Comments on the Paradoxicality of Zen Koans[*]
By Michael E. Levin

The Journal of Chinese Philosophy
V. 3 (1976)
pp. 281-290

Copyright 1976 by D. Reidel Publishing Company




Professor Chung-ying Cheng's attempt[1] to apply some techniques of contemporary logical theory to the analysis of the koan dialogic exchanges of Zen Buddhism is laudatory. He claims in particular that koans are paradoxes in a suitably generalized sense of "paradox," and that the point of their paradoxicality is a group of ontological principles. However, while welcoming his effort, I have serious reservations about Cheng's success either in explicating what is paralogistic about koans or in explaining their role in Zen metaphysics. I will argue here that: (1) So far as Cheng has shown, no koan is, strictly speaking, paradoxical. (2) Cheng's mistake arises from a failure to see that one sentence can be used in a speech-act to assert another sentence. (3) The principles of ontic non-commitment and intersubstitutability (explained below) misconstrue the purport of koans.



It will aid clarity to examine at the outset Cheng's attempt to broaden the structure of paradoxes. He calls 'P is true if and only if [2] P is false' the '(A) form' of paradox (84) and proposes (85) six alternative forms, of which

        (B)     P is meaningful iff P is not meaningful
        (C)    P is intelligible iff P is not intelligible

are typical. He claims that

        (H)     P is q iff P is not q

is a generalization of all these forms.

    This is a misconception. The general form of a paradox is a proof, for some S, of S iff -S. Any sentence of any of the forms (B) through (H) are of this form. If we let S be 'P is meaningful', (B) becomes

        (B')     S iff -S




and so on for (C) - (H). In these cases the sentence both affirmed and denied is a second-order sentence ascribing a semantic predicate to another sentence. But this is just to say that we have a second-order claim of the form (A). The variety of semantic predicates that generate statements equivalent to their own denial does not induce a corresponding variety in types of paradox, any more than paradoxes generated by set-theoretical paradoxes do. Consider Russell's, in which a set K is proved to be a member of itself iff it is not a member of itself. Here the property asserted and denied of its (non-sentential) subject is: membership in K. Of a particular thing, K as it happens, we prove that it both does and does not have this property. But it would be absurd to list 'x is a member of K iff x is not a member of K' as a further form of paradox. We would have to go on and list 'x is the set of all ordinals iff x is not the set of all ordinals' as a yet further form, and so on.

    Thus, if koans give rise to paradoxes (in a second-order language), there must be some sentence P and some semantic property M - truth, meaningfulness, or whatnot - such that the koan states or entails M (P) and -M (P). The interdeducibility of, say, the meaningfulness and meaninglessness of P would show that both 'P is meaningful' and its denial are provable. We would have a statement of form (B), but (B) is not an alternative to but a special case of (A). An alleged Zen paradox of this form, then, would be a paradox of the familiar sort, and must be judged as such.

    Cheng's argument that koans do in fact generate sentences or exchanges both meaningful and meaningless rests on their role in the search after Enlightenment. A koan provokes Enlightenment by an unintelligible question, statement or exchange. But, by provoking enlightenment the question, statement or exchange ipso facto becomes the reverse of unintelligible. Its very absurdity makes it profoundly sensible. As Cheng puts it,


Either the question or the answer or their conjunction... in a Zen dialogic exchange may appear to be... lacking relevance for revealing the ultimate truth and thus lacking relevance for inducing or testing Enlightenment, yet at the same time it is held... by the Zen masters that it is in virtue of... such lack of relevance for revealing the ultimate truth that that question or that answer or their conjunction... are genuinely relevant for revealing the ultimate truth (85-6).

    I will defer until Section II Cheng's more specific claim that the Zen




exchanges become relevant by making perspicuous the pointlessness of ontic commitment For now J want to ask if Cheng is accurately describing what takes place in a koan. Is the statement within the koan simultaneously meaningful and meaningless? I shall suggest that the answer is 'No', but that, rather, by uttering a meaningless assertion in a koan the Zen master makes another assertion which is quite meaningful, and it is this other assertion which is 'relevant for revealing the ultimate truth'. If so, there is no paradox in or entailed by the koan, since it is one sentence which is meaningless and another which is meaningful. What causes confusion, I suspect, is that the first is used to assert the second.

