Philosophy against Misosophy


Michael R. Butler


Essays by Me

Essays by Others



The Transcendental Argument for God’s Existence

Michael R. Butler


III. Transcendental Arguments in Recent Philosophical Literature


Though transcendental arguments (TAs here-after) have been used by various philosophers since Aristotle,[41] it is with the publication of P. F. Strawson’s Individuals,[42] that TAs have become a prominent fixture in contemporary philosophy. The discussion of the structure and nature of TAs has generated a good deal of controversy.  In what follows, I will overview the contemporary literature on TAs with the aim of setting the most common criticisms of TAG in sharper focus. 


A. The Nature of Transcendental Arguments

Before plunging into the current debate over TAs it will helpful to first sketch the form TAs gene-rally take as well review a few definitions proffered in the literature.  The aim here is not so much to analyze the nature of TAs—this would require a paper of its own—but to get something of a feel for what kind of arguments they are.[43]

A TA takes on (roughly) the following form: For x to be the case, y must also be the case because y is the necessary precondition of x; since x is the case, y must be the case. By itself there is nothing particularly distinguished about this form of argu-ment.  For with it I could argue that having parents is the precondition of having grandparents, and since I have grandparents, I must have parents. Though this shares a structure similar to that of a TA, it is not a TA because it appeals a posteriori knowledge—in this case, knowledge of basic biolo-gical facts.  TAs are distinguished from this type of argument by the fact that they appeal only to a priori knowledge—what we can know without any appeal to experience.  From this it follows that form alone is not the distinguishing feature that sets TAs apart from other arguments.[44]  Most recent commentators are in agreement with this conclusion.  Grayling is representative:

[T]o argue, or reason, or proceed transcendentally, or to employ stan-dard philosophical techniques tran-scendentally, is just to argue or pro-ceed, etc., with a certain aim in mind and a certain subject-matter to hand ...there is nothing distinctive about the form of TAs, and that what is distinctive about them is their aim and subject matter.[45]

What, then, is the aim and subject matter of TAs?  Before answering this, it is helpful to consider what is not the aim or subject of TAs.  TAs should not be confused with paradigm-case and/or polar concept arguments popularized by Austin, Ryle and others of the so-called Oxford school.[46]  For while these types of arguments share a similar form with TAs, they differ greatly in the type of conclusion that is inferred.  Specifically, TAs have highly generalized conclusions while polar concept arguments are much narrower.  A brief comparison should bring out this distinction.  Austin argues that the skeptic’s appeal to illusion does not work because the term ‘illusion’ makes sense only in a context of having some real things to compare with it and thus everything could not be an illusion (or better put, it makes no sense to say everything is an illusion).  Assuming this argument works, the conclusion in somewhat parochial: it defeats only one particular skeptical challenge.  The skeptic, though, can simply propose to toss away both words and offer a fresh challenge.  A TA aims at something more cosmopolitan. Unlike polar con-cept arguments, TAs attempt to demonstrate that x (whatever x may be) is a precondition of experi-ence. In a word, then, the difference between a TA and a polar concept argument is one of scope; the latter asks what are the necessary preconditions for the intelligible use of a small set of terms, the former is concerned with the use of a much larger set.

With this aim in mind, we are now in a position to survey a few representative definitions found in the contemporary literature. According to Anthony Bruekner a TA is:

an argument that elucidates the con-ditions for the possibility of some fun-damental phenomenon whose exis-tence is unchallenged or uncontrover-sial in the philosophical context in which the argument is propound-ed.  Such an argument proceeds de-ductively, from a premise asserting the existence of some basic pheno-menon (such as meaningful dis-course, conceptualization of objective states of affairs, or the practice of making promises), to a conclusion as-serting the existence of some interes-ting, substantive enabling conditions for that phenomenon.[47]

Brueckner points to three salient ingredients that are involved in TAs. First, they begin with some unchallenged or uncontroversial phenome-non or experience (e.g., I speak a language or I have an idea of a single spatio-temporal system of material things). Next, from this phenomenon they proceed deductively to a conclusion.  Finally, this conclusion is a non-trivial condition for the possi-bility of the phenomenon.  Grayling further refines the nature of a TA’s conclusion:

... the aim of transcendental argu-ments is to establish the conditions necessary for experience, or experi-ence of a certain kind, in general; and, at their most controversial, to estab-lish conclusions about the nature and existence of an external world, or other minds, derived from paying attention to what has to be the case for there to be experience, or for experience to be as it is.[48]

The conclusion of a TA either tells us something about the conditions necessary for experience or something about the nature of reality.  That it, the conclusion of a TA may either be conceptual or on-tological.

Ross Harrison adds one more characterization of TAs:

Transcendental arguments seek to answer scepticism by showing that the things doubted by the sceptic are in fact preconditions for the skepti-cism to make sense. Hence the skep-ticism is either meaningless or false. A transcendental argument works by finding the preconditions of meaning-ful thought or judgment. For example, skepticism about other minds sug-gests that only the thinkers them-selves might have sensations. A tran-scendental argument which answered this scepticism would show that a precondition for thinking oneself to have sensations is that others do so as well. Expressing the skepticism in-volves thinking oneself to have sen-sation; and the argument shows that if this thought is expressible, then it is also false.[49]

According to Harrison, the refutation of the skeptic is the primary objective of a TA.  This view is supported by other commentators of TAs.  Barry Stroud, for example, maintains, “transcendental arguments are supposed to demonstrate the im-possibility or illegitimacy of [the] skeptical chal-lenge by proving that certain concepts are neces-sary for thought or experience.”[50] Similarly, Strawson asserts that one who “advances such an argument [transcendental] may begin with a pre-mise which the skeptic does not challenge ... and then proceed to argue [what the] necessary condi-tion of the possibility of such experience is.”

