the act of naming (that is, the act of assigning a single sign to

represent a batch of things). For as Pears reminds us, naming is

neither simply artificial (indeterminate) nor simply natural

(determinate), neither forced on us nor performed by us in a

spirit of pure whimsy. To signify or to name such-and-such by

so-and-so is "convenient but not absolutely necessary." Now this

is just what one would expect if the "things" that are batched

under a given name are what Peirce says they are-neither

physical, logical, nor psychological atoms, but rather signs which,

while giving us latitude for interpretation, resist some inter-

pretations more than others. Thus the picture of a universe

perfused by signs is a picture of a universe in which indeterminacy

is neither an illusion nor a peculiar property of a human artifact

called "language." A Peircian realist thus "explains" naming not

by claiming to "recognize the presence of a universal" as one

would recognize the presence of a bird in a bush-as one deter-

minate entity among others-but by taking the peculiar thing

about naming (the peculiar determinate compresence of nature

27 See 3.93 for Peirce's refutation of logical atomism, taken as the claim

that there are fully determinate signs: "The logical atom . . . must be one of

which every predicate may be universally affirmed or denied. . . . But an

absolutely determinate term cannot be realized, because, not being given by

sense, such a concept would have to be formed by synthesis, and there would

be no end to the synthesis because there is no limit to the number of possible

predicates."

2IO

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

and art, force and passivity, determinacy and indeterminacy)

to be a feature of the named as well as the namer. From the

point of view of the nominalist, of course, this merely amounts to

reading back into nature and "hypostatizing" a peculiarity of

our mind; Peirce's reply is that this charge of hypostatization

can be made only by one who has himself hypostatized the results

of analysis in some form of atomism, logical or otherwise.

This explanation of naming can be put in terms of the notion

of "Habit" as well as that of "Sign." Pears's description of a

"reasoned choice of usage" in terms of "convenience without

absolute necessity" is clearly reminiscent of Hume's remark on

''custom or habit": "the uniting principle among ideas is not to

be considered as an inseparable connexion" but as a "gentle

force, which commonly prevails, and is the cause why, among

other things, languages so nearly correspond to each other." 28 Even

apart from the triadic character which Peirce attributes to them

both, it is not hard to see how signs and habits can be thought of

as two ways of looking at the same phenomenon. Both are vague;

both are neither natural nor altogether conventional; a sign can

have many interpretations (while resisting some more than others)

and a habit can express itself at various times in various manners

(but not in any and every manner). The cash value of my giving

a name to a batch of things is my establishing a habit of cor-

relating tokens of a given sign with tokens of other signs. For

nominalists, reductionists, and atomists, this habit is either a

confused way of thinking about a set of reflexes, or else a sui

generis phenomenon occurring in human beings (which is more

or less what it was for Hume). But Peirce, looking at the universe

as perfused with habits as well as with signs, explains the conven-

ience of naming certain batches-of slicing up nature in certain

ways, and thereby developing certain habits of expectation-

by reference to the fact that nature has already sliced itself up

by developing habits on its own.29 Thus his realism can be seen

28 Treatise, ed. by Selby-Bigge (Oxford, I950), p. IO.

29 The fact that assigning a name creates a habit of expectation (which,

for Peirce, was the definition of a "belief") is basic to Peirce's thought. The

essential thesis in question has been exhibited with great elegance in Wilfrid

Sellars' "Concepts as Involving Laws and Inconceivable Without Them"

211

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTT

as the thesis that the "reasoned choice of usage" which is naming

is rational, in part, because of its respect for the rationality which

it encounters in nature. But for nature to be rational in this

sense does not mean that it "recognizes the same universals"

as does the mind, but simply that it contains the sort of determi-

nate indeterminations that our mind does.

