See the (?)fish swimming freely about—this is fish happiness.	You’re not a fish. How (whence) do you know fish happiness?
You’re not me; how (whence) do you know I don’t know fish happiness?	I’m not you so I certainly don’t know you. You’re certainly not fish and that’s enough to say you don’t know fish happiness.
Let’s go back to the beginning. When you said "how (whence) do you know fish happiness,” it was asking me already knowing I knew it. I knew it above the river.

Puzzles and Challenges

One might wonder why this simple exchange gets so much play and so little close analysis.[1] Few treat it as central to understanding Zhuangzi’s philosophy.[2] We don’t suspect that the conversation actually took place.[3] We don’t even know if Zhuangzi and Hui Shi were actually friends.[4] It’s not clear that anyone understands the point of the exchange.[5] Standard commentary consistently treats Zhuangzi as not seriously participating in the debate. His posture is described as "playful" or "dismissive" of Hui Shi’s logic, as being smugly assured of his mystical knowledge, or even "sarcastic."  Rather than crucially informing our view of Zhuangzi, we usually read the passage to "confirm" "what we already know" about both Zhuangzi and his fellow discussant, Hui Shi—the mystic v the logician.[6]

We can explain the fascination with the passage from a different direction, however. Commentators often fondly recite this story of Zhuangzi’s lament for Hui Shi:

Chuang-tzu, among the mourners in a funeral procession, was passing by the grave of Hui Shih. He turned round and said to his attendants:

"There was a man of Ying who, when he got a smear of plaster no thicker than a fly's wing on the tip of his nose, would make Carpenter Shih slice it off. Carpenter Shih would raise the wind whirling his hatchet, wait for the moment, and slice it; every speck of the plaster would be gone without hurt to the nose, while the man of Ying stood there perfectly composed.

Lord Yüan of Sung heard about it, summoned Carpenter Shih and said "Let me see you do it." "As for my side of the act," said Carpenter Shih, "I did use to be able to slice it off. However, my partner has been dead for a long time."

Since the Master died, I have had no one to use as a partner, no one with whom to talk about things." (Zhuangzi, chapter 24 tr. Angus Graham)

We understand the lament as the loss of a philosophical partnership, of two like minded but disagreeing intellectual companions engaged in the joys of productive philosophical argument. The lament paints a poignant picture of mutual respect, constructive engagement and what must have been a lot of common purpose and shared attitudes– albeit with a substantive pattern of disagreement. The disagreement should have been "fertile" for Zhuangzi’s intellectual growth. Zhuangzi literally could only reach the heights of his craft in shared conversation with Hui Shi. This picture fuels the Graham/Harbsmeier lament that we know too little about the writings and philosophy of Zhuangzi’s great and prolific friend (and perhaps teacher).[7]

The fish story portrays this constructive debating relation. Arguably, it’s the main one with an unequivocally "philosophical" content and style[8]—and a kind of philosophy even Westerners can recognize at first glimpse—epistemology![9]

Angus Graham proffers a related explanation of the distinctive value of the passage:

This is the only instance of disputation with Hui Shih in the Outer chapters (interest in the Sophists was already failing), and is remarkable for a playfulness which in parodying logical debate is more faithful to the detail of its structure than anything else in Chuang-tzu. When Hui Shih defended the paradoxes listed without explanation in `Below in the Empire' … he must have been talking like this, as the Later Mohists do in the explanations of some of their Canons; what a pity we never hear what he had to say!

Graham’s depiction here contains the main elements of the manifest image. The passage ridicules logic but also exhibits it in great detail. It is precisely the detail of this philosophical discussion that makes the passage difficult to interpret, however. Graham postulated a crucial clue when he suggested that the rather rare use of an^where in the question "From whence do you know . . . ?" helps make sense of Zhuangzi’s final comment.  It also links Zhuangzi’s position in the dialogue to the perspectivalism that is one of the philosophical themes of the Zhuangzi.  It’s a particularly strong theme in chapter 17, "Autumn Floods," where we find this dialogue. 

I accept and follow this Graham line.[10] That the conclusion Zhuangzi draws is a "positional" one is clearly implied by his concluding statement. Translations before Graham had seen this point, but few had linked it as closely to the language used as Graham’s analysis did.  His hypothesis about an^whence distributes the perspectival point throughout the dialogue.[11] I won’t rely heavily on this lexical claim but focus on the rest of the structure of the argument—which, I shall argue, fills out the perspectival analysis.

Graham’s insight gives us a line on Zhuangzi’s concluding remark, but doesn’t really solve the other philosophical puzzles posed by the dialogue.[12] How is the final remark a resolution or a plausible conclusion to the disagreement? What perspectival point is Zhuangzi’s argument supposed to have made? Aren’t there internal contradictions in each discussant’s position? Isn’t Zhuangzi’s penultimate riposte a non-sequitur? Is Hui Shi right that Zhuangzi is forced by his own logic to acknowledge he doesn’t know fish happiness? And so forth.  While accepting Graham’s insight, I venture some answers to these remaining questions and attempt to do it in a way that illuminates the distinctive form of philosophical perspectivalism in epistemology exemplified in the Zhuangzi text.

While I do make Zhuangzi’s point in this debate consistent with the perspectivalism of this chapter and with a theme of relativism/perspectivalism that runs through much of the rest ot the Zhuangzi, I do not accept as a condition on the adequacy of an interpretation, that a passage must be made consistent with an entire book.[13] It surely is possible to imagine that the correct interpretation of the passage would expose or illustrate an incoherence or contradiction in another Zhuangzi position.[14] Despite the frequent appeal to whole text coherence as a standard of good interpretation, the coherence relevant to issues of meaning is the coherence of the entire community’s discourse.

