categories and transcendental arguments

M.S. GRAM

The University of Iowa

C A T E G O R I E S A N D T R A N S C E N D E N T A L A R G U M E N T S

Philosophers have long claimed the. existence of a kind of argument which can establish the existence of primitive concepts in our conceptual scheme. What makes a concept primitive has not been stated precisely by philosophers purporting to offer transcendental arguments. Kant says that a primitive concept is a principle o.f the understanding; and, more recently, Strawson has identified such concepts with what he calls the massive central core. of human thinking. 1 There are, indeed, properties which primitive concepts have.; but both characterizations are defective, and for the same reasons. To say either that a concept is a principle of understanding or that it belongs to the massive, core o.f human thinking is merely to. say that a

concept is a means by which we think o.f something. And this is true of primitive and ordinary concepts alike. To, say, on the other hand, that certain concepts are principles o.r that they are parts of a core is merdy to repeat the distinction to. be clarified. I offer the following as a characteristic which will distinguish primitive from other concepts : A concept is primitive if it is instantiated when any other concept is instantiated. An argument is transcendental if it estabiishes which concepts have this property. Thus if there are transcendental arguments, we are in possession of a means of solving a perennial metaphysical problem. We will be able. to distinguish between which of our concepts are empirical and which are concepts of generic o.r categorial features of the world. And this will be no mean accomplishment. To have a way of distinguishing between categorial and empirical concepts is to. show that there is a proper domain o,f metaphysics not reducible to any other domain.

There are two, things that should be noted at the outset. One concerns the adequacy of my provisional criterion for primitiveness of concepts; the other, the. relation between the theory of transcendental arguments and the theory of concepts. I take. them in turn. My criterion is vacuously satisfied by so-called analytic concepts. Thus the concept of something that is either F or -F is instantiated when any other concept is instantiated. An analytic concept can be constructed out of any concept whatever if that concept is o.nly supplemented by a disjunction and the contradictory concept; and it would seem to follow that any concept whatever can thus be made to be

~252

CATEGORIES AND TRANSCENDENTAL ARGUMENTS

primitive. This fact does not, however, vitiate my criterion. For that criterion can be defended by saying that it is vacuously satisfied by analytic concepts while it can be significantly satisfied only by concepts that are nonanalytic.

What is important here is to note that the restriction just placed on my criterion is not ad hoc. Analytic concepts are, to be sure, instantiated by every conceptual framework; but they are also compatible with the: existence of any such framework; hence, the idea of an analytic concept is too broad to recover the sense in which the notion of primitiveness is used in metaphysics and epistemology.

One more thing should be noted at the outset : Arguments purporting to show that there are primitive concepts can be formulated independently of the distinction between concepts and the things which instantiate them. To prove that a concept is primitive is to show that the things instantiating them are primitive; hence, to, show that a concept is primitive does not depend upon characteristics which are peculiar to concepts. Any argument which undertakes to show that there are primitive concepts whose instantia- tions need not be primitive would not count as a transcendental argument. For all such an argument would show is that we choose certain concepts as primitive. It would not show that they in fact are primitive. The success or failure of a transcendental argument cannot, therefore, depend upon any characteristic peculiar to concepts; thus any reference to the theory of concepts can be safely omitted in asking about the possibility of transcendental arguments.

1 do not believe that there are any transcendental arguments. I do not, that is, believe that there is a specific kind of argument the conclusion of which shows that a concept must be primitive. I shall argue to this conclusion in two stages. First, I shall consider the conditions which philosophers have laid down for such arguments, showing that these conditions do not distinguish a transcendental from any other kind of argument. Secondly, I shall examine several specimens of transcendental arguments in order to ask whether they satisfy these conditions. And I shall argue that no argument can fulfill the conditions which philoso?hers have set down for transcendental arguments.