Thus the real difference between Pears's view of naming

as sui generis and Peirce's "explanation" of it as one more appear-

ance of that pervasive feature of nature and thought which he

calls Thirdness is indeed, if one likes to call it so, verbal. To say

that nature has habits or utters signs is not to take us any deeper

into our own expecting and talking than we were already. If we

ask for an explanation of the "possibility" of any given instance

of naming we shall still have to look at the details of the particular

situation (although if we are Peircians we shall look at these

details through different spectacles). Peirce's application of his

categorial scheme to the act of naming can be seen as simply an

elaboration of Pears's (crucial) remark that "No description of

any item of mental furniture which included only its momentary

properties and not its habitual use could possibly explain the

generality of thought" (p. 56n.). But equally, both this remark

and the whole of Pears's discussion can be seen as an illustration

of Peirce's key heuristic principle-that Thirdness is not reducible

(Philosophy of Science, XV (1948), 287-3I5). The point expressed by the title

of this article is clearly seen by Wittgenstein when he says, "If language is to

be a means of communication, there must be agreement not only in definitions

but also (queer as this may sound) in judgments" (No. 242).

The thesis that nature has habits (which in modern jargon is the thesis that

dispositional predicates are irreducible) is the starting point of Peirce's solution

to the problem of induction (cf. 5.I70, 5.457). This is unfortunately too large

a topic to enter on here; suffice it to say that he is here putting in metaphysical

language the conclusion at which present-day inquiry is now gradually arriving:

that the control which nature exercises over our inductive inferences appears

not only in the results of experiment and observation but in the construction

of frameworks within which we observe and experiment, and that this latter

sort of control is not reducible to the former. Cf. Nelson Goodman, Fact,

Fiction, and Forecast (Cambridge, Mass., I955), p. I17 and ch. iv, passim. When

this conclusion is more broadly stated, it appears as Professor Polanyi's view

that all inquiry (not merely bad science or metaphysics) is "circular" and has

an "epicyclical character"; cf. his Personal Knowledge (Chicago, I958), pp.

288 if. and chs. ix and x, passim.

2I2

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

to Secondness and Firstness. In applying Peirce's categories to

the act of naming, no new exit has been found from the maze of

words; but perhaps something has happened to the man who

feels caught in the maze analogous to what happens to a prisoner

who takes to heart the realization that all men are, in one sense

or another, prisoners. All that has happened is that the maze of

words has been enlarged by more words, but what more can we

reasonably hope for, once we are convinced that language cannot

be transcended?

III

In this final section I want to press one step further the analogy

between some manifestations of Peirce's doctrine of the irreduci-

bility of Thirdness and some of the insights that underlie the anti-

reductionist revolt in contemporary empiricism. A whole range

of distinctions which Wittgenstein makes in the Philosophical

Investigations can be seen as clustering around the distinction

between "causal" and "logical" determination (No. 220).

Causal determination is what using language would be like if it

were a matter of being "inspired" by an "intuition" (Nos. 2I3,

232). Logical determination, on the other hand, is what goes on

when we follow a rule.30

The program of enforcing this distinction covers a great deal

of what Wittgenstein accomplishes in the Philosophical Investigations

(especially in the first half of Part I). In his analysis of meaning and

understanding he spends much of his time exhibiting how useless

it is to look for "some item of mental furniture" which will

"guide" or "compel" or "influence" one to, for example, utter

a certain sound when one sees a certain letter (Nos. i69, I70).

Reductionism as applied to the activities of meaning and under-

standing appears as the attempt to find an "intermediary between

the sign and the fact" (No. 94) which will supply a "lever," the

felt pressure of which causes us (No. I 70) to mean or understand

just so-and-so in just such-and-such a way. Wittgenstein's critique

of the notion of "mental states," centering on the fact that we

30 Cf. Nos. i69-I76, i98-220, and especially No. i98, where Wittgenstein

explicitly contrasts causal connection with custom.