Coherence of a line of thought is not a bad thing, of course. While it is not semantically required, coherence with some larger theme in the Zhuangzi provides an answer to an explanatory question of textual origin or motivation—what is the reason for including this dialogue here in the text? There are other ways of answering the question besides expository coherence—political commentary, changing interests, animosity toward persons or groups, accident etc.  However, an interpretation that make the point of this exchange a way of clarifying or bolstering some position taken elsewhere in the text can give us a straightforward account of more of the detail of the passage.[15] As a way of explaining text inclusion, broad coherence has advantages over external explanations—e.g., the decline in interest in logical matters, political developments, defection, etc. External causes can partially explain some "tenor" of a passage, but the interpreter still tends to rely on coherence to explain inferentially articulated detail.[16] The more detail we consider, the harder it is to explain without appealing to inferential patterns among sentences.

Ironically, part of the historical difficulty in understanding this exchange may lie precisely in tying it too closely to perceived author coherence—coherence with the "manifest image" of the two speakers. The problem is not merely Zhuangzi’s image as a mystic, but includes Hui Shi’s converse image as a logician. Neither image, I will argue, helps us explain the detail of this exchange.

Rather than explaining the dispute, the manifest image tends to explain it away.  It becomes a familiar miscommunication between a logician and a mystic. Pre-identification of philosophical position prevents us from construing this dialogue as a productive philosophical discussion between cooperating intellects—the kind of discussion that would underlie Zhuangzi’s lament about Hui Shi’s passing.

I propose, instead, to flip the standard analysis on its head to explain the deep structure of the dialogue as manifest in its fine detail. Zhuangzi is the more skillful dialectician leading the intuitionist (Hui Shi) in a logical trap.  The trap is a close relative of one that catches Hui Shi other places and the discussion reveals a common pattern of agreement and disagreement between these two ancient thinkers—they agree in their perspectival relativism and disagree on how to formulate its implications.

In the Zhuangzi as a whole, I believe Hui Shi thematically plays the role of sharing the relativist point of view (maybe even originating it), but of misstating the consequences of that shared position. He frequently lands in contradictions which Zhuangzi manages to avoid. Zhuangzi, the more logically consistent of the two, exposes these contradictions in their disputes and shows how more careful reasoning from their shared position should go. Arguably, this picture of their relationship better explains Zhuangzi’s lament. He honed his insights into their position by these dialectic exchanges revealing Hui Shi’s mistakes.  The dialogues help him (Zhuangzi) think more carefully about what are the valid conclusions of his relativism.

Graham’s used his locative analysis of an^whence to explain mainly Zhuangzi’s closing remark—but rather than developing the perspectival theme in detail, he combined that insight with the common view that Zhuangzi is playfully dismissing "logic."  Looking for the alternate picture that is of the debate as a reductio of one of Hui Shi’s illegitimate inferences from relativism should more thoroughly explain the role of this locative way of asking "how do you know?" 

The mystic v. logician theme doesn’t make clear sense of the fond picture we have of fertile philosophical discussion between the two. If Zhuangzi tended to ignore, make fun of and dismissively laugh off Hui Shi's logically sound arguments, in what sense could he have gained anything or needed stimulation from Hui Shi? The argumentative exchanges represented by that common picture would have been philosophically barren. The problem with the orthodox analysis of this particular discussion is that someone looking at the dialogue knowing nothing about the two participants would have a hard time telling which of the two was more committed to picky, analytic debating points.

If the first key to unraveling the exchange is to escape from the grip of the mystic v logician image, the second is to note that Zhuangzi shifts the debate by semantic ascent to the epistemic principle to which Hui Shi is committed. The issue is not simply "does Zhuangzi know?" but to "what is the appropriate standard of attributing knowledge?" We can view the discussion as an example of a productive philosophical exchange by tracing how Zhuangzi’s questions lead Hui Shi to a reductio of his assumption that only one perspective (the subject’s) counts as real knowing (e.g., such things as pleasure and knowledge). Zhuangzi’s rival view is that multiple standards for attributing knowledge (even of such things) are appropriate for different contexts and different kinds of knowledge claims.

What I do here, accordingly, is focus on steps in the discussion to explain the detail in ways that make the exchange "productive."  The picture that will emerge is a quite different one of a Zhuangzi who out-thinks Hui Shi in his own terms—a Zhuangzi who, far from haughtily dismissing reasoning, accepts and meets a reasoned challenge by inferring a better result than that posed by his opponent. We have a passage that shows us how a Zhuangzi could possibly have honed and perfected his philosophical position from a real life exchange of the sort abbreviated here.

This result, like the mystic-logician alternative, does fit the dialogue into a coherent (but alternative) picture of the philosophical relation between Zhuangzi and Hui Shi.  They are fellow relativists who disagree about how correctly to formulate the consequences of their relativism.  However, the result does not prove, against modern textual skeptics, that the two actually were friends or that the relationship pictured in the stories was a historically accurate one. It can be made into a more coherent exchange and the advantages of that, again, are not that such is a condition of adequacy, but that it accounts for more detail explaining inclusion of this dialogue in the collection. The same explanation of detail could equally explain the "invention" of the dialogue (if that’s what it was). The invented detail is still motivated by inferential relations of the responses of the two personae—whether real or fictional.

From our own purposes in trying to understand Chinese thinking, this passage is one of a small cluster of examples of reasoning in ancient Chinese texts that Sinologists recognize as having a surface resemblance to Western philosophy more than to the manifest image of Chinese thought. As usual, the greater value of such examples lies in careful analysis of them to identify and elaborate the deep conceptual differences behind the surface similarity. Difference usually stands out more starkly against the background of the familiar. The traditionalist identifies "Western style reasoning" to justify skipping over it. From the point of view of any serious student of Chinese thought, these are precisely the cases that deserve the most careful analysis and scrutiny.

The Analysis

See the Fish swimming at leisure

Zhuangzi starts with a simple inference[17] ostensibly the beginning of a conversation.[18] Translators worry about the correct way to translate the modifier Zhuangzi uses here of the fish.[19] We will not dwell on the issue here.  He describes the fish using the phrase chu ^comeout you^swim:roam cong^follow:from rong^easy (Come out and swim freely and leisurely) then says: "ci^this yu^fish zhi^'s le^pleasure."  (This is fish-happiness.) The inferred consequent has two crucial ambiguities.