I. The Kantian Heritage

Kant lays down two conditions for transcendental arguments. I propose to separate them from his doctrine of synthetic a priori propositions and his

253

M.S. GRAM

theory of intuition. For you can accept Kant's theory of transcendental arguments and not be committed to accepting either of these latter doctrines. Kant's conditions, then, are these. First, a transcendental proof shows the truth of propositions claiming the application of primitive concepts to experience. This condition is stated in the Kritik. A typical statement of it

is this :

Through concepts of the understanding pure reason does, indeed, establish secure principles, not however directly from concepts alone, but always only indirectly through the relation of these concepts to something altogether contingent, namely, possible experience. 2

A transcendental proof will establish that primitive concepts are applicable to possible experience. The second condition for such a proof is that an appeal to experience for proof of transcendental propositions is circular; hence, an appeal to experience to refute ~ transcendental proposition is bound to imply the very proposition it is intended to refute. Kant puts it this way: He says that "though it [a transcendental proposition] needs proof, it should be called a principle, not a theorem, because it has the peculiar character that it makes possible the very experience which is its own ground of proof. TM

There is one immediate consequence of these conditions as Kant states them: The first condition is both question-begging and superfluous. It is question-begging because acceptance of it as a condition merely presupposes an independent way of showing which concepts are primitive. It is superfluous because, as we shall see, Kant's second condition alone will, if successful, do everything that this condition purports to do. And this is merely a consequence of Kant's own program: He set out to, identify primitive concepts in the Metaphysical Deduction of the first Kritik; hence his ability to find such concepts does not hinge on his ability to give a transcendental argument. But consistent as this may be with Kant's program, it nonetheless shows that Kant's first condition for a transcendental argument makes it impossible for such an argument to establish which of our concepts are primitive.

It is possible, however, to reformulate Kant's account of transcendental arguments so that it will serve as an independent means of certifying certain concepts to be primitive. Any argument fulfilling Kant's second condition

- - that the denial of a transcendental proposition assumes the truth of the claim - - would fulfill Kant's first condition. Thus in showing that a

254


proposition has a self-refuting denial, Kant would be showing that it con- tains a primitive concept. A concept is primitive for Kant when it makes experience in general possible. I take this to mean that all objects of experience must fall under a primitive concept. And if this is so, there cannot be a description of an experience which refutes a proposition that is both transcendental and true which does not itself imply that claim. To say of a proposition that it has a self-refuting denial is, then, to, imply that it claims the universal application of a concept to e x p e r i e n c e - - which is just to, claim that the concept in question is primitive in Kant's sense.

But is Kant's second condition an effective criterion of a transcendental argument ? I think not. The operative phrase in Kant's statement of the criterion is "ground of p roo f " : A transcendental proposition is said to make possible the very experience which is its ground of proof. This is the reason why the denial of such a proposition is said to be self-refuting. But to say, as Kant does, that the negation of a transcendental proposition is self-refuting does not mean that the mere denial of a transcendental proposition implies its contradictory. To say that a transcendental proposition makes possible its own ground of proof means that the only grounds available for supporting such a denial would only imply the proposition being denied.

Consider as example of this kind of argument the way in which Kant undertakes to establish the propositions in the Analytic of Principles. The negation of these propositions does not imply the propositions being denied : To deny that every event has a cause or that every appearance is either a substance or an accident does not imply that some events do, have causes or that there are some appearances that are either a substance or an accident. What does imply transcendental propositions of the kind I have just mentioned is the set of propositions describing evidence for the negation of such propositions. Thus to describe experiences that are not experiences of causal sequences or substances and accidents is to imply both the original transcendental propositions and their negations.

This, then, is the condition which Kant lays down for every transcendental argument. But what is wrong with it is not that propositions do not fulfill it but rather that a large class of analytic propositions would also fulfill such a requirement. I recognize that, on Kant's theory, synthetic a priori

propositions are meant to fulfill this condition as well. Whether this claim is true is beside the point. What is not beside the point is that, even if there

255

M.S. GRAM

are synthetic a prio~'i propositions that fulfill Kant's second condition, Kant's theory of transcendental arguments does not permit him to distinguish a proposition containing a primitive concept from an analytic proposition. The sense in which a transcendental proposition makes possible its own ground of proof is just that in which one very large, class of analytic propositions makes possible its ground of proof. And if this is true, Kant will not have succeeded in distinguishing genuinely transcendental propositions from those which are not.