While it is true that most (though certainly not all) contemporary TAs are used as skepticism-refuting arguments, this aim is not a necessary feature of them. Indeed, the view that TAs are essentially anti-skeptical in nature rests upon a historical mistake.  This mistake is revolves around a faulty interpretation of Kant’s understanding of a TA.  In the next section I offer an analysis of the nature of Kantian TAs.


B. Kant and Transcendental Arguments

Before analyzing and evaluating contemporary TAs, it is obligatory to say a few words about Kant.  Contemporary proponents of TAs, such as Strawson, often cite Kant as their inspiration. The TAs they offer, however, differ in at least one fundamental way from Kantian TAs.  Whereas Kant’s understanding of a priori knowledge, and hence transcendental reasoning, was closely tied to his view that the forms of sensibility and concepts of the understanding in some way consti-tute experience,[51] contemporary TAs tend to avoid anything resembling Kant's transcendental idealism.

Because of this, a number of philosophers have accused contemporary advocates of TAs, such as Strawson, of denuding Kant’s TAs of their distinc-tiveness. Indeed, some go so far as to claim that the contemporary reformulations do not deserve the title of “transcendental argument” at all. Before one is tempted to cast aside this whole debate as a petty etymological squabble,[52] it should be realized that more than nomenclature or lexical proprietorship is at stake. Kant is, among other things, the father of TAs.[53] Thus it would be a rather odd conclusion to state that his arguments were not TAs. Indeed, one is tempted to say that whatever Kant is doing in the first Critique, he is arguing transcendentally.  This is perhaps what Jaako Hintikka is getting at when he states that “the first order of business in any discussion of such arguments is to try to see what Kant under-stood by the term.”[54]  Ross Harrison is even more explicit, “Since Kant invented the label, any-thing properly called a ‘transcendental’ argument must have some analogy to the arguments which Kant used.”[55] These considerations have weight, and so, taking Hintikka's advice, understanding Kant's use of transcendental arguments shall be my first order of business.

In his preface to the second edition, Kant men-tions in a note that the only addition to the Critique was the Refutation of Idealism (Bxl), the purpose of which was to once and for all do away with the scandal of philosophy: that the existence of things outside us must be accepted merely on faith. A cluster of controversial issues revolve around the interpretation of the Refutation that are of import for the contemporary debate about TAs.  Should, for example, the Refutation be viewed as Kant’s main TA?  Much of the recent literature on TAs in-sists that is the case.  But this raises an immediate question.  If the Refutation is indeed Kant’s main TA, how are we to explain its absence in the first edition?  A related issue concerns the Refutation’s place in understanding Kant’s transcendental pro-ject. Even if it is not Kant’s main TA, it appears at least to be the clearest. Should we therefore use it at a heuristic in understanding the more difficult arguments contained in the Deduction?  A final issue has to is whether the Refutation is an exam-ple of a transcendental argument at all.  Since the answer of this last question is needed before we tackle the other questions, it is to this I shall turn.

To help answer this question, a preliminary question needs to be addressed. What did Kant mean by the term, ‘transcendental’?  Perhaps the clearest expression is found in the Introduction of the Critique:

I entitle transcendental all knowledge which is occupied not so much with objects as with the mode of our know-ledge of objects insofar as this mode of knowledge is to be possible a priori (A11/B25).[56]

Elsewhere Kant says that the only way we can know something about objects a priori is knowing what we put into them (Bxviii).  He also flatly de-nies that the word ‘transcendental’ has any refer-ence to knowledge of things; it refers only to cogni-tive faculties.[57]  Hence, transcendental know-ledge is knowledge of what our cognitive faculties “impose” upon the world. From this Hintikka con-cludes:

Thus a transcendental argument is for Kant one which shows the possibility of a certain type of synthetic know-ledge a priori by showing how it is due to those activities of ours by means of which the knowledge in question is obtained[58] (Hintikka, p. 275).

Hintika rightly takes Kant’s main TAs to be those in the Transcendental Aesthetic and Tran-scendental Deduction.  Though he admits that they are not as lucid as they could be (a mild under-statement), Kant indicates his intention to show that a priori knowledge always involves the active processes of the mind.  In a programmatic pas-sage, Kant writes: “But only the productive synthe-sis of the imagination can take place a priori” (A118).  

What then of the Refutation?  Hintikka (again rightly) maintains that the Refutation should not be viewed as Kant’s paradigm TA.  Indeed, he thinks it is not even a TA at all. It is an ad hominem argu-ment showing that the idealist contradicts his own position in the very process of stating his position. In other words, this is a very different kind of argu-ment than those found in the Aesthetic and the De-duction.[59]

A further consideration bolsters his claim.  It is difficult to construe the Refutation as the paradigm TA in the Critique since it was a later addition to the second edition.  And unless one wants to argue that the first and second editions offer significantly different philosophical outlooks (which nobody does) it is absurd to think of it as his central TA.  The retort to this is that while the Refutation is not Kant’s central TA, it is his clearest example of one.  Thus in understanding it we can better under-stand his more difficult ones in the Aesthetic and Deduction.  I will return to this contention present-ly.  What is important to note now is that Kantian TAs always involve appeals to constitutive or productive knowledge.  Appeal to transcendental psychology is thus the sine qua non of TAs.

If Hintikka is right, most of what passes as TAs today are, as we shall see, un-Kantian.  He further adds that if they are un-Kantian, they are spurious.  What this second charge amount to, though, is not clear. If he uses “spurious TAs” as a synonym for “un-Kantian TAs,” there is no problem. But if he that “spurious TAs” are not TAs at all, then we have to ask why this is the case. The next question we must address, then, is whether his take on Kant is correct.