1)      In classical Chinese, knowledge and other propositional contexts are usually rendered as complex noun phrases—using a possessive between subject and predicate.  (See the discussion below in [How do you know?]) Translators can choose to render Zhuangzi’s conclusion in propositional form: "[the] fish are happy." As we will see, the ambiguous formulation may motivate the disagreement.  Hui Shi’s skeptical response reads Zhuangzi’s statement "this is fish happiness" as entailing cognitive access to something like Nagel’s [20] "what it is like for a fish to be happy." His skepticism is more clearly in order on that reading. In the alternate reading, it is a more familiar form of  "other minds" skepticism—you can’t know that fish are happy (even if they were to laugh out loud).

2)      The character le^{pleasure:happy} need not be read in the strongly subjective way that Western philosophers normally assign to our concept of ‘pleasure’ or ‘happiness’. As Fingarette urged years ago [21], Chinese psychological terms seldom have such exclusively subjectivist meanings. They evoke inner states by describing human embodiment in situations—as we so often see in Chinese poetry. [22] In other words, le could imply only that the fish are engaged in a pleasant activity—leisure or playful behavior. That insight makes the inferential gap between Zhuangzi’s antecedent evidence and his conclusion seem much less precarious. Those fish swim at their leisure/play/enjoyment (as opposed to searching for food). We could analyze Zhuangzi’s position adverbially" See the way of swimming—that is leisurely swimming for a fish. They are taking a stroll—as we are.

How do you know?

Hui Shi’s first "move" in the "Happy Fish dialogue" illustrates an interesting parallel between ancient Chinese and modern Western language games.[23] When we assert P, someone may respond with "how do you know P."  While it may seem too obvious to comment on, the reliance on this norm of assertion will figure crucially later in the dialogue.  Briefly, the "way" of assertion is one that is efficient for a community relying on transmission of information. Those who hear our assertions take them on our authority and may appropriately transmit them to others.  We, in effect, implicitly license the hearer to repeat our claims to third parties.  The hearer, accordingly, has a right (perhaps even a duty) to assure herself that you know what you assert—that the license to repeat it is valid. So the conversational norm governing assertion is that you should be prepared to respond to a challenge to your right to claim knowledge of what you have said.[24]

When Zhuangzi asserts "That's fish happiness" the rules of ancient Chinese like those of modern English allow Hui Shi's challenge: "how do you know that?" Zhuangzi implicitly acknowledges the appropriateness of the challenge by responding (and implicitly relying on the principle himself). His response answers a question with a question, which some regard as impolite, but is a common practice among philosophers. We’ll discuss this further when we return to Zhuangzi’s response below.

A second interesting parallel is that the challenge normally invites one to give your way of accessing the knowledge—"how do you know" (as in most translations). The form isn’t merely "do you know?" A hearer asks you to give your reasons so she can decide whether or not to acknowledge your claim to know what you assert. The interesting departure in this dialogue is that the natural and more familiar he^how yi^with zhi^know is replaced here by what Graham identified as the locative question form "an^whence zhi^know?" (whence do you know?). The form of the challenge is perspectival. Implicitly, this form invites Zhuangzi to give a perspectival account of his point of view or route of access to the knowledge.

Hui Shi does not simply rely on the norm allowing a "how do you know" challenge. He also motivates it by an inference from "zi^you/master fei ^is-not yu^fish" (you are not (the/a) fish).  We can fit Graham’s hypothesis of the perspectival context of the discussion to this challenge. The Zhuangzi famously uses the perspective of different species and their different standards of beauty, comfort, as an illustration of the complexity of the notion of different points of view.

Hui Shi, as a rule, takes relativism to entail something about a correct perspective. In this case, that for judging if a thing is in a state the correct perspective is the thing’s own.  If a thing takes something to be beautiful, then it is beautiful. Zhuangzi does not know the happiness from the fish’s perspective and, accordingly, should make no assertion about it. 

Hui Shi’s skeptical challenge, thus, implicitly commits him to their being a single correct standard for claiming knowledge of fish happiness—the self or subjective standard. That's the standard appropriate for both interpreting Zhuangzi’s comment and for endorsing that Zhuangzi knows what he asserts. That the debate from this point shifts to discussion of Hui Shi’s proposed standard is a crucial key to this interpretation.

Notice that Zhuangzi already preceded his claim with an inference. His implicit answer to "how do you know" had already been given. Zhuangzi explicitly inferred it from their manner of swimming (under his interpretation of it). He didn’t originally phrase this in a perspectival way. It was a simple inference from an (interpreted) observation.

Hui Shi’s position that knowledge of such things can only be the direct, non-inferential knowledge entails we should reject Zhuangzi's implicit standard. Only an immediate, intuitive apprehension of the happiness itself counts as 'knowing', not an inference from some observation. Thus his question not only poses a standard, but implicitly rejects the one already in play--the inferential, indirect third-person standard Zhuangzi explicitly relied on in his opening comment.

Hui Shi’s challenge, like the original question, exploits the structural ambiguity of propositional contexts in ancient Chinese mentioned above. Common assertions may exhibit the ambiguity, but knowledge claims almost always do. Knowledge contexts are rarely explicitly propositional—as they would be in English. [25]  In Indo-European languages, we would structure the challenge using the familiar grammatical pair: propositional belief v. knowledge. What we know (or merely believe) is normally expressed as a that-phrase. We know or believe that P—where P can be replaced by a whole grammatical sentence.

Chinese zhi^know typically takes a noun-phrase object. The noun phrase is formed by nominalizing the S-P sentence that occurs in the that-clause of Western belief contexts: The "thing" known is S’s P.  What we know is a thing’s property.  I have argued in previous publications, [26] that this feature (together with the non-sentential features of counterparts of a belief-structure in Chinese) explains a tendency, now common among interpreters of Chinese philosophy, to treat knowledge claims on the model of knowing of or knowing how rather than knowing that. Hui Shi's implicit standard, thus, exploits the nominalized reading we noticed above. Again, it makes his skepticism more intuitively plausible. The correct standard for an assertion may depend on the form of the claim. The ambiguity here is a function of the ambiguity in the form of Hui Shi's challenge.