2I3

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD ROR TY

cannot find any particular experience which is "an experience of

being influenced" (No. I76), provides the answer to those who

think it a flaw in Hume not to have produced somehow an

impression of the "gentle force" of habit. The answer consists

in pointing up the categorial gap between any given here-and-now

"state" and the temporal stretch which is a use (No. I38) or the

application of a rule (No. I40), or the having of a habit (all the

time, whether it is being exhibited or not). But in destroying

the myth that we sense a certain "psychological" compulsion

which we can distinguish from the "logical" compulsion which

it purportedly underlies (No. I40), Wittgenstein shows that we

can and must distinguish logical from causal determination

(No. 220). This distinction, I now want to show, is tied up with

Peirce's central distinction between the indeterminate and the

determinate.

In its most general form, Wittgenstein's "master argument"

against all forms of reductionism is that they generate infinite

regresses, and this is also Peirce's master argument against

Cartesian intuitionism. Wittgenstein's use of this form of argument

is clearest in his insistence on the vagueness of rules (and, a fortiori,

of concepts). If a rule were perfectly definite and nonvague-if

it left nothing to the discretion of its applier-then the applier

would need a rule to determine the application of this rule, and

so ad infinitum: "What does a game look like that is everywhere

bounded by rules? ... whose rules never let a doubt creep in,

but stop up all the cracks where it might? Cannot we imagine

a rule determining the operation of a rule, and a doubt which

it removes-and so on ?"31 If we admit that the application of a

rule requires something to mediate between the rule and its

instances, we are at once caught in the "third man" regress-

for the mediator can only be another rule.32 But if the rule is

31 The remainder of the paragraph (No. 84) from which this is quoted

goes on to point out that leaving room for doubt does not paralyze action;

cf. No. 87. This should be compared with Peirce's discussion of "make-believe

doubt" (e.g., at 5.265).

32 See Wittgenstein's discussion at No. 86 of the notion that we need a schema

to tell us how to use a table, another schema to tell us how to use the first, etc.

This should be compared with Kant's notion of "schemata" (K. d. r. V.,

A 137-B I76 if.) and Peirce's critique thereof (5.53 I)

214

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

something as sharp-edged and definite as a mental state-if it

comes like a flash (Nos. I39, I9I, I97) and "intimates" (Nos. 222,

232) to me what I am to do-then it will require a mediator:

"But this initial segment of a series obviously admitted of various

interpretations (e.g., by means of algebraic expressions) and so you

must first have chosen one such interpretation." Not at all. . . . A

doubt was possible in certain circumstances. But that is not to say that

I did doubt, or even could doubt....

So it must have been intuition that removed this doubt? If intuition

is an inner voice-how do I know how I am to obey it? And how

do I know that it doesn't mislead me? For if it can guide me right,

it can also guide me wrong.

((Intuition an unnecessary shuffle.))

If you have to have an intuition in order to develop the series 1234 ...

you must also have one in order to develop the series 2222...

[Nos. 2I3-214].

Whenever we try to resolve the indeterminacy which is in doubt

by an appeal to intuition, we let ourselves in for being forced to

postulate a superintuition which will let us know whether this

intuition is the appropriate one for the occasion (compare Nos.

190-192). As Peirce puts it:

Now it is plainly one thing to have an intuition and another to know

intuitively that it is an intuition, and the question is whether these

two things, distinguishable in thought, are invariably connected, so

that we can always intuitively distinguish between an intuition and

a cognition determined by another.... There is no evidence that we

have this faculty, except that we seem to feel we have it. But the

weight of the testimony depends entirely on our being supposed to

have the power of distinguishing in this feeling whether the feeling

be the result of education, old associations, etc., or whether it is an

intuitive cognition; or, in other words, it depends on presupposing

the very matter testified to [5.2I4].

What Peirce calls "knowing that it is an intuition" is pragmatical-

ly equivalent to "knowing that it does not mislead me." In still

more general phrasing, both men are saying that the kind of

33 On not being able to find in ourselves a faculty which we feel is there,

cf. Wittgenstein, No. I76.

215

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTT

indetermination involved in understanding or following a rule is

not removed by postulating any determinate entity to eliminate it,

for the original indefiniteness reappears at the level of this new

entity. 3 Whether one postulates mental states, or formation

rules of an ideal language, or Leibnizian essences, the vagueness

which one is trying to expel will turn up again in the instrument

of expulsion (for example, in criteria for detecting a given mental

state, or in the metalanguage used to state formation rules).