According to Grayling, Hintikka is wrong—or as he says, Hintikka gets “the wrong end of the stick altogether.”[60] He tries to prove this by pointing to Kant’s use of the metaphor of a legal deduction at the beginning of the Transcendental Deduction (A84/B116).  There Kant asserts that, like a legal deduction which tries to justify a particular claim or possession (i.e. prove that the claim or possession was lawfully obtained), so a transcendental deduc-tion tries to justify the employment of certain concepts.  And it does this in a way that is different from an empirical deduction.  The latter type of de-duction merely shows how a concept is acquired through experience and “therefore concerns, not its legitimacy, but only its de facto mode of origination” (A85/B117).[61] Grayling immediately concludes from this that “the task [of transcen-dental deductions] is to provide a vindication of title, not to show where the concepts come from.”

The problem with Grayling’s reading of Kant is that to Kant, these are not separate questions.  In order to vindicate the use of a concept one must be able to demonstrate where the concepts come from.  This is shown by Kant’s later remarks on transcendental proofs in The Discipline of Pure Reason.  There he says that transcendental proofs are always direct or ostensive and explains this by saying that this type of proof “is that which combines with the conviction of its truth insight into the sources of its truth” (A789/B817, my emphasis).  In other words, transcendental proofs or deductions are ostensive because they point to the source or origin of truth.

As Hintikka has already noted, it is abundantly clear that the main transcendental proofs in the Critique are found in the Aesthetic and Deduction. It is also clear that understanding the TAs in them is extremely difficult.  This is why many take the comparatively easier argument of the Refutation as a model to help understand them.  And since the basic form of the Refutation is a reductio [ad absurdum], it follows that the Aesthetic and Deduction must also be reductiones. This practice has been a common if not often stated assumption among Kant scholars: they approach his more difficult transcendental proofs by comparing them to easier and allegedly analogous ones. That is, they pay attention to Kant’s alleged practice rather than his theory.[62] This has resulted in much confusion as to what the exact nature of Kant’s argument is.[63] It is precisely because of the clarity of the passage at A789/B817 that it should be used as a heuristic device to understand Kant’s main transcendental proofs.[64]  Thus the way to understand these difficult arguments is via his me-tatheoretical comments in the Disciplines of Pure Reason.

But what then are we to do with the above pas-sage from Kant (A85/B117)? Based on Grayling’s reading, Kant must be contradicting himself. This throws doubt on Grayling’s reading. Does Kant not contrast a transcendental deduction with an empi-rical deduction that depends on the “mere” origin of the concept? Does this not clearly state that whatever transcendental deductions turn out to be, they are not concerned with the source of the concept in question?  A careful study of the text leads us to a negative answer.  Grayling mistakes the gist of what Kant is getting at.  Kant indeed is making a contrast between transcendental and empirical deductions, but this contrast is not that the latter is concerned with the mode of origination while the former is not.  Rather the latter is con-cerned with only its de facto or empirical mode of origination (e.g. the actual perception of an object) while the latter is concerned with the a priori mode of origination (in this case, the transcendental ego).  Grayling’s contention is based on a profound mis-reading of the text. And so it is he, not Hintikka, who gets the wrong end of the stick.

Before we acquiesce to Hintikka and claim that contemporary TAs are un-Kantian because they do not make appeals to constructive knowledge, it is important to point out the obvious. Kant did use an argument analogous to contemporary TAs in the Refutation and the Second Analogy. Even Hintikka admits this much. So in another sense it is not cor-rect to call contemporary TAs un-Kantian. They are indeed Kantian arguments, but they are not Kantian transcendental arguments. What then of the charge that they are spurious? At this point the question has become a mere linguistic matter. Nothing of philosophical important hangs on the answer at all.

Since it has been the practice to call contempo-rary TAs ‘TAs’ and since, as Walker has observed, Kant did not take out a copyright on the term,[65]there seems to be no pressing need for a change in nomenclature. It is little more than schoolmarmism (pace Hintikka and Harrison) to insist that all TAs must comply with tight Kantian strictures in order to warrant the honorific title “Transcendental Argu-ment.”  Thus it is legitimate to continue the practice of calling contemporary TAs ‘TAs’, bearing in mind that not all TAs are Kantian in nature.


C. Strawson

Over the last century there have been many examples of TAs in the analytic philosophy tradi-tion.  Frege, Wittgenstein, Davidson, Putnam and Searle all have offered TAs for conclusions as various as the necessity of senses (meanings), the impossibility of a private language, the claim that beliefs are generally true, the claim that radical skepticism is false and the truth of external realism.[66] It is with the publication of Strawson’s Individuals in 1959, however, that transcendental arguments came to the forefront of philosophical inquiry.

Strawson has spent much of his career resuscitating what he thinks are Kant's important insights while at the same time removing the dross.  His Individuals and later The Bounds of Sense[67] make extensive use of TAs while at the same time eschewing synthetic a prioris and transcendental psychology. Strawson thinks that once the swamp of intuition and the rest of Kant’s transcendental psychology are put aside, there re-mains a core insight into how our conceptual scheme works. According to Strawson, Kant’s work is best seen as analysis of our conceptual scheme rather than a transcendental deduction of catego-ries.  By taking away from Kant what he considers to be this core insight, Strawson employs TAs to solve two traditional problems in epistemology: skepticism about other minds and skepticism about the external world.  Since the use of TAs against the latter has a longer pedigree, I will concentrate on it.  Whatever is said about this TA, however, applies, mutatis mutandis, to the other.

Strawson’s argument against the external world skeptic is deceivingly simple.