That he does not in fact succeed can be. shown as follows. Notice that Kant talks about experience as a possible ground of proof for transcendental propositions. He says that an appeal to experience would make such a proof circular because transcendental propositions make. possible the very experience which would be used to, establish them. And the same' would hold for any attempt to disprove transcendental propositions: The experience used to disprove them would imply the very propositions which are being disproved. This is a characteristic which does not distinguish transcendental propositions from those' analytic propositions which are true by explicit definitions. Consider the proposition, "All brothers are male siblings." This and kindred propositions make. possible, their own ground of proof as long as experience is considered as such a ground. Thus consider a description of the experience which will refute the proposition in question. It is

this : 1) Some brothers are not male siblings.

But from (1) it follows that 2) All brothers are male siblings.

This fulfills Kant's second condition. For the analytic proposition in question here makes possible its ground of woof as long as that ground is a description o,f what is in the world; hence, a description of evidence that would refute such a proposition would also. imply the original proposition in the same way that (1) implies (2). But why does (1) imply (2) ? The reason is that (1) is an implicit contradiction. It has this status because the explicit definitions of the terms involved permit us to deduce a contradiction from (1). And since a contradiction implies any proposition, (1) implies its own negation. It is not an admissible answer to this to say that (1) might be introduced into the argument as a synthetic proposition. If this were an admissible, move, then (1) would not be evidence refuting the original proposition at all just because the sense of at least one of the

256

CATEGORIES AND T R A N S C E N D E N T A L ARGUMENTS

terms which both propositions have in common would have to, be, different :

To change the sense of any of the expressions two, propositions have, in common would prevent one f rom contradicting or implying the other. And

there is another reason why stipulating that (1) is synthetic cannot be done : To make. (1) synthetic would fail to explain how the description of the evidence adduced to refute a transcendental proposition is self-refuting. And this must be explained if we are to make sense out of Kant's claim that a transcendental proposition makes possible its own experimental ground of

proof; hence., (1) must stand as analytic. What I have been arguing so far is, however, limited. I have been trying

to show only that it is impossible, to, hold both that a transcendental proposition is analytic and at the same. time. that the propositions purporting to refute them are synthetic. But this leaves open another way of re.-

formulating what I have been calling Kant's second condition so that we can distinguish propositions that are properly transcendental from those that

are analytic. W e might hold that the. class of transcendental propositions is the same as the class of propositions which are both synthetic and which

have self-refuting negations. I f this modified condition could be fulfilled, then the objection I have been making to, Kant 's conception of a transcendental argument would fail.

The modified condition cannot, however, be fulfilled. Let us suppose,

first, a specimen of a transcendental proposition which is synthetic in that it has a self-consistent negation and, secondly, a synthetic proposition which

contradicts such a proposition. But it is easy to see that the second proposition need not imply the first proposition. Thus we are left where we were before. A proposition negating a transcendental proposition is either analytic or synthetic. If it is the former, it will imply the transcendental proposition only if that proposition is itself analytic. And if it is the latter, it need not imply the transcendental proposition at all. In either case, then, Kant's second condition is not fulfilled.

My conclusion, then is that what makes an argument transcendental in

Kant's sense is not something that can distinguish a transcendental f rom an analytic proposition which is tree by virtue of explicit definitions. To say of a proposition that the description of an experience which refutes it merely implies that proposition is to say that the description in question implies a contradiction. And this does not give us a way of discovering

primitive concepts at all. I am not denying that such a criterion exhausts

257

M.S. GRAM

Kant's ways of establishing which concepts are primitive. My present point is only that these ways do not depend upon a characteristic which can be attributed to a kind of argument scheme. Everybody knows that Kant's way of specifying these primitive concepts in the body of the first Kritik occurs in the Metaphysical Deduction; but what is wrong with this procedure is that he specifies such concepts by enumeration and thus does not give us a characteristic argument by which the concepts he cites in that deduction can be segregated from any other concepts. But if this is so, then there are, so far as Kant's description of a transcendental argument goes, no transcendental arguments. For if an argument is transcendental if and only if it fulfills the conditions Kant lays down, such an argument will not be distinguishable from the kind of argument which can be constructed out of the denial of any analytic proposition which is tree by explicit definition; hence, no such argument can show any concept to be primitive.*

II. The New Transcendentalism : P. F. Strawson

Strawson is the most recent proponent of a kind of argument which purports to show which of our concepts are primitive. I take him to accept the criterion that I mentioned at the outset : A concept is primitive when the instantiation of any other concept implies the instantiation of the primitive concept? Strawson's suggestion of finding such concepts is this : To deny that they are instantiated is to pretend "to accept a conceptual scheme, but at the same time quietly [to reject] one of the conditions of its em- ployment. TM You have produced a transcendental argument, then, when you have shown that the denial of a proposition implies the existence of a conceptual scheme which it explicitly rejects.