    It is well to look at an example independent of present concerns of a speaker using a sentence S to assert a sentence S' whose truth conditions differ from those of S. In fact, it is an example in which S is false and S' true. Suppose I see wife-beater M whom I mistakenly believe is holding a glass of whiskey (it's really ginger-ale). I say to you: 'The man with the whiskey is a wife-beater'. Now, suppose there is another man in the room who I have not noticed, who is holding a glass of whiskey, but who isn't a wife-beater. Now S is false, since the man with the whiskey is not a wife-beater. But if my mistake about M is pointed out to me, I wouldn't admit that what I had said was false. I would allow only that I had misidentified the man I was calling a wife-beater. Since S was false but what I asserted was true, I must have asserted something other than S when I uttered S. Evidently I was asserting S': 'M is a wife-beater'. I was using 'The man with the whiskey is a wife-beater' to assert that M is a wife-beater.[3]

    I would suggest that something similar happens in a koan. A sentence which is meaningless (or false, etc) is used to make another assertion which is not. (Koans of course differ from cocktail party mix-ups in being intentional rather than inadvertent.) Take a concrete example. The student asks the Master "When all things are reduced to the One, where does the One reduce to?". The Master replies: "When I was in Ch'ing, I bought a robe that weighed seven chin."[4] This koan exhibits the form Cheng calls QXpR. There is nothing absurd about the question Q or the response R if each is considered in isolation; yet taking R as a response to Q yields a meaningless dialogue. R has nothing to do with Q. Yet the student achieves Enlightenment (because, according to Cheng, he sees that no ontology is adequate).




    I will grant provisionally that the very inappropriateness of R to Q has revealed something very appropriate. But the critical question is whether R, 'When I was in Ch'ing, I bought a robe of seven chin' has been shown to be or transformed into an appropriate answer to Q. Has it been shown that - (QXpR), or QXpR iff -(QXpR) (only this, I argued, will yield a genuine paradox)? I think not. Rather, in saying 'I had a robe weighing seven chin', the Master is asserting (if Cheng is right) that no ontology is adequate. Now this latter sentence, 'No ontology is adequate', is the appropriate remark, appropriate to the student's search for Enlightenment. What is appropriate in this context, then, is not the answer or assertion of the master that he had a robe of seven chin, but rather another assertion that the Master used 'I had a robe of seven chin' to make: namely, 'No ontology is adequate'. We do not have any single assertion which is apposite iff it is inapposite. The sentence 'I had a robe of seven chin' retains its simple irrelevance to the student's question.

    My argument applies as well to straightforward koanic assertions of the form Rp (90), such as (14): 'I am him yet he is not me' (87). I4 is not intelligible iff it is unintelligible. It, presumably, is just unintelligible. However, the speech-act of uttering 14 makes an assertion about, arguendo, ontology, which while not absurd is also a quite different assertion than I4. So we don't have any single sentence which is intelligible iff it is unintelligible.

    There are two replies that Cheng might give. One is to claim that the distinction between sentence uttered and sentence an utterance is used to assert does not apply here, since it is the self-same sentence 'I had a robe of seven chin' which is both meaningless (as an answer to the student's question) and meaningful (as a pointer to Enlightenment). But even if this much is granted there is still no paradox. A paradox 'in relation' (85) requires that one thing, in this case a sentence, stand in a relation R (in this case 'is meaningful for') and - R to one and the same thing. But this does not hold here. The sentence is meaningless for the student's question and not meaningless for Enlightenment.