There is no doubt that we have the idea of a single spatio-temporal sys-tem of material things; the idea of every material thing at any time being spatially related, in various ways at various times, to every other at every time. There is no doubt at all that this is our conceptual scheme. Now I say that a condition of our ha-ving this conceptual scheme is the unquestioning acceptance of particu-lar identity in at least some cases of non-continuous observation.[68]

Except perhaps the last sentence, Strawson's claim seems to be perfectly correct. It is just a fact that we think that every material object is in some relation to every other physical object—spatially and temporally.  Of course, we do not usually speak of having a spatiotemporal system of ma-terial things, nor are we generally cognizant of this fact.  But the way we use language demonstrates that we presuppose this conceptual scheme.  A few pedestrian examples prove this rather easi-ly.  Think of a father telling his son to move his bicycle off of the driveway so he can park the fami-ly car or a dry cleaner who tells the customer to pick his suit up on Tuesday morning.  In both of these scenarios, all individuals involved presup-pose that material objects share a spatial relation to each other (either the car or the bicycle can be parked in a particular spot but not both) and temporal relations (the suit that is dropped off today will be the same one that will be ready on Tuesday morning).  If they did not presuppose these relationships communication would quickly break down.

But now the skeptic comes along and chal-lenges these presuppositions perhaps the car dis-appears when no one is looking and a different (though similar) one appears when it is observed again. What the skeptic is challenging us to do is justify our belief that our conceptual scheme is true. To the skeptic Strawson replies:

He pretends to accept a conceptual scheme, but at the same time quietly rejects one of the conditions of its employment.  Thus his doubts are un-real, not simply because they are lo-gically irresoluble doubts, but be-cause they amount to the rejection of the whole conceptual scheme within which alone such doubts make sense.  So, naturally enough, the alternative to doubt which he offers us is the suggestion that we do not really, or should not really, have the conceptual scheme that we do have; that we do not really or should not really, mean what we think we mean, what we do mean.  But this alterna-tive is absurd.  For the whole process of reasoning only starts because the scheme is as it is; and we cannot change it even if we would.[69]

The claim is that in order to get his doubts off the ground, the skeptic must presuppose the very conceptual scheme he calls into question.  And this illicit importation uncovers the very absurdity of his position.

It is important to realize that Strawson thought[70] that this once and for all defeats the external world skeptic. Were this the case, it would have been a true milestone in the history of philosophy. Like most bold claims from well-recognized philo-sophers, though, it has become somewhat of a cause célèbre attracting many defenders and critics.


D. Objections


1. Stroud

One of the most trenchant critics of Strawson is Stroud.  In his elegant article entitled simply “Tran-scendental Arguments,”[71] Stroud sets out to show that TAs in general and Strawson’s in particu-lar fail to deliver on their promise of silencing the skeptic.

Stroud takes Strawson’s argument to be:

1. We think of the world as containing objective particulars in a single spatio-temporal system.

2. If we think of the world as con-taining objective particulars in a sin-gle spatiotemporal system, then we are able to identify and re-identify particulars.

3. If we can re-identify particulars, then we have satisfiable criteria on the basis of which we can make re-identification.

6. Objects continue to exist unper-ceived.

Stroud thinks the argument stops here.  And if this is the case, it is plainly a non sequitur.  Stroud has no qualms with the move from (1) to (3).  The problem is that they do not get Strawson to (6). Additional premises are, therefore, needed, and Stroud is ready to oblige with

4. If we know that the best criteria we have for the re-identification of parti-culars have been satisfied, then we know that objects continue to exist unperceived.       


5. We sometimes know that the best criteria we have for the re-identifica-tion of particulars have been satis-fied.       

Without these or similar premises, the argu-ment is not valid.  But since (5) is a statement of fact, the skeptic would not let Strawson (or any-body) get away with it. So from this Stroud con-cludes that Strawson is wrong to take the skeptic as denying (6). Instead he should be taken as de-nying whether we can know the truth of (5); whe-ther we are justified in ever asserting it—and this is far different from the exotic Berkeleyan claim that objects cease to exist when not observed. So (5) is superfluous.  All that is really needed is (1)—(4), which say that if we think of the world as contain-ing objective particulars in a single spatiotemporal system, then we can know whether objects conti-nue to exist unperceived.  Or better put, if it is meaningful to speak of objective particulars, then we must have a way of knowing whether they con-tinue to exist unperceived.  But this is a verification principle, a principle that maintains that for propo-sitions to be meaningful, there must be a way, at least in principle, of knowing or testing whether they are true or false.

The problem with this is if TAs rely upon an im-plicit verification principle, then TAs turn out to be superfluous for the skeptic is directly refuted by such a principle.  If we sometimes know that the best criteria we have for the re-identification of particulars have been satisfied, then we know that skepticism is false.

Later in the article Stroud goes on to generalize this point by maintaining that is hard to imagine that there could be any TA that does not in some way rely on a verification principle to bridge the gap between belief and reality.  The reason being is that any TA will maintain that denying the truth of a given discourse presupposes that discourse.  But this will not do since the skeptic can simply bite the bullet and say that the whole discourse is mud-dled.  And because the skeptic has this route al-ways open, the transcendental arguer will have to make a universal claim and hold that all discourse is meaningless unless a certain set of propositions (Stroud calls it “the privileged class”) are neces-sarily true given the general discourse.  There are several problems with this, but the main one is that this does not set the skeptical problem to rest.  All this proves is that the privileged class must be believed; it does not prove that the privileged-class propositions are in fact true. To ensure this, a verification principle is necessary. But, again, why spend so much time discussing the intricacies of a TA when the verification principle is all that is ne-cessary to vanquish the skeptic directly?

But if we have the latter, the former is super-fluous and, thus, there is nothing special about transcendental arguments.  To put it crassly, to di-vorce meaning from truth is to render TAs impo-tent, but to conjoin meaning and truth is to render them (TAs) extraneous.