How do you know I don’t know?

Simply to respond to the challenge, Zhuangzi would cite some means of access to what he (allegedly) knows. Here he doesn't.  (As we saw, he had already stated his ground for the assertion.)  Instead, he turns the discussion to the first-person standard presupposed in Hui Shi's challenge and its implied rejection of Zhuangzi’s third person standard. Zhuangzi’s response is not merely the impolite evasion of answering a question with another question. It’s a counter-attack.  Hui Shi committed himself to a privileged status for the first-person standard.  Zhuangzi interprets this as the non-relativist claim that correctly to claim to know X's inner state, you must be X.

To assume Zhuangzi here responds directly to the challenge is to read it as a defense of a dogmatic or anti-skeptical position.  Thus commentators read it (as Beckwith complained above fn. 5) as simply saying "I know what I know" or as Chinn (1997) does as anti-skeptical. If it’s merely that, then Zhuangzi’s answer is also evasive and impolite—both conversationally wasteful responses. Zhuangzi’s response can be viewed as both principled and appropriate and extremely efficient conversationally if we acknowledge the ascent to a philosophical issue.

Zhuangzi's response illustrates the "semantic ascent" typical of philosophy.  He focuses on a "second level" or meta-claim. Zhuangzi does not reject the assumption that he should be accountable for warranting his claim to know what he asserts. Technically, that wasn’t at issue in Hui Shi’s challenge since Zhuangzi had prefaced his claim with his reason.  The question was, 'whence' – from what perspective—does he know it? Zhuangzi can fairly turn his attention to the interpretation Hui Shi places on his assertion in insisting on the first person perspective. Hui Shi has read "this [is] fish happiness" as "I am in direct contact with fish happiness—the kind of contact a fish would have." For that reading, the presupposed standard "you must be fish" is warranted.

Zhuangzi’s pluralist position should commit him to their being multiple interpretations of his original assertion. Each would be governed by different standards of knowing. Hui Shi's insisting only his "inner" or "subjective" standard is right implies only the matching interpretation of "this is fish happiness." Zhuangzi now leads Hui Shi into a reductio of his view that the "inner" is the only correct standard. The reductio here lacks the volubility of Plato’s Socratic performances, so we must tease the steps out of the inferential details of the dialogue.

More ambiguities lurk in the Ancient Chinese grammar that will be over-specified in almost any English translation. First, classical Chinese common and proper nouns work alike grammatically in this example. I argued years ago[27] that the Chinese dialecticians were theoretically treating all nouns as logically singular terms. Thus, Zhuangzi’s shift from ‘fish’ to ‘me’ is not as jarring as the parallel inference would be in English. What Hui Shi presupposes is not classic "other-minds" skepticism. Technically, his implied standard is "being of the same species" not "being the very same individual subjectivity."  Zhuangzi here extrapolates that standard, but he needn’t be doing it on the basis of a subjective analysis of [aspects of] consciousness. In effect, he is simply taking the grammatically ambiguous standard, "one must be X to know X’s F"--where F can be replaced by either le^happiness or zhi^knowledge. In replacing yu^fish with wo^I in the X position, Zhuangzi is replaces one singular term with another in the formula schemata.

Second, Zhuangzi’s extrapolating the standard from knowing le^happiness to knowing zhi^knowledge would also be grating on its face if delivered in an English dialogue. Even if we take both to be internal states, [28] we might still object. One is cognitive and the other affective.  We might think it easier to know other rational people’s cognitive states than their affective states.

This cognitive-affective distinction was not as marked in ancient Chinese, witness the common translation of xin as ‘heart-mind’. Zhi^know seems to have a wide range of uses which, according to definitions in the Mohist Canon, includes something like ‘consciousness"[29] (e.g., Canon: A24 When you sleep, your zhi^{know (consciousness)} does not zhi^know).  Further, if we take Hui Shi to be skeptical about whether Zhuangzi knows "what it is like to be a happy fish" then it is arguable that the state of knowledge in question consists in duplicating the inner state in one’s own inner field. This should make Zhuangzi's extrapolation of the principle from le^happiness to zhi^know somewhat less objectionable. If that's what knowing is, then knowing one's knowing must involve the same difficulty. Ancient Chinese thinkers seldom seem to have operated with an implicit "true belief plus an account" concept of zhi^know. 

Interestingly, however, Zhuangzi’s counter-question at this point is a relatively rare example of a fully propositional object of zhi^know. (An^whence zhi^know wo^I bu^not zhi^know yu^fish zhi^'s le^play:leisure?) [30]  Presumably, Zhuangzi could have put his point in nominalized form, though it would have been a bit of fun saying it (an^whence zhi^know wo^I zhi^'s bu^not zhi^know yu^fish zhi^'s le^play:leisure?).  The change, however, could be significant.  An explicit propositional form exploits the ambiguity of the "fish [are] happy" in a way that favors Zhuangzi’s claim to knowledge. The question looks more like an ordinary other-minds question, can we know that X (Zhuangzi or the fish) is/are in state Y (happy or knowing).

We might object, on Hui Shi’s behalf, that he has not claimed to know that Zhuangzi doesn’t know. He’s only asked Zhuangzi to justify his cognitive "right" to the assertion.  Hui Shi might have responded, "I don’t know if you know or not—that’s precisely what I was asking you." To take that position, Hui Shi would implicitly abandon his exclusive reliance on the subjective standard for attributing knowledge. He could be taken innocently and openly to have asked what other standard is involved.

In his next response, however, Hui Shi acknowledges the exclusivist reading of his challenge as implying that Zhuangzi doesn’t know.[31] If we accept this interpretation of the challenge, then the norm of assertion we discussed at the beginning comes back into play. If he’s committed to "Zhuangzi doesn’t know. . .", he too should be prepared to respond to a challenge, "how/whence do you know that Zhuangzi doesn’t know?"  If the only standard is his subjective standard, then Hui Shi should withdraw his implicit claim on the same grounds that he argues Zhuangzi should withdraw his original comment about the fish.