    A more radical move (suggested by Cheng in conversation) is to construe the Master as trying to dissolve accepted frames of reference by showing that any sentence can be used to make any assertion. (This would subserve ontic intersubstitution, cf. 97.) This, I think, is something




that cannot be shown, since for any S it is S's standard meaning that allows it to be used, when it is so used, to assert some S'S. The use of a sentence to make a non-standard assertion is parasitic upon its standard interpretation. Thus, I can use 'The man with the whiskey is a wife-beater' to assert 'M is a wife-beater' because 'The man with the whiskey is a wife-beater' retains its standard interpretation as the ascription of the predicate 'is a wife-beater'. If that understanding were relaxed, I could not use 'The man with the whiskey is a wife-beater' to assert 'M is a wife-beater'.

    Notice that my argument against the paradoxicality of koans does not in the end depend on my claim that (A) is the canonical form of paradoxes. Even if we allow that proving a sentence to be both meaningful and meaningless is a new kind of paradox, I have; if correct, shown that a koan does no such thing. The new kind of paradox still demands that it is one sentence that satisfies schema (B), and this is what koans do not give us.



The strangeness of the koans, then, is the use of an absurd sentence S to assert a non-absurd sentence S'. What S' is it that the Master is using S to make, and how and why are sentences like S suitable vehicles for S'? I would prefer to leave such questions to scholars of greater learning and sensitivity than mine. Yet, however presumptuous it is, I want now to argue that Cheng has misinterpreted the upshot of the Zen koan.

    Professor Cheng claims that koans reveal "a framework in which no reference to any category of things is made" (91). "[T] he emptiness of reference for this ontological structure [the one at issue in the koan in question] is justified by the emptiness of reference to the totality of ontological structures" (92). Cheng calls this perspicuous moral of the koans "the principle of ontic-non-commitment" (PON) and draws as a corollary the principle of "ontological substitution"; since no ontology is adequate, one ontology is as good as (and hence may be substituted for) another. Koans prompt reflection on these principles, according to Cheng, since only by assuming them can we make sense of koans ("resolve the paradoxicality of the given Zen paradox" [92]) (92, lines 21/35; 93, 22/32).

    Let me propose a formulation of Cheng's point in a contemporary




Western idiom. Cheng finds the point of the koan to be the transcendence of what Quine calls the problem of ontological commitment. One of the main aims of science, according to Quine, is to find the most adequate set of entities for understanding the world. Zen wants to make us see that this aim is unsatisfiable. Nor does it want to do this just as a matter of intellectual clarity. The point of ontic non-commitment is in its turn spiritual. While in the West we tend to think of the problem of ontological commitment as a purely intellectual exercise of assent to some or other ontological options, Zen regards entertaining the problem as itself part of attachment: allowing one's tranquility of mind to depend on the vicissitudes of external things (93). Transcending the need for theoretical commitment in ontology is a phase of spiritual transcendence.[5]

    My impression, as against this, is that PON is insufficiently general to capture the upshot of all koans, and just one possible way of rendering intelligible the distinctively 'ontological' koans.

    One ground for my suspicion is not only the extreme specificity of PON - do the Master's responses all have to do with his referring terms? - but the fact that PON makes good sense only in a Fregean-Quinean framework which was clearly not the one ancient Buddhists used in formulating their questions. It would be short-sighted never to view theses in one tradition as responses to problems in another. But surely here, where we are dealing with a problem whose formulation requires a background of first order logic, we must be on our guard.

    Upon reflection, moreover, the link between the provocative arbitrariness of the koans and the interpretation of them via PON is much looser than Cheng supposes. Cheng understands this arbitrariness as an arbitrariness of reference: the Zen exchange has a 'surface semantic structure' which cannot be correlated with or embedded in any commonsensical interpretation of the reference of the interlocutors. Cheng is right if he is saying that it is difficult to think of a translation, consistent with any of a wide variety of assumptions about the psychology and rationality of the participants, which gives a common subject-matter to the question and answer in a koan. Call an interpretation of a dialogue which specifies a common reference or subject-matter in an acceptable set of sortals a focus. It is indeed an aspect of an arbitrary exchange that no focus can be specified for it. But surely this is just one of a great many glosses, not the only one singled out, nor incompatible with any others,




that a non sequitur can support. The framework undermined by complete arbitrariness need not be the specific framework of ontic inquiry; it can, for example, be the more general framework whose principal assumption is that theoretical inquiry has a point. We shall look below  at such an alternative gloss.