2. Körner

One of the most prevalent criticisms of tran-scendental arguments was first raised in the con-temporary literature by Stephan Körner.[72] rner argues that while Kant's arguments provide sufficient conditions for human experience (in this case the categories of understanding), they are not necessary conditions.

The person propounding a transcen-dental argument assumes that every and any thinker employs the same categorical framework as he does himself, and tries to show that, and why, the employment of this particu-lar framework is ‘necessary.’  The de-fect of all transcendental arguments is their failure to provide a uniqueness proof, i.e., the demonstration that the categorial framework is unique.[73]

The objection is that while it may be possible to prove that some conceptual scheme is sufficient for experience, it is not possible to prove that it is necessary. The reason that no uniqueness proof is possible is laid out nicely by rner. rner con-tends that there are only three possible ways of establishing a schema's uniquness.

First, to demonstrate the schema’s uniqueness by comparing it with ex-perience undifferentiated by any me-thod of prior differentiation . . . . Se-cond, to demonstrate the schema’s uniqueness by comparing it with its possible competitors . . . . Thirdly, one might propose to examine the schema and its application entirely from within the schema itself, i.e., by means of statements belonging to it.[74]

These three possibilities are subject to powerful criticisms, however.  The first possibility cannot es-tablish a schema’s uniqueness by comparing it with undifferentiated experience, since in order for a comparison to be made, statements about the un-differentiated experience would have to be ex-pressed by making recourse to some prior differ-entiation of experience.  That is, there is no way to formulate a statement about experience without differentiating experience.  But even if  such a comparison could be made, all that could be de-monstrated is that the schema under consideration does represent or reflect the undifferentiated ex-perience. But this does not establish the unique-ness of the schema since it may be that other schemas also represent or reflect the undifferen-tiated experience as well.

Körner offers two criticisms of the second possi-bility. First, in order to establish the uniqueness of one schema, all other possible schemes would have to be exhibited and refuted.  But as Rorty has pointed out, we could never know whether all possible schemas have been exhibited.  “Nothing in heaven or earth could set limits to what we can in principle conceive.”[75] No matter how many al-ternative schemas one can conceive and refute, there is always the possibility that there is another schema (or several!) that has not yet been con-ceived.  Griffiths sums up this criticism nicely:

[T]o establish [the uniqueness of a schema] would involve an exami-nation of all possible conditions. But by what criterion could we determine that all possible conditions had been enumerated? Further, this means an examination of all possible conditions in detail: a humanly impossible task.[76]

Second, Körner maintains that the very act of demonstrating a schema’s uniqueness by compar-ing it with another schema is contradictory.  Such a demonstration would presuppose that another schema, a schema distinct from the one whose uniqueness is trying to be proved, can possibly re-present or reflect undifferentiated experience.  But if this is presupposed, the demonstration is self-contradictory.  “[This] requires anyone who uses it to admit before the argument begins that what he is trying to prove is false, otherwise he could not even try to prove it.”[77]

The third possibility (attempting to establish the uniqueness of a schema from within the schema itself) is quickly dispatched by Körner.  All that can be shown by such a proof is how the schema differentiates a region of experience.  It cannot show that it is the only possibly schema able to differentiate that region.


E. In Defense of Transcendental Arguments


1. Stroud

Stroud’s argument against TAs is that they ei-ther rely on dubious factual premises or that they require a verification principle.  Either way, the TA will not do the requisite work against the skeptic since he will either reject the factual premise out of hand or reject the verification principle upon which the TA depends.  Thus Stroud places defenders of TAs on the horns of a dilemma: either the verifi-cation principle must be dropped and with it the claim that the TA gets us to what the world must be like, or the verification principle can be maintained, but only at the price of rendering the TA unneces-sary—the verification principle answers the skeptic directly.  Almost all recent defenders of TAs have tried to meet this dilemma by tackling the first horn.[78]  Three basic strategies have been employed in this endeavor.

First, Stroud himself puts forth the possibility of denying the skeptic’s claim that it is sufficient that we must merely think the world is a certain way (have a certain conceptual scheme) and not that the world must be a certain way.[79] The skeptic is just wrong in asserting that mere belief is enough. The only way to account for the fact that our con-ceptual scheme must be believed (or, conversely, to say that it makes no sense to question our con-ceptual scheme) is that it must be true.  Our con-ceptual scheme is impossible to deny because it “corresponds” to the way in which the world actually is.

Stroud points out that it is extremely difficult to conceive of a defense for the strong modal claim of this position (i.e., the world must be this way in order to make experience possible).

. . . [H]ow can truths about the world which appear to say or imply nothing about human thought or experience be shown to be genuinely necessary conditions of such psychological facts as that we think and experience things in a certain way, from which the proofs begin?[80]  

Not only has no argument been advanced toward this end, but what such an argument would look like is difficult to imagine.

Second, a defender of TAs may accept Stroud’s contention that such arguments merely give us the conclusion that we must believe the world to be a certain way and not that it really is that way.  In order to bridge the gap from belief to the world, the defender need not make recourse to a verification principle, but can contend that the way in which we conceive of the world constitutes the way the world actually is.  That is, there is in reality no gap between our thoughts and the world since the world does not exist independently of our conception of it.  This, of course, is a version of idealism. Given idealism, the move from our conceptual scheme to reality is immediate and the TAs that make this idealist assumption do indeed tell us something about the world.[81]

Accepting idealism is a high price to pay to get TAs to work, however. Indeed, Strawson’s reinter-pretation of Kant was motivated by the attempt to purge out his problematic transcendental idealism and preserve his ingenious insights into the analysis of concepts.  Idealism is, thus, the very thing that Strawson (and other proponents of TAs) attempts to get away from. Moreover, a  mover to-ward idealism will not in actuality strengthen a TA since idealism is itself subject to serious criticisms (e.g. those of Reid, Moore and Russell).  And even if a good case can be made for idealism, the assump-tion of idealism renders the anti-skeptical TAs un-necessary since idealism refutes skepticism directly.