Hui Shi ignores another way he might avoid the trap Zhuangzi is setting for him. He might insist on the interpretation of his standard as species based, not as an individual or subjectivist standard. This would undermine the parallel Zhuangzi draws between the two knowledge commitments.  He opts not to draw either of the distinctions that could extricate him from Zhuangzi’s reductio.

Right! Not being you, I don’t know you; you, not being fish don’t know fish—that’s the whole of it!.

Hui Shi’s next move takes Zhuangzi’s bait. He clings to his subjective standard while he accepts:

(a)    That in claiming something, one should know it.

(b)   That his subjective principle should be applied to different individuals as well as to species,

(c)    That the standard applies to zhi^knowledge as it does to affective attitudes like le^pleasure and

(d)   That his challenge amounts to claiming that Zhuangzi does not know fish happiness.

Hui Shi then draws two dangerous conclusions.  "I don’t know your knowledge state" and "you don’t know the fish’s le^pleasure state."  But the second contradicts the first when we combine it with (a) the norm of assertion. Hui Shi is committed to both that he knows and that he doesn’t know what Zhuangzi knows. 

In drawing the first conclusion, Hui Shi could have used his subjective standard in two ways: a direct and an inferential way. Being Hui Shi, he knows directly that he doesn’t know what Zhuangzi knows. However, Hui Shi chose to formulate the conclusion inferentially.  He re-committed himself to his standard and inferred from it that he doesn’t know Zhuangzi’s knowing. What he does conflicts with his commitment to the inner-perspective principle. If his principle is correct, he should not apply it. He has implicitly committed himself to two different standards—knowing (what it is like first person) and knowing by inference. Both commitments, still, lead to his conclusion that he doesn’t know Zhuangzi’s "knowing."

Then he uses his standard again in an inference to "Zhuangzi doesn’t know the fish are happy."  This time, however, his commitment and what he does with it lead to different results.  He can’t claim to know what he asserts (that Zhuangzi doesn't know fish happiness) by using his inner-perspective method.  The trap is closing.  Not noticing his own predicament, he declares his case "complete." His triumphant tone of confidence attaches to the proposition known by violating his own principle.  It is knowing from the outside, at a distance—not the subjective principle he champions. If he knows it—as the triumphal refrain signals, then his principle must not be the only way of knowing.

Zhuangzi is right—Hui Shi can’t consistently express his objection to Zhuangzi’s knowing the fish while relying on his subjective standard for attributing knowledge. If he tries to express his skepticism, Hui Shi ends up committed both to knowing Zhuangzi’s cognitive state and to not knowing it. The conflicting commitments come in different ways from his principle of privileging the inner perspective—but following the way of knowing he favors, he must conclude he doesn't know if Zhuangzi knows fish happiness or not. If he accepts that the subjective principle is one of a plurality of correct tests of knowing, then he can escape the contradiction. However, that would allow Zhuangzi to return to the inference stated at the beginning of the dialogue. 

Hui Shi’s in a bind. He needs inference from the principle to justify his claim that Zhuangzi doesn’t know, but his principle rules out knowledge by inference. His only escape is to say he doesn't know if Zhuangzi knows or not—and to withdraw his assertion. He may continue to believe that Zhuangzi does not know what he asserted, but Hui Shi, relying on the principles he does, should not comment on Zhuangzi's assertion. If Zhuangzi were committed to Hui Shi's principles, he would have asserted nothing as well. One consequence of Hui Shi's "way" of knowing is the familiar anti-language conclusion. Neither should have spoken.

Thus, this dialogue does not depict a rigorously logical Hui Shi trapping Zhuangzi, who escapes by mischievously refusing to acknowledge logic. It rather shows a Hui Shi trapped by his persistent tendency of slipping from relativistic premises to absolutist conclusions that conflict with his relativism. He starts by being perspectival about joy, but slips into making an absolute claim about the standard of knowing it.

Let’s return to the beginning

The account dominated by the manifest image of Zhuangzi (humorous mystic) and Hui Shi (logician) would, at this point, treat Zhuangzi as cavalierly dismissing Hui Shi’s thoroughly logical proof, changing the subject, and just verbally goofing around to display his distain for logic. We no longer have to treat it this way. Hui Shi’s original challenge allowed Zhuangzi to invoke philosophical ascent. Zhuangzi refocused the issue away from the question of whether Zhuangzi knows and addressed instead to Hui Shi's consistency in using his subjective principle to question Zhuangzi's right to the assertion. Does that principle allow Hui Shi to dispute a claim to knowledge?  He ends up trumpeting a conclusive claim to know that violates the principle. He must now acknowledge that we have other ways of knowing besides "from the inside."

This denouement is notoriously recondite. One tempting confusion stems from the common assumption that Zhuangzi is denying that Hui Shi knows what he knows. On the contrary, Zhuangzi position is epistemic pluralism; a multiplicity of standards govern judgments of what counts as knowing in different situations. He can and should allow that Hui Shi does know what Zhuangzi knows.

This confusion may tempt us to misread Zhuangzi when he recommends following the discussion back to its start. Then it becomes easy to suspect, as I thought years ago, that Zhuangzi is about to resort to a silly and inconsistent sophistry. Ordinarily, in asking "how do you X" we presuppose that you have done or can do X.  "How do you make the yo-yo sleep?"  Then the question seeks information about what method you used in (admittedly) actually doing X. The form of the 'how' question ordinarily implies one acknowledges the ability itself. Is Zhuangzi simply relying on this verbal trick?  In asking "‘how do you know?’ you presupposed that I knew."[32]

I was tempted to read Zhuangzi in this way, in part, because I was still caught up in the familiar view that Zhuangzi was trapped by Hui Shi’s logic. That motivation largely disappears when we see that it’s Hui Shi who is trapped. Zhuangzi has lured him into a reductio of his absolutist epistemic theory. Even then, the reading hardly comports with the rest of the standard view of a Zhuangzi, dodging logic from his playful, mystical perspective. On this account, he evades it by using a verbal trick, a paradigm of philosophical sophistry. It’s a sophistry because, in a challenge to knowledge, the phrase "how do you know" normally has the opposite presupposition.[33]   It yields a Zhuangzi behaving like a bad analytic thinker—not a mystic.