But are there such arguments ? Strawson gives us three principal candidates for such an argument. They purport to show that material bodies are the basic particulars of our conceptual scheme. I shall take them in turn.

A. Tl~e ~rgument for the Continued Existence of Particulars" Despite Discontinuity of Observation. The present argument purports to establish the following transcendental proposition :

P : "There are perceptual particulars which are continuous although our observation is not."

To deny P, for Strawson, is to say that the identity of particulars through time is "something feigned or illusory or at best doubtful. ''7 There are two distinct arguments given for the truth o,f P. The first runs like this :

258


1) " [ W ] e are never willing to, ascribe particular-identity. 8 [the contradictory of P]

2) We have a "new, a different, spatial system for each new continuous stretch of observation. ''9

[from (1) by the lemma that every ascription of particular-identity assumes the existence of a single spatio-temporal framework]

3) There is "no question of doubts about the identity of an item in one system with an item in another. ''1'~

This argument, although it is valid, is anything but transcendental. So far from implying P, it merely establishes our knowledge of the falsity of P. To show that the skeptic is wrong only in that he merely doubts when he can in fact know merely establishes what is not a primitive concept.

But there is another argument which Strawson gives for P. It runs as s :

1) No perceptual particular is continuous through space and time. [the contradictory of P]

2) To say that any particular is discontinuous implies that something else must be continuous : "We cannot attach one occasion to, another unless,

from occasion to occasion, we can reidentify elements common to different occas ions . ' ,11

3) If something must be continuous through interrupted observation, then (1) is false.

[from (2)]

The crucial move here is from (1) to (2). And it is governed by the following lemma :

To say that any particular is discontinuous assumes the existence of a single, inclusive, particular which is the spatio-temporal system in which the two particulars are located.

The lemma is, I believe, true.: One's ability to say that two perceptual items are numerically different assumes that they differ with respect to something that does not change; hence, identification o.f particulars as different implies identification of a particular which is continuous.

Is this argument transcendental ? It does imply, to be sure, a proposition contradicting the claim that no perceptual item is continous through time; hence, anyone wanting to hold that no particulars are continuous must also adopt a conceptual scheme in which at least one particular is continuous in order to say that there, are discontinuous particulars. But there are two comments to be made about the argument. One., it does not prove P but

259

M.S. GRAM

rather a considerably weaker version; namely, that at least one particular is continuous. The conclusion of the argument is compatible with the fact that all particulars save one are discontinuous. Two, the conclusion does not establish that the concept of a particular is primitive - - let alone the concept of a material body - - but only that the concept of a continuous spatio4emporal system is logically prior to our ability to identify any particular.

B. The Argument Against Massive Reduplication. The transcendental claim here is that there are unique descriptions or, equivalently, that identifying descriptions apply uniquely. That it is a transcendental claim is to be shown, according to Strawson, by examining the consequences of accepting the contrary claim; namely, that no, definite description need apply uniquely. The argument refuting this view runs as follows :

1) Since particulars cannot be directly located (i.e., demonstratively identified), they must be described in general terms. ~:2

[ex hypothesi] 2) A particular which cannot be directly located cannot be given a unique

description. 13 [by definition]

3) But "even though the particular in question cannot itself be demonstratively identified, it may be identified by a description which relates it uniquely to, another particular which can be demonstratively identified. 1~

[from (1) and (2) by the lemma that one identifies particulars u- niquely either by unique descriptions or by relating them to particulars that have such descriptions]

4) Thus it is false to say that no particulars can be uniquely identified. [from (1), (2), and (3)]

The strategy of this argument is simple: To describe any particular in general terms implies the existence of unique descriptions. The existence of unique descriptions is a condition of the existence of general descriptions of particulars.