Third, one may concede that Stroud is correct in arguing that TAs must rely on a verification principle to get from our conceptual scheme to the world, but rather than give up on TAs altogether, propose a more modest view of what TAs are supposed to prove.[82]  Grayling, for example, makes a distinction between two types of TAs—option-A and option-B.  These two options differen-tiate between TAs that have metaphysical conclu-sions (“The world must be thus and so”) and those that have conceptual ones (“We must not deny thus and so because of our conceptual scheme”).  For obvious reasons, Grayling is pessimistic about the prospect for option-A TAs.  As for option-B TAs, they have troubles of their own.  Such arguments attempt to establish that given our conceptual scheme, certain concepts must be pre-supposed.  That is, it makes no sense to question our conceptual scheme since, in so doing, we must presuppose it.  But the skeptic’s reply to this line can be anticipated: Even if certain concepts are es-sential for this conceptual scheme, it may not be for another.  Our current scheme might indeed pre-suppose causation, but that does not mean that other possible conceptual schemes must do so as well.  This difficulty is addressed in the following section.


2. Körner

Körner’s criticisms of the first and third possible ways in establishing a conceptual scheme’s uni-queness are sound.  There is no way to prove the uniqueness of a conceptual scheme by comparing that scheme to undifferentiated experience or by testing its application from within the conceptual scheme itself.  His criticisms of the second possibi-lity are, however, subject to serious criticism.

Körner, recall, argued that the second type of proof, the proof that attempts to demonstrate the uniqueness of a given conceptual scheme by comparing it with its possible competitors, is impossible for two reasons.  The first is that it is “self-contradictory in attempting a ‘demonstration’ of the schema’s uniqueness, by conceding that the schema was not unique.”  What Körner means by this is that in attempting to establish the unique-ness of a particular conceptual scheme by com-paring it with possible competitors, this com-parison denies the very uniqueness of the scheme it is trying to establish.

This argument, however, rests upon a confu-sion.  It is true that if such a uniqueness proof as-sumes that there are other legitimate conceptual schemes that are in competition with the one that it is being established, such a proof would be self-defeating.  To say, for example, that such and such a conceptual scheme is the only possible one and then go on to compare it with another genuine conceptual scheme is an absurd affair.  However, the other possible conceptual schemes that are compared to the one that is trying to be estab-lished by a uniqueness proof need not be con-sidered as genuine competitors.  These other con-ceptual schemes appear to be in competition with ours, but on closer examination, it is show that they are not.  Such an argument would look like something as follows: “One would think that this supposed conceptual scheme poses a challenge to our conceptual scheme, but on closer examination, this supposed conceptual scheme is, in actuality, not a genuine conceptual scheme.”

The distinction between a genuinely competing conceptual scheme and a merely apparent com-peting conceptual scheme shows that we can make sense of a uniqueness proof that compares one conceptual scheme with its possible com-petitors.  But here we face a further diffi-culty.  Granted that conception of such a proof is not self-defeating, how could such a proof actually proceed?  There are at least to possible ways.  First, one can argue that the competing conceptual schemes violate some necessary pre-condition of experience.  That is, the competing conceptual schemes are not genuine because that are unable to account for the experience we have.  Second, one can argue that all other pos-sible conceptual schemes are dependent upon (and thus reducible to) the one he is trying to establish as unique.

Schaper contends that the first option (demon-strating that a supposed competitor scheme vio-lates some necessary precondition of experience) is out of the question.[83]  This is because, on the one hand, if the necessary precondition of experi-ence belongs to the supposedly unique conceptual scheme, the argument is question-begging.  All that is being demonstrated is that because the competing scheme does not adhere to the sup-posedly unique scheme’s necessary preconditions of experience, it is not a genuine competitor.  This is hardly a persuasive argument.  But if, on the other hand, the necessary preconditions of experi-ence do not belong the supposedly unique concep-tual scheme, that scheme is thereby shown to be not unique.  Either way, the first option of demon-strating uniqueness is impossible.

The second option is to argue for the unique-ness of a particular conceptual scheme by proving that it is the only possible conceptual scheme. Such an argument is advanced by Donald David-son.[84]  Davidson contends that the notion of a completely foreign conceptual scheme that philo-sophers such as Quine and Kuhn advance is inco-herent. Although his arguments are subtle and tied in with his extensional semantics, the gist of his contention is not difficult to understand.  In order to recognize something as an alternative concep-tual scheme, we must be able to map it onto our own conceptual scheme.  If a conceptual scheme is so different from ours that we are not able to ac-complish such mappings, we would not even re-cognize it as a competing conceptual scheme.  This is because the only way for us to recognize some-thing as a competing conceptual scheme is that we compare it to our own.  When no such comparison is possible (where two “paradigms” are “incommensurable,” to borrow from Kuhn), there would be no way for us to recognize it as a con-ceptual scheme at all.   In other words, the notion of an inconceivable (i.e., unrecognizable) concep-tual scheme is thus incoherent.

This brief discussion of Davidson is a nice segue into Körner’s second reason why this kind of uni-queness proof is impossible.  Recall that in order to establish the uniqueness of one conceptual scheme, all other possible conceptual schemes must be refuted.  But surely this is an impossible task since we do not have the time, let alone the capacity, to conceive of every possible conceptual scheme and then proceed to refute them one by one.  If this were indeed the task, there would be no hope of ever accomplishing a uniqueness proof.  But fortunately for TAs and their defenders, this is not what they set out to accomplish.  Rather than refuting every possible alternative conceptual scheme, TAs endeavor to simply refute one—the negation of conceptual scheme being defen-ded.  Førster comments are insightful.