I now have three connected worries about reading Zhuangzi's penultimate sentence in this way. 

·        First, it is a weak sophistry – an almost textbook case of the Wittgenstein's analytic philosopher's disease. We get caught in a puzzle because we impose a rigid interpretation on a grammatical form ("How do you. . . ?") then carry it from its "home" use to a different "language game" (knowledge challenges) where it plays a different role.

·        Second, Zhuangzi's own prior response clearly commits him to the opposite (and proper) reading of Hui Shi’s challenge. His own challenge clearly interprets Hui Shi’s prior question as entailing that he (Zhuangzi) does not know. Zhuangzi thus acknowledges that Hui Shi's "whence" implied doubt, not endorsement of Zhuangzi’s knowledge claim. The line of interpretation makes Zhuangzi dishonest as well as sophistic. 

·        Third, it is difficult to see how to fit this verbal trick into the flow of the dialogue. It neither picks up on the reductio of Hui Shi’s use of his subjectivity principle nor leads effectively to the puzzling conclusion: "I knew it from above the Hao [river]." The verbal trickery neither illuminates nor develops the perspectival thrust of the discussion.

To develop an alternative way of understanding this passage, let’s ourselves go back to some basics. First, we should focus on the implicit pragmatic theory of language in traditional Daoist texts—the core role of language is normative guidance, not factual representation. This is one key to seeing how Zhuangzi’s theory fits with his version of perspectival skepticism. Commitments to different norms of term use (description) give rise to different directions of guidance.  Second, if we regard Hui Shi's Ten relativist "theses" as an inspiration to Zhuangzi, then we need to identify how he thinks he improves on Hui Shi’s position. My thesis is that Zhuangzi avoids incoherent derivation of absolutist (monist) conclusions. Hui Shi, like careless relativists everywhere, keeps being tempted to draw absolute conclusions from his relativist premises.

The first item helps make Zhuangzi seem refreshingly "modern." Naturalist pragmatism now portrays epistemology and semantics as being deeply normative and modern developments in these fields are borrowing analyses developed in ethics, aesthetics and political theory. In Zhuangzi’s terms, we can think of daos of speaking, of interpreting, of inferring, and asserting there are of acting and painting. It naturally becomes obvious that, as in ethics, the standards for these normative epistemic judgments are "relative" to the judging situations, in which different ends, values, purposes, roles and so forth—in a word, with different contextual daos. The standard of what counts as 'knowing' in responding to a Cartesian or twin-earth skeptic is different from the standard appropriate for more ordinary contexts of assertion. The standard of recognizing barns in "barn façade county" is not the same as it is in Montana.

This is a way to understand why the paradigm case arguments don’t successfully defeat a Cartesian skeptic. While it is true that the "normal" standards for the use of ‘know’ do not require certainty, when a skeptic points out a somewhat fantastic possibility of error, simply citing the "normal" use doesn’t effectively respond to the context of his argument. It can merely show why, in normal situations, we don’t have to use ‘know’ in the skeptic’s way. The skeptic can escape by recommending a reform of normal usage. At that point, decisive rebuttal would have to show that his recommendation is incoherent in some way.

Consider a related situation within Sinology. Professor Ames and I both routinely deal with beginning students or interested non-Sinologists who ask us what an inscription or passage means. We look at the passage and simply "read it." We treat meaning in that context as "obvious" to someone who has learned Chinese. Our answers to these beginning students sound like observation sentences. We simply look at the passage and "see "the meaning.  Given our training and the context of the question, neither of us needs any implicit inference from the evidence to answer the question.

Where we stand at this moment, however, the situation is different. We are at a Sinology conference, surrounded by other "experts" in Classical Chinese, debating about the meaning of some familiar passage like the one before us.  In this context, for either of us to react to the disagreement before us by something that sounds like a simple observations sentence of the type we used with our freshman students would be "out of place"!  No one would take such "observations" seriously as warranted in this context.  Neither of us is here entitled to claim knowledge in that direct, perceptual way. The situation imposes a different standard on both of us.  We have essentially similar training and mastery of Chinese and we both "know what the other knows." Since we disagree, the question of which of us knows the meaning must now rely on a standard more appropriate to the context of the question. Merely claiming to "see the meaning" would be semantically "inappropriate." This is a twist on Hume’s contrast of his views when dines, plays backgammon or converses with friends, he is not plagued by the doubts that come when he does philosophy.[34]

Zhuangzi's intriguing form of skepticism makes a similar point. He does it explicitly in the "Gaptooth" discussion about "knowing that you do not know."  There his skeptical mood starts with doubts that we can know something simply because everyone agrees to it. Clearly, this philosophical doubt implies a higher standard of knowledge than we use for "ordinary" knowledge claims. Normally (i.e., outside of philosophical discussions of skepticism) we accept the "authority" of others and of our own past beliefs. However, as soon as we formulate this normal attitude as an explicit rule, we would reject it. It, in fact, creates a contradiction, because one of the things everyone "knows" is: what everyone "knows" is often wrong.[35] As the skeptic, Wang Ni, continues to reject formulations of skepticism, and ends with something like the Pyrrhonian conclusion that we don't know if we know. But Wang Ni ultimately bases that conclusion on doubts about correct word use. "How do I know that what I call knowing is not ignorance? How do I know that what I call ignorance is not knowing?"[36]

Now let’s return to Zhuangzi's troubling final remark for another look.  We need to construe it slightly differently to make the discussion follow a coherent line.  Hui Shi has been caught in a reductio because he tries to restrict knowledge to the norm appropriate for a first person assertion—the (subjectivity) standard. Zhuangzi would presumably allow that there are two ways of knowing about affective states. The way I can know I am happy is different from the way you can know it. Hui Shi wants to "privilege" one of these ways of calling something "knowing."  Only the inner should be so classified. Yet, he can’t avoid relying on a third person standard (using inference rather than direct awareness) if he wants to conclude use his privileged standard to draw any conclusions about knowing—in particular to conclude that Zhuangzi does not know fish-happiness.