Now "causal determination" (as Wittgenstein uses the term)

is among relations what "bare particulars" or Leibnizian monads

are among the terms of relations-that is, it is the archetype of

determinacy. The paradigm case of causal determination, in this

distinctive sense, is Hume's rebound of billiard balls (Peirce's

Secondness qua Efficient Causality), to which (reverting to

Pears's terms) absolute necessity is central and "convenience"

is quite irrelevant: /3cL rather than 6'ayK7q. The temptation

which mechanism held out to the empiricists of the nineteenth

century was to "reduce" the acvadyKcr of Hume's "gentle force"

to the Wta of corpuscles or electric charges. Peirce, in the meta-

physical application of his categories, was reacting against this

attempt, and Wittgenstein was reacting against its successor-

the attempt to perform the same reduction "epistemologically"

(or logisch-philosophisch) rather than psychologically. Their language

is therefore different, but the patterns of arguments they use,

and the kinds of similarities and dissimilarities to which they

point, are as similar as are their enemies. Both are fighting

against the "Ockhamistic" prejudice that the determinate always

lurks-actually, and not merely potentially-behind the indeter-

minate. Both recognize the sense in which we cannot break out

of the cluster of things which Peirce calls Thirds and whose

workings Wittgenstein calls "logical determination" (for example,

signs, words, habits, rules, meanings, games, understanding) to

something more definite which will somehow replace these things.

The general thesis that reductionism, nominalism, and intuition-

ism are at one in trying to transcend language is, so to speak, the

methodological way of putting what Peirce put metaphysically

34 The argument sketched on p. 203 above can easily be seen to be a special

case of this general line of argument.

2 i6

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

when he said that they were at one in trying to avoid admitting

the reality of Thirdness.

We should now be able to see the problem of naming as a

special case of a more general problem. One of the forms in which

the distinction between causal and logical determination appears

in the Philosophical Investigations is the replacement of the dyadic

"mirroring" relationship of sign and fact in the Tractatus by

the triad "X means Y to Z," where Z is something (for example,

a human being) which is not sharp-edged, is not fully determinate

(like an atomic fact or a picture of one), but is quite fuzzy around

the edges. Broadly speaking, the reason for this fuzziness is

that even this triad is incomplete, because it contains implicit

reference to a context which is indefinite in extent. If we regard

this context as grouping itself around the language user rather

than around his signs or their referents, we can see it as the

fact that behind the use of every rule in a language game there

are the rest of the rules of the game.35 Beyond these rules, in

turn, one glimpses the rules of still another game, in terms of

which the first one has meaning. And so on. Now a reductionist,

invoking the "cannot get started" argument36 to avoid this

creeping contextualism, will usually retreat to the topic of

learning rules rather than using them-and, in particular, to the

putatively primitive process of learning names, a process whose

context, in so far as it has one, is presumably exclusively causal

rather than logical in character. But here we are met by the

opening argument of the Investigations, against the usual notion

of how this learning works. Now in the light of our description

of the Wittgensteinian "master argument," we can see that

Pears's exhibition of the sui generis character of naming is equiv-

alent to this opening argument. For the central thesis of both

is that, granting that in ostensive definition we do exit from

the maze of words, this exit does not give us a touchstone to

35 Cf. No. 3I on separating the character of being a king from the rest of

the game of chess and, generally, the critique of ostensive definition which

stretches roughly from No. 26 to No. 38.

36 The term is due to Alan Pasch (op. cit., pp. i62-i63). Pasch's counter-

argument, to which Wittgenstein would certainly have been sympathetic, is,

"After all, what cannot get started without a' lowest-level language is not

language but analysis."