A transcendental argument ... in order to establish a particular condition of knowledge or experience, proceeds by considering an alternative, that is, the negation of the condition, and subsequently demonstrates its inter-nal incoherence. Clearly, this ex-hausts the field of possible alterna-tives to this condition. For although one may perhaps imagine different philosophical positions or conceptions based on the negation of the original condition, this would not add to the number of alternatives to it.[85]

Since there is nothing particularly daunting about disproving the negation of a conceptual scheme, a uniqueness proof for a conceptual scheme is certainly not impossible.

An objection may be raised at this point that, while this may be what a TA sets out to do, in practice it only refutes a particular competitor of the conceptual scheme that it is trying to estab-lish.  So, for example, Strawson’s argument—that in order to re-identify particulars we must have an objectivity condition—proceeds by refuting the sense datum hypothesis.  But notice that the sense datum hypothesis is simply one version of the negation of the objectivity condition (the “non-objectivity condition”).   In refuting a particular version of this “non-objectivity condition,” Straw-son intends to refute the “non-objectivity condi-tion” in general.  In other words, the refutation of one version of the “non-objectivity condition” is meant to show that all versions of such a condition are reducible to absurdity.  Of course one may argue that in refuting the sense datum hypothe-sis Strawson is only showing a problem intrinsic to it and not the “non-objectivity condition” in gener-al.  But in order to make this objection work, an argument must be provided by the skeptic to show that this is indeed the case.  If no such argument is provided (and it is difficult to image what such an argument would look like), the refutation of one particular version of the “non-objectivity condi-tion” should is nothing less than a refutation of “non-objectivity condition” in general.  And be-cause the “non-objectivity condition” is the nega-tion of the objectivity condition, a refutation of the former provides a proof for the latter.[86]



[41] In Metaphysics (v. 1061a 5-1062b) Aristotle demonstrates the transcendental necessity of the law of contradiction.

[42] Individuals: An Essay in Descriptive Metaphysics (London: Methuen & Co., Ltd., 1959).

[43] It should be noted at the outset that some philosophers argue that there is nothing at all distinctive about TAs.  Moltke Gram is, perhaps, the leading advocate of this view. The following articles by Gram are representative: “Transcendental Ar-guments,” Noûs, vol. 5, no. 1 (February 1971), 15-26, “Must We Revisit Transcendental Arguments?” Philosophical Studies 31 (1977), 235-248.  Gram’s arguments, however, have not been persuasive to most commentators on TAs.

[44] See T. E. Wilkerson, “Transcendental Argu-ments,” Philosophical Quarterly 20 (1970), 200-212 and Ralph C. S. Walker, Kant: The Arguments of the Philosophers (London: Routledge and Kegan Paul, 1978) for insightful discussions of the logic of TAs.

[45] The Refutation of Skepticism (London: Duck-worth, 1985), 94.

[46] Michael Dummett denies that there is (was) such a school.  Assuming he is correct, the name is still useful and in this context innocuous.  See his “Oxford Philosophy” in Truth and Other Enigmas (Cambridge: Harvard University Press, 1978), pp. 431-436.

[47] “Transcendental Argument” in Robert Audi, ed., The Cambridge Dictionary of Philosophy, 2nd ed. (New York: Cambridge University Press, 1999), 925.

[48] “Transcendental Arguments” in Jona-than Dancy and Ernest Sosa, eds., A Companion to Epistemology (Oxford: Basil Blackwell Ltd., 1992), 507.

[49] “Transcendental Arguments” in Edward Craig, ed., Routledge Encyclopedia of Philosophy, vol. 9 (New York: Routledge, 1998) 452.

[50] Barry Stroud, “Transcendental Argu-ments,” Journal of Philosophy 65 (1968), 241-256.

[51] Even a cursory reading of the Critique makes this clear.  In the very beginning remarks of the Introduction to the second edition Kant states: “But though all our knowledge begins with experi-ence, it does not follow that it all arises out of experience.  For it may well be that even our empirical knowledge is made up of what we receive through impressions and of what our own faculty of knowledge (sensible impressions serving merely as the occasion) supplies from itself” (B 1).

[52] Although some of the debate is indeed trivial.  For although Kant made the term a technical one and, thus, must be conceded a certain propriety over it, philosophers should be more concerned with the success and failures of the arguments at hand rather than lexicographical particularities—Johnson, after all, was modest enough to call himself a harmless drudge.

[53] Although Kant never actually used the term “transcendental argument”; his closest term is “Transzendentalen Deduktion”.

[54] “Transcendental Arguments: Genuine and Spurious,” Noûs, Vol. 6 (1972), p. 274.

[55] “Atemporal Necessities of Thought; or, How Not to Bury Philosophy in History,” Reading Kant: New Perspectives on Transcendental Arguments (Oxford: Blackwell, 1989), p. 42.

[56] All references from the Critique of Pure Reason are from Norman Kemp Smith, trans. (New York: St. Martin’s Press, 1965).

[57] Prolegomena, p. 294 (Academy edition). Cited in Jaakko Hintikka, “Transcendental Arguments: Genuine and Spurious,” Noûs, Vol. 6 (1972), p. 275.

[58] Hintikka, 275.

[59] For slightly different reasons, David Bell also considers it a mistake to view The Refutation as a transcendental argument.  See his “Transcendental Arguments and Non-Naturalistic Anti-Realism,” in Robert Stern, ed., Transcendental Arguments: Problems and Prospects (Oxford: Clarendon Press, 1999), 189-210.

[60] The Refutation of Scepticism, 80.

[61] Grayling misleadingly paraphrases this phrase as “mere ‘mode of origination.’” Grayling, The Refutation of Scepticism, 80.