But is this a transcendental argument ? The claim that there are only general descriptions of particulars is made to imply the existence of unique descriptions because such a daim implies the existence of particulars. This would fulfill the condition which Strawson lays down for transcendental arguments were it not the case that the argument in question is question- begging. When you deny that there are such things as unique descrip-

260


tions, part of what you deny is that there is a unique spatio-temporal

framework within which particulars are located. For this, too, is a particular. And in one respect, at least, all particulars are in the same leaky boat : For the denial of unique descriptions diberh~upt is also a denial of a unique spatio-temporal system; hence, the existence of such a system cannot be. assumed in a demonstration of unique descriptions. But once you have denied the. existence of such a framework, the claim that there, are. no unique descriptions does not imply its opposite. I conclude that the present argument is not transcendental. To say that a claim assumes the conceptual framework which it denies is to, say that it assumes that framework to verify that claim. And this is very different from saying, as the present argument says, that a transcendental argument may exempt one entity from the class of entities whose existence it calls into question.

C. The Argument for the' Existence of Material Objects. The last specimen of transcendental reasoning is an argument purporting to shove that there are material objects, is The argument is as follows.

1) There are no material bodies. [ex hypothesi]

2) The only things that exist are minds or their properties, thoughts, ideas, and sensations.

[ f rom (1) by the lemma that whatever exists must be either material bodies and their properties or minds and their properties]

3) But the identification of mental entities is parasitical on material bodies : The identification of anything mental requires that we. identify it by reference to. a material body.

4) Therefore, there are material bodies. [from (1), (2), and (3 ) ]

The crucial premise here is (3). And that premise rests on the impossibility of identifying an entity which Strawson calls pure consciousness. TM Such an entity cannot exist because the assumption of its existence entails the denial of the distinction between my experience and the experience of others; hence, we cannot identify experiences by references to, something that is purely mental. But since the only other alternative, is the recognition of material bodies and their proporties, we must recognize that material bodies exist in order to identify something that is purely mental? 7

But this argument is as question-begging as the one before it. The argument here presupposes the impossibility of a pure consciousness - - which is just to presuppose the impossibility of the claim that everything in the world

261

M.S. GRAM

is a mind or property of a mind. Strawson argues against the possibility of a pure consciousness as fotlows. He claims that the description of a world in which only mental items existed would entail the collapse of the distinc-

tion between qualitative and numerical difference. Thus to say that only mental items exist is to allow only temporal points to localize them; but to say that something is located at two points of time does not establish that there are two items rather than one located in this way. And so Strawson concludes that the assumption o.f a purely idealistic world would entail the

collapse of the distinction between numerical and qualitative identity; and the further conclusion is drawn that individual center's of consciousness cannot be distinguished on an idealistic theory. Hence the assertion that there exist only mental items is said to imply the very conceptual scheme it pur-

ports to, deny. But is this really the case ? I do not believe that the collapse of the

distinction between numerical and qualitative identity is entailed by denying

the existence of material objects. The only way in which the denial of material objects can be made to imply the existence of material objects is by the addition of the further premise that any claim implying the collapse of numerical and qualitative identity in the items of our experience must imply the existence of material bodies. But why should anyone think that such a conclusion follows ? There is, of course, one way to, get it to follow. I am

referring to, the. method of stipulation. And although the way in which Strawson practises that method is subtle, what he' does comes, nonethdess,

to the same thing. What the denial o r material objects implies is that we are not able accurately to describe the conditions o.f the identification of mental items in our world. But if this is a consequence o.f the hypothesis that there are no, material objects, then, according to Strawson, that hypothesis cannot be true.. For it conflicts with the truth of other propositions whose truth we must acknowledge if we are correctly to. describe the con-

ceptual scheme we have. But this isl after all, question-begging. To say that the denial of the

existence of material bodies implies the collapse of the distinction between numerical and qualitative identity is simply not true. What is true is that such an implication holds if you assume that what accounts for this distinction is spatial location. And this is just where the argument becomes question-begging. What is being assumed is that spatial location must account for numerical difference. And this, in turn, assumes that the only

262


thing that accounts for spatial location is something that is not mental at

all. The basic assumption of the argument, then, is that what accounts for numerical difference is also what accounts for the distinction between mate-

rial and mental objects. And yet this is the very thing that is being denied by the claim that there are no material objects. But once you have robbed the argument of this question-begging assumption, all you are committed to doing is to. give another criterion for the identification of particulars. And all this amounts to is the recognition that such criteria must be different

in the idealist's world. But this is very different from saying, as Strawson's argument would force us to. say, that the denial of the existence of material

objects implies the existence of these objects.