    Professor Cheng has the Master doing what Strawson has the sceptic doing at the beginning of Individuals: "We are caught in a conflict... if we feel we have to look for the reference of the Zen question in virtue of its surface semantic demand on the one hand and at the same time feel we have to abolish the surface semantic structure and reformulate it in light of some standard framework of reference on the other" (91). There is indeed a conflict between looking for the reference of a Zen question (What entities is the question about?) and the fact that no answer to this question will also answer the question 'What is the reference of the reply to the Zen question?' My point is that this impossibility, this conflict, is not the only one. The requirement of focus is just one assumption that cannot be satisfied. An absurdist exchange violates a whole range of assumptions about communication; the assumption that a focus exists is just one. Thus, if the purpose of the exchange is to stimulate critical reflection on the assumptions it violates, it is arbitrary to pick on the existence of focuses as the assumption we are to question. (It wouldn't be wrong to read Oedipus the King as a study of the consequences of poorly-marked intersections; just peculiarly narrow.) Thus, even waiving the historical anomaly of Cheng's thesis, his way of looking at the koans must be considered arbitrary.

    This is not to deny that some koans do raise questions about the presuppositions of the use of referring terms. I1: 'Show me your original face before you were born' and I3: 'Don't call it a pitcher but tell me what [a pitcher] is' are examples. However, the questions they raise are not such as suggest that we "forego all ontological commitments to all possible semantic categories" (92). What they suggest is that we will have to forego singular terms for non-existents (I1) and bare particulars (I3). I will even suggest a bit later that the assumption that PON is always in force leads to a misinterpretation of some koans.

    I want now to confirm my rejection of Cheng's gloss on koans by suggesting another one which is intrinsically as plausible. Reverting to the jargon of Section I, I want to sketch an account alternative to




Cheng's of the assertion that the koan speech-act is used to make. What I will say is rather traditional, but I want to try to say it as carefully as possible.

    Human beings have a tendency to 'live in their heads'. This phrase covers several facts. First, men have a tendency to overtheorize. Some things are ruined by too much thinking on them, things which are essentially matters of experience. What is more, almost anything can be source of immediate experience, and so almost anything is vulnerable to ruination by too much theorizing. The second fact is this. Such theorizing usually presents itself phenomenologically as internal verbalization, and the internal verbalization often insinuates itself between ourselves and the thing experienced. This is how the thinking interrupts experience and how it leaves us with only our verbalizations. This leads to the third fact: when our theoretical internal verbalization is interposed between ourselves and external things, the object of our awareness becomes ourselves. It is we who are doing the theorizing, and to be aware of the theoretical verbalizing is to be aware of ourselves. This state of mind is undesirable, for it is a commonplace that our happiest moments come when we are not conscious of ourselves, and that most forms of consciousness of self are baneful. It is hard to say why this is so; perhaps the resources of a self are much more limited than the resources of the world, so only an object-directed consciousness can satisfy the human appetite for variety.

    The disadvantageousness of this state leaves us with a problem: how can a man with a propensity for injecting his theorizing between himself and the world be coaxed out of doing this? I would suggest that this is the problem the Zen master is addressing, and the koan is his answer. One technique is out; ironically, the very technique I've been using. It does no good to mount an argument about the disadvantages of living in one's head. This would be one more theory, one more verbal construction for the unenlightened to interpose between himself and the world. The activity has got to be halted, and what the Zen masters realized is that it can't be halted by arguing, however subtly and cogently, that it has got to be halted.

    The point of the koan, then, is to halt living in one's head by presenting inescapably candidate objects for immediate experience. The objects are presented in contexts normally reserved for verbal theorizing, since the abrupt shift of context makes them perspicuous. Thus, when the student




is lost in a cloud of metaphysics surrounding the One, the master turns his attention to a robe. He turns the student's attention: he doesn't say "Your attention would be better spent on a robe, for by seeking fulfillment in speculation you are like a dog chasing its tail in the hope of nourishment." This is an interesting argument, and the odds are the student would pursue it. The Master shows without saying the advantages of experience. He could in fact do this by adverting to a river or a fox; he could clout the student. Anything would do - that is what is insightful about Cheng's principle of ontic substitutability.