2I7

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTT

which we can recur and on which we can rest our understanding

of learning and using language. As Wittgenstein points out: one

of the differences between obeying an inspiration and following

a rule is that in the former case we cannot teach somebody else

our technique (No. 232). The appeal to ostensive definition in

any of its forms ("just recognizing" color samples, or similarities,

or universals) is simply a special case of intuitionism, and is to be

met with the same set of arguments as any other such case-

namely, that such an appeal is either a confession of failure or

else the generator of an infinite regress. Specifically, the sort of

infinite regress generated is indicated by Pears when he says that

the answer offered by realism to the question, "Why are we able

to name things as we do?"

* . . could not be informative even if it were detailed; since there

could be a non-circular answer to the question, "What universal?"

only if the exit from the maze of words were made at some different

point, which would merely put off the moment of embarrassment

from which in the end neither speech nor thought can be saved

[P. 54].

The infinite regress is generated by successive puttings off of

the moment of embarrassment, performed by successive requests to

"just recognize" something.37

Given this view of the negative dialectic which Pears's article

exhibits as a special case of the general anti-reductionist and

anti-intuitionist dialectic exhibited by Peirce and Wittgenstein,

we can now compare the "positive" sides of their doctrines.

Without attempting to turn the Philosophical Investigations into a

work on ontology, it is perhaps not unduly distorting to say

37 It is perhaps worth noting that an accusation that one's opponent is

caught in a vicious circle is almost always answered by his attempt to break

the circle by moving up to a higher logical type-e.g., by appealing to an

intuition of the intuitional character of the experience which was originally

presented as itself a complete explanation. The counterploy is then to chase

one's opponent up a ladder of types (converting the vicious circle into a vicious

helix) until you get tired, at which point you accuse him of generating an

infinite regress. Thus the "vicious circle" and the "infinite regress" ploys form

the two horns of a destructive dilemma and each is, in a sense, incomplete

without the other. A fuller statement of Ieirce's and Wittgenstein's anti-

reductionist master argument would bring this point out.

2 i8

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

that Wittgenstein, like Peirce, is insisting on the reality of vague-

ness. When he argues against Frege and "Ockhamists" generally

that a concept's having vague boundaries does not prevent it

from being a concept, precisely because it does not prevent it

from being used (No. 7'), he is articulating the germ of Peirce's

thesis that realism (in the sense of the irreducibility of the in-

determinate) and pragmatism reciprocally entail each other.

The character and the importance of the pragmatism of both men

becomes evident if we remark that both, despite their use of the

anti-reductionist master argument sketched above, insist on the

unavoidability of certain "harmless" infinite regresses. For

Peirce there is potentially a sign behind every sign, and for

Wittgenstein there is potentially a language game behind every

language game; but both consider these regresses harmless on the

pragmatic ground that practice does not require the actualization

of these potentialities.