II. The Possibility of Any Future Transcendental Argument The basic structure of the arguments I have been considering is this. To

show that the identification of any item of experience, implies the identification of some other item is to, show that our concept of the latter is more basic. And if there are items in our experience identification of which

is assumed by identification of any other item in experience, the concept of the former would be primitive. This feature of a transcendental argument would explain why the statement of the grounds on which the. denial of the conclusion of such an argument would merely imply that conclusion.

The denial of such a conclusion would imply that there are items of experience which do not require the identification of certain other entities for their

own identification. And if the proposition being denied is transcendental, then it would be possible to show that the denial of the proposition tacitly

assumes the very entities whose existence' is being denied. Strawson's candidates for transcendental arguments are meant to illus-

trate this pattern. Thus the denial that there are continuants is cast in the form of the claim that instantaneous particulars do not require the existence of continuants as conditions for their identification. The denial that there are unique descriptions is presented in the form of the claim, that we do

not have to assume the uniqueness of anything in experience in order to describe items within experience. And the denial that there are material

objects is set forth as the claim that the items we do experience do. not require material objects for their identification. All of the arguments I have been considering, then, have this in common : The self-refuting character of the denial of a transcendental proposition is exhibited when it is asserted

263

M.S. GRAM

that the. conditions of identifying a certain entity do, not require the existence' of the entity claimed to be categorial.

I do not deny that arguments having this characteristic can be produced. I think that Strawson has produced such an argument when he shows that the identification of momentary items in our experience implies the existence of at least one item that is not momentary. This argument does not, to be sure, prove as much as Strawson thinks it proves. It does not, for example, prove that the basic particulars of our conceptual schemes are continuants, for it establishes only that there is one continuant in our conceptual scheme - - which is quite compatible: with the claim that all other particulars of that scheme are momentary. This does not, however, prevent the argument Strawson gives from fulfilling the requirement he lays down s transcendental arguments.

But does this mean that there are in fact transcendental arguments even though most of the specimens I have been examining fall short of this status ? I do, not think so. There is a way of showing that, even though there are arguments which can be made to fulfill the condition laid down for transcendental arguments, such arguments do not show which of our concepts are categories. Showing that one. item of experience is identificationally dependent upon another is not enough to show that the concepts involved in this relation are categories. And this is not shown just because it has not been shown that the identificational dependence is asymmetrical. Thus arguments showing identificational dependence do not exclude the possibility that the so-called primitive item in our experience in fact depends upon the identification of items that are not considered to be primitive. And if it can be shown that such a possibility cannot be excluded, then it will follow that no argument purporting to show identification dependence can show which of our concepts are primitive.

Consider once again the argument showing that the identification of momentary items in experience depends upon the existence of something permanent. It is also the case that what is permanent depends for its identification upon what is momentary. What the argument in question establishes is that we are able to identify items in experience as numerically different because they are different within the same spatio-temporal framework. But this does not establisti that our concept of such a framework is primitive or catego.rial. For we identify the spatio-temporal framework as permanent from occasion to occasion by identifying the occasions as numerically

264


different. We can discover that the system of spatio-temporal relations is

permanent only if we can discover that the moments of our observation are different.

That what is permanent in experience depends for its identification on what is momentary can perhaps be shown as follows. To say o,f something that it is permanent can mean at least two things, only one of which is

relevant to the present argument. To say that something is permanent can mean that it cannot change; and this cannot be interpreted to mean that it

is self-identical. This is, however, a sense of "permanence" that can be truly predicated o,f even momentary entities. There is a second sense of "permanence," however, which is peculiar to temporal passage. To say that

something is permanent is to say that it is numerically the same at different moments of time.. This sense of "permanence" is relevant to the identifi-

cation of momentary items of experience. And the condition of recognizing something as permanent is the ability to say truly that it can be assigned different temporal predicates. In this sense, then, nothing is temporally permanent unless it can have characteristics that are momentary.