[62] Graham Bird is one of the few who makes his commitment to this practice explicit. “Kant’s Tran-scendental Arguments,” in Eva Schaper and Wil-helm Vossenhuhl, eds., Reading Kant (Oxford: Blackwell, 1989).

[63] Perhaps the most famous example of this is found in Strawson’s misreading of the Deduction in his The Bounds of Sense (London: Methuen, 1966).

[64] By ‘clarity’ I mean that it is clear what Kant intended the goal of transcendental proofs to be.  I do not mean that it is at all clear how Kant actually goes about doing it.

[65] “Transcendental Arguments and Scepticism,” in Schaper and Vossenkuhl, eds., Reading Kant, p. 56.

[66] Gottlob Frege, “Thoughts,” in Logical Investi-gations, P. T. Geach, ed. and trans. (New Haven: Yale University Press, 1977); Ludwig Wittgen-stein, Philosophical Investigations, trans. by G. E. M. Anscombe (Oxford: Blackwell, 1958); Donald Davidson, Inquiries into Truth and Interpretation (Oxford: Oxford University Press, 1984); Hilary Putnam, Reason, Truth, and History (Cambridge: Cambridge University Press, 1981); John R. Searle, The Construction of Social Reality (London: Penguin, 1995).

[67] The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason (London: Methuen, 1966).

[68] Individuals, 35.

[69] Individuals, p. 35.

[70] Strawson has since changed his mind.  See his Skepticism and Naturalism: Some Varieties (New York: Columbia University Press, 1985).

[71] Journal of Philosophy 65 (1968), 241-56.

[72] In a fascinating study of early Post-Kantian-ism, Paul Franks shows that Gottlob Schulze offered a similar objection to Kant and his K. L. Reinhold in his Aenesidemus (1792).  “Transcen-dental Arguments, Reason, and Scepticism: Con-temporary Debates and the Origins of Post-Kantianism,” in Stern, ed., Transcendental Argu-ments, 111-45.

[73] Categorial Frameworks (Oxford: Blackwell, 1974), 72. Quoted in Eckart Førster, “How Are Transcendental Arguments Possible?” in Schaper and Vossekuhl, 15. Roderick Chisholm’s strictures on TAs are similar to Körner’s. See his “What Is a Transcendental Argument?” in The Foundations of Knowing (Minneapolis, MN: University of  Minnesota Press, 1982), 98. Originally published in Neue Hefte für Philosophie 14: 19-22.

[74] “The Impossibility of Transcendental Deduc-tions,” The Monist, 51 (1967), 320-21.

[75] Richard Rorty, “Transcendental Arguments, Self-Reference, and Pragmatism,” in P. Bieri, R.-P.  Horstmann, and L. Kròger, eds. Transcendental Arguments and Science (Dordrecht: D. Reidel Pub-lishing Co., 1979), 82.

[76] A. Phillips Griffiths, “Transcendental Argu-ments,” Proceedings of the Aristotelian Society, suppl. Vol. 43 (1969), 171.

[77] Eva Schaper, “Arguing Transcendentally,” Kantstudien, 63 (1972), 107.

[78] Richard Rorty actually tackles the second horn. He agrees with Stroud that a verification principle is necessary, but not the empiricist kind that Stroud actually tackles the second horn.  He agrees with Stroud that a verification principle is necessary, but not the empiricist kind that Stroud maintains.  Rorty argues that this empiricist form of verificationism is not only dubious, but demon-strably false. TAs that rely on such a principle are therefore directly refuted. However, Rorty main-tains that a Peircean form of verification (to know the meaning of a term is to know inferential rela-tions), is not obviously false and can do the requi-site work of answering the skeptic.  But this answer to the skeptic takes us no further that Stroud.  It merely shows that the skeptic must assume the truth of our conceptual scheme in order to call it into question.  But this, of course, does not estab-lish anything about the way the world is. Richard Rorty, “Verificationism and Transcendental Argu-ment,” Noûs, Vol. 5, n. 1, (February 1971), 3-14.  See Anthony Brueckner, “Transcendental Ar-guments I,” Noûs, 17 (November, 1983), 551-575 for a trenchant criticism of Rorty.

[79] “Kantian Argument, Conceptual Capacities, and Invulnerability,” in Paulo Parrini, ed., Kant and Contemporary Epistemology (Dordrecht: Kluwer, 1994).

[80] Stroud, “Kantian Argument,” 234 quoted in Stern, ed., Transcendental Arguments, 7.

[81] Bernard Williams suggests that both Kant and contemporary philosophers who employ TAs impli-citly assume some form of idealism.  See his Prob-lems of the Self (Cambridge: Cambridge University Press, 1973).

[82] This move is the most common in the TA lite-rature.  See, for example, Peter Hacker, “Are Tran-scendental Arguments a Version of Verification-ism?American Philosophical Quarterly, Vol. 9, No. 1 (January 1972), 78-85; Eckart Førster, “How are Transcendental Arguments Possible? in Reading Kant, 3-20, Ralph C. S. Walker, “Induction and Transcendental Argument,” in Stern, ed., Transcen-dental Arguments, 13-29, Robert Stern, “On Kant’s Response to Hume: The Second Analogy as Tran-scendental Argument,” also in Stern, 47-66.

[83] Schaper, “Arguing Transcendentally,” 107-8.  Interestingly, Schaper thinks that this type of proof is the only one possible.

[84] “On the Very Idea of a Conceptual Scheme,” in Donald Davidson, Inquiries into Truth and Interpretation (Oxford: Clarendon Press, 1984), 183-198.

[85] Førster, ”How Are Transcendental Arguments Possible?,” 15.

[86] The example in this paragraph is due to Førster.


Next IV. TAG Again and V. Conclusion 


Beginning of Article and Table of Contents