The difference between harmful and harmless infinite regresses

will be clearer if we revert to a metaphor of Section II and talk

about the difference between the depth and the breadth of an

explanation. The intuitionist and the reductionist look for some-

thing determinate underlying the indeterminate-something of

which the indeterminate is an epiphenomenon. In the anti-

reductionist master argument, this determinate thing is shown to

be itself indeterminate and to require the postulation of a still

deeper-lying determinate, and so on. Whether one thinks of this

regress as going down in search of the Deepest or up the ladder

of logical types in search of the Highest is less important than

what I shall call the vertical character of both searches. This

character consists in the fact that one looks at each new step as

a transition to a new level-a level which is a necessary condition

for the existence of the previous step(s). That is, at level n one

looks at the existence of level n-i as possible only by virtue of n,

which itself exists a se. Such a regress can be contrasted with what

I shall call a horizontal one, in which each new step gives us

something which is of essentially the same kind as what we had

at the last, but something which renders the last step more

determinate than it was. The relation between step n and step n-i

is thus not like the relation between creator and created, but like

21Q

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTr

that between a mystifying book and a brilliant commentary on

it; the book was there already, even though perhaps nobody

could make much of it until the commentary came along. Nor

would there be anything surprising in somebody writing a com-

mentary on the commentary, and so ad infinitum. Movement

along a horizontal regress lacks the sort of jolts we feel whenever

we are forced to a new level in a vertical regress (the sort of jolt

felt by the child when the question "Who made God?" first

occurs to him) and it also lacks the sense of utter futility which

grips us when we realize that we can always be forced to move

on from any level of a vertical regress. The reason a vertical

regress can be condemned to futility by the anti-reductionist

master argument is that it attempts the impossible task of making

determinate the relationship between the purely determinate

and the purely indeterminate (resembling in this the task of

explaining creatio ex nihilo, or the Concrete Universal). The

reason why a horizontal regress cannot be destroyed by the same

argument is that it takes the datum as it finds it, as a determinate

indetermination, and realizes that all further steps will also

produce determinate indeterminations which, while they can

render the original datum more determinate, cannot, because of

their own indetermination, render it perfectly determinate.38

The relation of this notion of a horizontal regress to prag-

matism will perhaps be clearer if we glance at the example of

religious explanation used in the last paragraph. Does it harm

our relation to God if we realize the futility of understanding

his relation to the world? To some men, this is indeed harmful;

they may, for instance, cease to be able to pray. Such men are

analogous to those who commit the classic reductionist error of

expecting the results of analysis to exist prior to the analysis,

and who take the ill success of an analysis to entail the irreality

of the datum. But, as "existentialist" as well as pragmatist

38 The horizontal vs. vertical metaphor can be pressed a bit further, as

follows: a horizontal regress exhibits a process which actually takes place in

time. It is thus stretched out along time's arrow, whereas the steps of a vertical

regress are connected by atemporal relations of "presupposition." Vertical

regresses are, in fact, horizontal ones turned on end by people with unprag-

matical minds and a preference for eternity over time.

220

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

theologians tell us, such a realization need not be harmful (and

may well be therapeutic). When Tillich, for instance, tells us

that the task of theology is no more and no less than to "corre-

late" revelation with the culture of the times, and that revelation

prior to such correlation is, in so far forth, indeterminate, he is

saying that theological explanation takes a horizontal rather than

a vertical form. But since theology is, after all, not the religious

life but simply one (more or less optional) expression of it, this

eternally indeterminate character of theological explanation is

harmless. Just so, the fact that our understanding of how we

follow a rule or give a name will be permanently vague does not

interfere with our actually obeying rules and naming things. This

resolution of the indeterminacy of interpretation is put by

Wittgenstein as follows:

This was our paradox: no course of action could be determined by

a rule, because every course of action can be made out to accord

with the rule. The answer was: if everything can be made out to

accord with a rule, then it can also be made out to conflict with it.

And so there would be neither accord nor conflict here.

It can be seen that there is a misunderstanding here from the mere

fact that in the course of our argument we give one interpretation

after another; as if each one contented us at least for a moment, until

we thought of yet another standing behind it. What this shows is that

there is a way of grasping a rule which is not an interpretation, but

which is exhibited in what we call "obeying the rule" and "going

against it" in actual cases.

Hence there is an inclination to say: every action according to the

rule is an interpretation. But we ought to restrict the term "inter-

pretation" to the substitution of one expression of the rule for another.

And hence also "obeying a rule" is a practice. And to think one is

obeying a rule is not to obey a rule [Nos. 20I-202].

The permanent possibility of practice is what renders harmless the

indefinite horizontal regress of interpretations, oscillating as

they do between the purely determinate ("nothing accords with

the rule") and the purely indeterminate ("everything accords

with it").