There is an obvious but mistaken response to this move = . And it runs like this. The demand I have made for a criterion of identifying the per-

manence of the system of spatio-temperaI relations might be rejected as inappropriate. For the sense in which the system if permanent might be regarded as different from that in which continuants within the system are

permanent. For, it might be argued, the system itself does not itself en- dure through time and occupy different spatial positions. Ordinary continuants do; hence, it seems that we do not need a criterion for permanence when we say that the system itself is permanent. The demand for a criterion arises only if an entity changes in some respect; but if there is

no respect in which an enity changes, then it would seem that a criterion to distinguish permanence from change with respect to, that entity cannot be legitimately made.

But the fact is that the distinction between permanence and change can be made with regard to the system itself. The demand for a criterion is a demand for stating the way in which we recognize something. And since permanence is just as much an element in our experience as change, it is legitimate to ask how we recognize the presence of permanence in our experience. In this respect, then, the: permanence of the system is no more exempted f rom the demand for a criterion of recognition than is the numeri-

265

M.S. GRAM

cal difference of items within the system. That the system can be perman-

ent depends upon the existence of momentary entities. To say that the sys-

tem is permanent presupposes that we can say that some of the items within

the system are momentary; hence, unless you can say that there are momen-

tary items in experience, you cannot say that there is any item which is

permanent. Thus the conclusion of the original argument can be turned as

follows. It is tree that the identification of momentary items assumes a

permanent element. But it is just as true that the identification of a perman-

ent item assumes momentary itema, identification of these two elements is in-

terdependent; hence, an argument which shows the identificational depen-

dence o f one upon the other is incapable of showing that the concept of one is more basic than that of the other.

What I have offered thus far is only an example of how transcendental

arguments can go wrong. The point can, however, be generalized to cover

any attempted demonstration of identification dependence. Consider the con-

cept of any item in experience which is said to be identificationally prim-

itive in that we must identify every other item of experience with respect

to it. We are always free to demand a criterion 0.f identification for the

supposedly primitive, item. And the reason we can make such a demand is

simple: We can demand a criterion of identification for every item in ex-

perience. And since a primitive item in experience cannot provide its own

criterion, the only items available to. us are those which we are forced to

call derivative. If this is so, then it is impossible, to show with respect

to any item in our experience that it is identificationally primitive.

Some philosophers might resist the conclusion that I have just drawn

by arguing that there are contexts in experience in which identification

does not assume a criterion. And if this is true, then it would appear

that there, is a glaring exception to my claim that we can demand a cri-

terion or recognition for every item in our experience. The argument pur-

porting to demonstrate this exception runs as follows. There are, it is said, at least two questions that we can ask about any item of experience.

For one thing, we can ask whether it is self-identical. For another, we can ask whether it is the same something we have previously experienced.

We need a criterion to answer the second question; but we do not, it is argued, require a criterion in order to ascribe self-identity to, an item

of experience. And the reason we do not require a criterion in such contexts is that the demand for a criterion allegedly arises only when you

266


want to discover whether something is the same through a plurality of pre-

sentations. The conclusion, then, is that there is a sense in which no

item of experience needs a criterion for identification. The same point can be put differently as follows : A criterion is required for recognition

of items in experience but not for ascribing the predicate o,f self-identity to them.

The focegoing argument can be undercut in two ways. It can be shown,

first, that its conclusion can be conceded without thereby resuscitating transcendental arguments. And it can be shown, secondly, that the conclusion of the argument does not really co.nstit:~.te an exception to my claim

that we can demand a criterion for every item of experience. Take these counter-arguments in turn. W e can concede that ascriptions of self-identity

do not require criteria. But does this show that there are identificationally primitive items in our experience ? I think not. Ascriptions of self-

identity are made for every item of our experience quite independently of whether it is primitive or not. And so it follows that sense in which items of experience do not require criteria does not permit any inference

to the conclusion that the items in question are identificationally primi-

tive. But this is not all. The conclusion of the argument in question does

not present a genuine exception to my claim that every item of experience requires a criterion. All the present argument establishes is that the ascription of self-identity to an item of experience does not require the prior identification of anything else. And this is very different from saying that there are items in experience which require no criterion of iden-

tification at all. It is only this latter claim that is incompatible with the claim I am making. But the argument here, so far from establishing

this claim, is in fact irrelevant to, it. The search for a transcendental argument, then, is futile. For even,

as sometimes happens, if an argument for identificational dependence can be produced, it is not the kind of argument that can establish which of

our concepts are primitive. And it is, a fortioH, not the kind of argm- ment which can distinguish the kinds of concepts with which a metaphysi- cian deals from any other kinds of concepts. The aim of a program, of ontology based on transcendental arguments is the enumeration of identificationally primitive elements in experience. And such an inventocy would include those items upon which every other item in experience is identifi-