    It supports this view of koans that Professor Cheng himself sometimes hints at Zen's emphasis on immediate experience without developing the implications of his hints. He says in a footnote that the principle of 'contextual demonstration', closely allied to ontic substitutability, could also be called the principle of experiential reconstruction "as it is intended to indicate the fact that after ontological reduction reality will be experienced in whatever way it happens to be experienced" (102). This latter, I have argued, is nearly the central point of the koan. How "reality will be experienced in whatever way it happens to be experienced" follows upon ontic reduction is something Cheng does not tell us. I suspect the cited passage reflects Cheng's awareness that the 'principle of experiential reconstruction' has a much more central place in Zen Buddhism and the institution of the koan than he is in a position to allow, and he tries to make it follow from the principle he has construed as the point of the koan. But it will not follow, so far as I can see, and this suggests that Cheng has erred in his extraction of principles from the koan.

    My rather traditional analysis of koans does at least as good a job as Cheng's, I think, in explicating the koans Cheng cites, although justifying this claim is beyond the scope of this article. By way of illustration, however, the reader might reflect on III3: 'Who is the Buddha?' / 'Your name is Hui-chao'. If the reader is convinced that the point of koans is PON he will take the speech-act III3 as an occasion to rethink the reference of 'Buddha' and 'Hui-chao' (perhaps to conclude that he can construe either as referring to anything). But does this not, intuitively speaking, miss the radical inconsecutiveness of the answer? Is it not more plausible to suppose that the Zen master is (in Wittgenstein's words) assembling reminders to the purpose of getting the student to live his life?




    A final point in favor of my interpretation as against Cheng's, and a point of independent philosophical interest, is this. I would claim, in the face of a long scholarly tradition, that this account of the complicated and disadvantageous state of living in one's head is entirely independent of the metaphysical question of the existence of a self or substantial ego. It is, first of all, entirely possible for a substantial Cartesian self to achieve 'emptiness', to get out of his (its?) head. Eliot says that a hint of this is " heard so deeply/ That it is not heard at all, but you are the music/ While the music lasts." Contrariwise, it is entirely possible that a Humean bundle theory of the self (a theory with a famous echo in Mahaayaana Buddhism) is correct, and yet the bundle, so to speak, be tormented by self-awareness. This happens whenever the object of a conscious state in a bundle is some other conscious state of that same bundle - perhaps a conscious state of verbal theorizing. Indeed, it might even be possible for a transitory state of a Humean bundle to have itself as object.[6] Since, then, the value and attainability of egolessness is independent of the problem of the self, the position that egolessness is to be pursued is not hostage to metaphysics, and to that extent is more acceptable than a position like PON which is. Cheng's interpretation makes the koan rest on the surely debatable claim that "An ontological structure of no reference should be recognized for all semantical structures of language." Koans as I construe them make a point acceptable even to the metaphysically more circumspect.






*    I wish to thank Margarita Garcia for helpful criticisms.

1.    'On Zen (Ch'an) Language and Zen Paradoxes', Journal of Chinese Philosophy 1 (1973), 77-102.

2.    I henceforth abbreviate 'if and only if' by 'iff'.

3.    I shall not be concerned with specifying further the distinctions between sentence, utterance, proposition, assertion or the contextual influences on these items. These are general issues in philosophy of language, and nothing I say here depends on their resolution. If the reader accepts the example of the tee-totaling wife-beater (derived, I think, from Donnellan), he concedes what I need.

4.    This is an adaptation of Cheng's 114 and 116: it is the form of the koan given by Suzuki. Zen Buddhism [New York, 1956]: 134-5.

5.    Needless to say, I do not know if Quine would approve of his criterion for ontological commitment being pressed into the cause of sartori.

6.    I owe this point to Ted Kornfeld.