221

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 RICHARD RORTT

Finally, if we turn from the theory of meaning to the theory

of knowledge, we can see how this pragmatism ties in with the

more familiar notion of pragmatism as mediating between realism

and idealism. Both realism and idealism are forms of intuitionism:

the former attributes a lot of little, completely determinate

intuitions to us; the latter attributes one great big, completely

determinate intuition to the Absolute. The permanent truth of

idealism is that every state of knowledge is indeterminate and

incomplete, whereas the permanent truth of realism is that each

state of knowing is what it is and no other thing. The recognition

that a series of knowings is a series of determinate indetermina-

tions does justice to both sides of this putative dilemma, but

this recognition is possible only if we grant that indeterminacy

is inescapable.39 Now the cash value of the distinction between an

appeal to practice and an appeal to intuition is just that an intuition

cannot be vague and still be an intuition, whereas an action is

neither vague nor nonvague. Your thought of obeying a rule can

be made as sharp-edged as you please, but the question of the

determinacy or indeterminacy of the act of obedience simply does

not arise.40 The regress of incomplete interpretations-a regress

39 The reader who is bothered by my free-and-easy use of "indeterminacy"

will find support for his suspicions in D. C. Williams' critique of "a confusion

which affects much of contemporary philosophy, the confusion between the

idea of the determinate and the idea of the determined. The new word 'determi-

nacy' or 'indeterminacy' seems expressly invented to elide and conceal the

distinction." ("The Sea-Fight Tomorrow" in Structure, Method and Meaning,

ed. by Henle, Kallen, and Langer, New York, 1951, pp. 292-293.) The point

cannot be argued in this space, but I should want to say that what pragmatism

invites us to see is precisely what the tradition which Professor Williams calls

"the extensionalistic view of things" never saw: that the determinate is such

only because it is determined by something. This general critique of extension-

alism and related strands of thought is put forward very clearly in W. D.

Oliver's Theory of Order (Yellow Springs, 195i), especially ch. iii.

40 As Wittgenstein suggests (No. 176) the inability to find an item of mental

furniture which is the experience of being influenced by a rule is the "germ

of the idea that the will is not a phenomenon." In Peircian language, this

Kantian point is expressed by saying that Secondness is not Thirdness. In the

language of language strata, we can say that terms like "vague" and "in-

definite" occur in the stratum in which we talk about interpretations (in

Wittgenstein's restricted sense of the "substitution of one expression of a rule

for another"), but not in the stratum in which we talk about actions qua

actions (if there is such a stratum).

222

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions
 PRAGMATISM AND LANGUAGE

of rules, habits, and signs standing behind rules, habits, and

signs-which we find in Peirce's and Wittgenstein's horizontal

regresses gives idealism its due. But reference to action, which

(because degree of determinacy is irrelevant to it) can take place

at any step in the eternally incomplete series of interpretations,

preserves the down-to-earth character of realism.

In this paper I have been emphasizing similarities between

Peirce and Wittgenstein, and I have played down the differences

between them. These differences are real and important-and I

have hinted at them in Section II, in discussing Pears's appeal

to a direct acquaintance with the naming process. There is

obviously no room here for a systematic exposition of these dif-

ferences, but I should like to suggest that such an exposition will

be most fruitful if it is given against the background of the similar-

ities which I have attempted to sketch. The lesson which Wittgen-

stein's later writings taught us about the nature of philosophy

loses much of its value when it is construed as an attack on system-

atic theory construction and technical vocabularies. Such a

reading of Wittgenstein confuses the real enemy-reductionism-

with the use of a tactic which is available to the reductionist

and the antireductionist alike. It uses the accidents of Wittgen-

stein's thought to criticize the accidents of previous philosophizing.

The followers of a great philosopher often confuse the tactics he

used in criticizing his predecessors with the master strategy of

his approach to philosophic problems. This happened to Peirce in

his own day, and there is evidence that it is happening to Wittgen-

stein in ours. Drawing historical parallels, in the fashion illustrated

by this paper, can be mischievous if done in a "reductionist"

spirit-treating one mode of philosophizing as merely a confused

approach to what some other mode has made plain. But if this

spirit is avoided (and I have done my best to avoid it here),

historical comparisons can help free us from preoccupation with

accidents of tactics and can direct us toward the crucial insights

which generate master strategies.

RICHARD RORTY

Wellesley College

223

This content downloaded from 159.237.12.65 on Wed, 23 Mar 2016 14:17:11 UTC
All use subject to JSTOR Terms and Conditions