267

M.S. GRAM

cationally dependent but which themselves do not depend upon any other items for identification. But there can be no such item of experience. To say of any item that it does not require a criterion of identification is to say that we can never know when we are in possession o f an item of that kind. But since any entity whose existence is claimed demands a criterion of identificatic~n, no entity can be identificationaUy primitive except at a price which cannot be paid; the price, namely, of rendering it impossible for us to, know when we have instances of such an entity.

N O T E S

i P.F. Strawson, Individuals (London : Methuen, 1959), P. xo.

2 A737; cf. A78z=Bszo. 3 A737; cf. A782=BBxo. 4 I have dealt with Kant 's general theory of transcendental arguments more exhaustively in my "Transcendental Arguments ," Nous , vol. 5 (February, z97:t), pp. I5-;~6. Cf. ~aakko Hint ikka 's animadversions on the foregoing in "Transcendental Argumen t s : Genuine and Spurious," Nous , vol. 6 (September, :t97z), PP. 274-8z; and. my reply, "Hin t ikka and

Spurious Transcendentalism," forthcoming. 5 He states this point in a confusing way; hence, he says that there are arguments which reveal "categories and concepts which, in their most fundamental character, change not at all" (Individuats, p. :to). This does not show any concept to be !~rimitive simply because no concept can change, whether it is primitive or not. To say that a concept has changed into another concept is to say - - what is self-contradictory - - that one entity has become another. But since the concept which has allegedly changed is not the same concept that it was, it is self-contradictory to say of any concept that it has become another concept.

Strawson, o~v. cir., p. 35. ? I ignore the ambiguity of this c l a im: It can mean either that P is false or that it is possible that P is false. Strawson would, in any case admit that it is logically possible for P to be false, for he wants to establish that the conjunction of the negation of P and certain

other propositions generates a contradiction.

8 Strawson, op. cir., p. 35- o Strawson, op. cit., p. 35. 10 Strawson, op. cir., p. 35- l l Strawson, op. cir., p. 32. 12 Strawson, op. cit.s p. zz. 13 Strawson, op. cir., p. 2:t. 1~ Strawson, op. cit.t p. 2:t. 15 INhat followa is a summary of one of the themes of Indivlduals, Chapter III, pp. 87ff. 13 Strawson, op. cit., pp. 98ff. 17 Kant gives a similar argument when he undertakes to refute idealism. The material idealist holds, according to Kant, that the only experience is inner experience; and if the idealigt is problematic, he holds that we infer the existence of outer things from inner experience. Kant seeks to refute both by deducing the falsity of the claim from the

assumption of its trttth, as follows : z) I am conscious of my own existence in time

[This is implied by the idealist claim that the only experience is inner experience.] z) There is something permanent in time.

[From (~) by virtue of the lemma that consciousness of one's own existence in time implies permanence.]

268


~) There is a pe rmanen t outs ide me as dis t inct f rom a representati.on of a pe rmanen t outs ide me.

[From (2) by the l e m m a that representa t ions are momen ta ry whi le that of which they are representa t ions is not.]

4) Thus it is fa lse that the only experience is inner experience.

[From (I), (2), (3), and (4)] I men t ion this paral lel to S t rawson only to leave it, for i t is obvious ly fal lacious. Kant takes it for g ran ted at B277 tha t the permanence of the self can be de termined only wi th respect to outer experience and that , consequently, we are " una b l e to perceive any deter- mina t ion of t ime saved through change in outer r e l a t ions . " W h a t Kant is do ing here, then, is this . H e is (:[) equat ing the pe rmanen t in t ime wi th subs tance (~228), (2) equat ing subs tance wi th m a t t e r (B228), and (~) concluding that wha t is pe rmanen t m u s t be outs ide us. W h a t wrecks the a r g u m e n t is that i t fa i l s to exclude the very poss ib i l i ty which it was mean t to exclude; the possibi l i ty , namely , that wha t we call the pe rmanen t in t ime is menta l and no~ mater ia l .

269

categories and transcendental arguments

Documents