We can now view Zhuangzi’s final comment as a helpful explanation to Hui Shi of his perplexity. Hui Shi's enthusiastic confidence in his argument shows that he does treat his conclusion about Zhuangzi's not knowing as firm, solid knowledge. So he uses and trusts the inferential, outside perspective, third-person standard. Zhuangzi has consistently relied on the same standard—and does so now in acknowledging that Hui Shi does know what he, Zhuangzi, knows. Zhuangzi knows the fish are happy and Hui Shi knows what Zhuangzi’s knows—what his state of knowledge is and how/whence he got it. Allowing a third person perspective removes any puzzle. Hui Shi knew at the beginning and his challenge was formulated precisely because he did know how Zhuangzi knew, on what he relied. He knew Zhuangzi wasn’t claiming to know the "what it was like to be a happy fish."  His problem came because he adopted a commitment to an unrealistic standard of knowing. His challenge was motivated by that standard and knowing that Zhuangzi knew in a way that did not follow the inner-perspective way of knowing. 

Hui Shi knew that the ground of Zhuangzi’s assertion was the observed manner of swimming. Hui Shi, standing with Zhuangzi "above the Hao" saw exactly the same thing and knew precisely in which sense, how and whence, Zhuangzi knew about the fish.

In laying out this discussion in terms of standards for the use of ‘know,’ I have occasionally helped myself to concepts, distinctions and tools of analysis to draw our attention of readers to relevant features of the argument. Some of these are terms that I have argued were not available to Zhuangzi and Hui Shi. Where we use sententials, 'truth' and 'inference' 'subjective' and 'objective', ancient Chinese thinkers used different notions, built around things like 'terms', 'assertability' 'way' and 'dependency'. So their focus would not be on inference or logic, but on what approving or disapproving of the use of a term in a context depends or relies on. They would speak of a way rather than a norm, of outer v. inner perspective rather than subjective-objective, and so on.  We should be able to go back and retrace all the steps in the reasoning in terms familiar in the Zhuangzi without losing the point.  I have scattered such rephrasing through my account, but not attempted to replace all of them. I leave that as an exercise to the interested reader.

The component of this orientation to discussion of theory of language (arguably contributed by Hui Shi) is the insight that what is appropriate to say depends on a perspective—which Zhuangzi broadened to include purposes (useful and useless), indexicals, and background assumptions or standards.  So it is perfectly in character, for this discussion to be shaped around the idea of what it is appropriate to say from various "points of view." Thus Graham’s insight about the "whence" points to a crucial aspect of the character of philosophical discussion in ancient China.

Perhaps as important as the revised image of a careful and principled philosopher Zhuangzi is the linked revision in our picture of Hui Shi—his foil. For this exchange at least, Hui Shi’s contribution shows no hint of sophistry.  The closest anyone comes to a pure sophistry is the rejected interpretation of the final passage from Zhuangzi. Nor is there a trace of logic in Hui Shi’s contribution. He is doing epistemology, but logical statements, per se, plays no role in his contribution. Logic figures when it helps us notice that he is caught in a reductio.  Nor do we find Hui Shi making or insisting on fine distinctions. On the contrary, he ignores crucial distinctions that might have saved him from Zhuangzi’s trap. Hui Shi’s standard is the direct standard characteristic of intuitionism as opposed to Zhuangzi’s reliance on inference from evidence at a distance. 

Conclusion

I have been experimenting with an alternative set of assumptions about the philosophical persona of Zhuangzi and Hui Shi in interpreting this widely cited dialogue. How much can we conclude from the enriched picture of this beloved story? Does it settle outstanding controversies about Zhuangzi? Abandoning the traditional view of Zhuangzi’s intellectual heritage and character has independent motivation.  A recent series of archeological discoveries has spawned some turmoil in textual theory of the Laozi. These discoveries have reinforced the scholarly doubt of the popular image of Laozi and Zhuangzi as prophet-and-follower. As recently as 1979, it was intellectually respectable to picture the Zhuangzi’s role , as Stuart Hackett did, as that of providing the reasoning, elaboration, exposition and defense of Laozi’s aphoristic insights.(Hackett 1979:56-57) We still find that explanation of Zhuangzi in "capsule" encyclopedia presentations, but scholarship has been steadily chipping away at the traditional picture.

The alternate story, however, also has deep roots in the traditions of interpretation, starting with Guoxiang’s "individualistic" reading of the Zhuangzi and its contrasts with Wang Bi’s take on the Laozi and found in Wang Xianqian’s hypothesis of a linguistic basis of Zhuangzi’s thinking. As early as 1948, Fung Yu-lan focused more on the influence of Hui Shi as of Laozi—though he still classifies Laozi and Zhuangzi as the 2^nd and 3^rd "phases" of Daoism (Fung 1948 ch. 6, 9-10). Graham gave the new picture enormous momentum with his early insights into the "Qiwulun," his discovery of technical theory of language insights scattered throughout the Zhuangzi, and his lovely aphorism "Zhuangzi never knew he was a Taoist." However, the image of Zhuangzi as a Laozi "development" persists as a background to interpretation despite the scholarly undermining of the story that originally grounded it.

This tour through the "fish happiness" passage adds to these historical and textual reasons for recasting our view of Zhuangzi.  It adduces an interpretive advantage of the rival picture--its explanatory power. It illustrates the interpretive advantages of substituting a rival picture: of a truly philosophical Zhuangzi who outdoes Hui Shi at a distinctively Chinese style of philosophical dialectic from a shared perspective of relativism in the philosophy of language. If we focus on the detailed inferential structure of the discussion, we can illuminate fine details that show Zhuangzi’s argument to be principled and responsive. The "playful mystic" picture, by contrast, motivates a policy of dismissing inferential detail on the hypothesis that Zhuangzi is not merely ignoring Hui Shi’s inferences but poking fun at them. We found instead a line of interpretation that portrays a Zhuangzi who takes his rival’s arguments and positions on board conscientiously, shows how they lead to incoherence, and offers a more sound way to develop their shared perspectival insight. We exploit the picture of a Zhuangzi who "sharpens his skill" with another dialectician—a Zhuangzi who draws consistent conclusions from the perspectival analysis.  Hui Shi, by contrast, shares the relativist starting point but mishandles its implications.

The form of Hui Shi’s weakness relative to Zhuangzi is consistent through several philosophical issues.  Hui Shi appears to draw metaphysical or absolute conclusions from relativist premises. An obvious form of the issue was first isolated (perhaps accidentally) by Graham[37], who argued that Hui Shi’s theses led to logic discrediting itself.  If everything is relative, then there are no real distinctions.[38] The "all is one" mystic did, after all, draw the correct conclusion from relativism. The statement of Hui Shi’s theses in Ch. 33 "the social world" indeed includes this conclusion—"The world is one body." —as Graham also noticed.  Graham saw that Zhuangzi rejects this conclusion in Ch. 2 "Qiwulun (Equalizing Things Discourse)." However, he did not put these insights together to revise his traditional view of Hui Shi as a logician and Zhuangzi as anti-logic. I would maintain that it is Hui Shi who is not simply a mystic but, as illustrated here, an intuitionist. Hui Shi was never a logician nor much of a sophist. He was simply a less consistent relativist than was Zhuangzi.

Careful reflection on this dialectical relation between the two thinkers should also help avoid common errors in expressing Zhuangzi’s relativism—in confusing it with Hui Shi’s. If an interpreter makes Hui Shi’s mistake in thinking about relativism, he or she would misconstrue Zhuangzi. For an example, consider Hui Shi’s conclusion that we should "love all things equally." Standard interpretations would attribute something very close to Zhuangzi—the most common translation of his "Qiwulun" (Chapter two) is "discussion on making things equal." 

We should not take Zhuangzi to be committed to the claim that all views are (equally) correct or wrong. From Zhuangzi’s relativist philosophy, no judgment about absolute value follows. The Zhuangzi neither takes its relativism to entail that all views are equally good nor that they are equally bad.  "Error theory" – the conclusion that all views (or all "ordinary" views) are wrong is like "all views are right" or "all views are equal" in purporting to be absolute judgments, made from the absolute perspective of nowhere.

The Zhuangzi seems rather to deny that we can make sense of these purported judgments. The cosmos does not make judgments. The point of view of the cosmos (or any absolutist point of view—including the perfect man) is so irrelevant as to be unintelligible to us as we seek ways in this world.

Zhuangzi himself carefully stays within the relativist perspective—balancing the skeptical awareness that his is a point of view with his remaining within it. He consistently resists Hui Shi’s invitations to state the "view from nowhere" on the world of perspectives.  From his actual perspective, Zhuangzi makes all kinds of judgments. Things are not equal.

From my (relativist) point of view, for example, it’s clear to me that the Christian (or Muslim or Hindu or Buddhist) fundamentalist is wrong—I might even say crazy, irrational and stupid!  Nothing in relativism requires me to stop making these judgments from within my perspective, nor from defending them and living by them. It only reminds me that there are other perspectives and that that, had I grown up in another or and remained in my own fundamentalist religious tradition, I would now see things quite differently. Perspectivalism, pluralism or relativism fosters only a mild skepticism inducing greater tolerance, but not the conclusion that I should abandon or refuse to express my natural, contextual judgments.

A relativist recognizes that hers is a point of view and that there are others. She is not, however, rationally committed to the normal expression of her position as "just a point of view." Her perspective has no work for the ‘just’ to do.  The "just" is not a part of the relativist insight—but belongs to the absolutist’s restatement of it.  The crux of the issue between Zhuangzi and Hui Shi is about the consistent way to state their relativism.

What about other matters of textual history? This alternate image does little to dispel the skeptical hypothesis about Zhuangzi and Hui Shi. Was there really any relationship between these two alleged philosophical friends? The detailed examination hardly suggests that the passage is a verbatim recording of an actual conversation. Even if there were an actual discussion behind this section, the exchange is extremely unlikely to have been so brief and with the points so efficiently condensed. While it could be a précis of a more lengthy discussion that actually took place between two philosophers, it’s certainly still plausible as a "staged" dialogue fictionally developing the supposed philosophical relationship. We can read the passage in the same way while assuming it contributes to the Zhuangzi persona rather than recording a history of the person.

The passage, accordingly, can do little to confirm the alternate story of "Zhuangzi’s lineage."  However, we needn’t surrender wholesale to historical skepticism.  Like all skeptical arguments, skepticism of the Zhuangzi - Hui Shi friendship makes assumptions and presupposes intellectual motivations. One expressed motivation for the doubt is the dominant image of the philosophical postures of Zhuangzi and Hui Shi. That image, as we saw, makes a productive philosophical friendship and exchanges unlikely. Consequently, a more plausible account that reveals a productive pattern of philosophical discussion would undermine that motivation. Zhuangzi and Hui Shi can be seen as working in the same "school" of relativistic analysis—albeit with different ways of formulating their conclusions.

With that motivation removed, the skepticism may start to look slightly "metaphysical".[39]  Given the commonplace observation that we know almost nothing about Zhuangzi personally beyond what we learn from the text, it seems that we can phrase the limit of skepticism this way: if we know anything about Zhuangzi, we access that knowledge via the channel that indicates he had a friendship with Hui Shi.[40] The stories of their exchanges do not suggest anything physically impossible and they reflect the real-life phases a relationship might go through (death of a spouse, professional separation and jealousy, death of the friend, etc.)

A final point concerns whether the passage undermines or supports the view of Zhuangzi as a skeptic.[41]  On the surface, Hui Shi raises a classic skeptical worry and Zhuangzi dismisses it. However, we have seen that the issue is not simply Zhuangzi’s knowing or not knowing, it is an issue of the standard of knowing (and the pragmatic or perspectival reasons for different standards in different contexts).  The happy fish dialogue turns out to be a gloss on the Zhuangzi Ch. 2 skeptical position:

How do I know that what I call ‘knowledge’ is not ignorance and what I call ‘ignorance’ is not knowing?

Endnotes