Part II, Chapter 7: Judgment of Attachment: Functional Clinging and Ultimate Truth¶

The preface offered one judgment of attachment for each of three traditions—Madhyamaka, Yogācāra, and tathāgatagarbha. For Madhyamaka, it was attachment to intellectual activity itself. The preceding six chapters of Part II addressed suffering, origin, cessation, the path, and verification. This chapter treats the last and most concealed point at the functional level: how a method devoted to breaking attachments cultivates new grasping within itself. It explains the mechanism behind that judgment in the preface.

Ultimate truth is a logical metaconstraint, not a level of truth

The starting point for understanding this chapter's criticism is to locate “ultimate truth” (paramārtha-satya) accurately. A common misreading of the two-truths framework within the Madhyamaka tradition takes ultimate truth to be a “higher level of reality”: a domain of existence beyond everyday experience, in which things exist in their “true way”—empty, without intrinsic nature. This is a Vedānta-style reading. It turns Nāgārjuna's framework into an ontological dualism: an ordinary everyday world, conventional truth, and a more real metaphysical world, ultimate truth.

Nāgārjuna's texts do not support this reading. In the Vigrahavyāvartanī he explicitly says he has no thesis (pratijñā); reductio is his entire operation. If ultimate truth were a richly furnished level of truth, Nāgārjuna would have positive propositions requiring active defense, in direct conflict with his own account of his method. The precise function of ultimate truth in his usage is a logical metaconstraint on the form of propositions. It prohibits propositional forms that attribute intrinsic nature to anything: the mode of assertion “X has a fixed intrinsic nature” is invalid. It says “Any proposition attributing intrinsic nature can be led into internal contradiction,” not “There is a world called the ultimate, in which things exist in an empty way.” This reading later received its definitive formulation in Volume 1, Part II, Essay 18, more precise than the three-character Chinese term “metaconstraint” (元约束). Chapter 6 of that essay treats the two truths as levels of assertion, rather than two worlds. Chapter 7 then fixes ultimate truth's final form as an invariant across all frameworks: a truth that cannot be evaded in any framework capable of individuating, asserting, and reasoning, rather than a truth seen from nowhere. Conventional truth is truth within a framework. The same essay distinguishes three objects of exclusion—states of affairs, frameworks, and claims—to answer the reflexive question “Does an invariant amount to saying nothing?” Appendix II of this volume already uses the language of invariants; the formulation here is brought into line with it.

This understanding gives the two-truths framework a different structure. Conventional truth is the contextual level where practice actually occurs: conditioned dharmas, causality, stages of practice, suffering, and its cessation all hold here. Ultimate truth is a critical metalevel. It supplies no positive content, only constraints on the validity of propositional forms. Their relation is that of “object language” to “metalanguage,” not “a lower world” to “a higher world.”¹

The gap left by “emptiness too is empty”

Nāgārjuna expressly warns against grasping emptiness as another real existent. Chapter 13 of the Mūlamadhyamakakārikā says: “The Great Sage taught the dharma of emptiness to free us from all views; those who further see emptiness as existent cannot be taught by the buddhas.”² Emptiness is medicine prescribed to free us from views. If the medicine itself is grasped as a view, no medicine can cure us. The later expression “emptiness too is empty” formalizes precisely this meaning of “the emptiness of emptiness” (śūnyatā-śūnyatā), one of the eighteen emptinesses. This step is internally necessary. If emptiness itself had intrinsic nature, it would become an eternal, unchanging metaphysical entity, structurally identical to the entities posited by the proponents of intrinsic nature whom Nāgārjuna refutes, differing only in name. “Emptiness too is empty” closes the back route to reifying emptiness and, in principle, also prevents grasping it as a position to be defended.

Yet the step “emptiness too is empty” strips the conclusion of metaphysical significance without constraining the argumentative method that produces it. Here lies a subtle but important gap: the output of the metaconstraint—the conclusion of emptiness—has been stripped of metaphysics, but the argumentative process that produces the metaconstraint itself—reductio, analysis of propositional forms, and the epistemological tools called “Madhyamaka right view”—has received no corresponding constraint.

The practical result is that practitioners able to use arguments for emptiness can succeed in “not grasping emptiness,” avoiding grasping it as a metaphysical entity, yet unknowingly grasp “that which can produce emptiness”: the ability to use reductio, sensitivity in recognizing propositions about intrinsic nature, the framework of “Madhyamaka right view” itself, and even the pursuit of direct realization of emptiness. Grasping these differs entirely in its manifestations from grasping emptiness, so “emptiness too is empty” does not cover it directly.

This is functional clinging. Its objects are the tools, methods, and frameworks that produce the conclusion, rather than a conclusion asserted as true. Someone can sincerely claim “I do not grasp the doctrinal content of emptiness” while deeply grasping “the ability to use this doctrinal framework” and “the way of reasoning within it.” The latter is a more concealed grasping, harder to notice in oneself precisely because it operates under the banner of “breaking grasping.”

The mechanism of functional clinging

The mechanism by which functional clinging forms is no mystery. Any complex skill requiring long training to acquire will create in its possessor a deep identification with the skill itself. Go players who have practiced for years identify with being Go players, not just with a few brilliant moves. The arrival of artificial intelligence has made this especially clear. When Go programs repeatedly defeated top professional players, the players experienced far more than “losing a few games.” The position of “the strongest,” toward which a lifetime of training had been directed, was suddenly canceled. Some openly said that subsequent effort had lost its former meaning, even giving this as a reason to leave professional Go.³ Not a fraction of their skill had been taken away; what was shaken was the identity attached to it. Painting presents the same case. Contemporary artists' most common criticism of AI-generated painting is that “AI paintings have no soul.” This sentence guards a way of being more than it describes paintings. Artists find meaning in painting and a place for their identity in mastery of the craft; the word “soul” guards that place. Programmers' public disputes about so-called vibe coding—programming primarily through AI generation—display the same structure openly as a sequence of discourse. First, this volume's self-imposed boundary must be stated: the following diagnoses the structure of discourse, not anyone's psychology (Introduction, Chapter 1). This passage was written in 2026. Its examples will age; the structure will not. Discourse observable on forums and social platforms broadly follows a fixed sequence. It begins with defending identity and skill, equally thoroughly at both ends. One end makes a commitment to purely manual programming; the other, afraid of falling behind, pursues every AI technique one by one. The object defended has merely shifted from old skills to new ones. When this defense is frustrated, discourse turns into a list of charges against AI: hallucinations, bugs, and architectural defects are repeatedly cataloged, followed by an overall refusal to acknowledge its technical progress, even outright rejection. The charges then spill into mockery of others, such as attacks on a nonexistent category of “liberal-arts students' vibe coding.” Every step in this sequence has public texts that can be consulted, and none requires the speaker to hold any mistaken “philosophical view.”

The skill-based content of Madhyamaka practice—using reductio, mastering the distinction between the two truths, and pursuing direct realization of emptiness—likewise produces this deep identification during training. In the context of the theory of the path, this identification is precisely a grasping. It fixes “I am someone who can analyze problems using Madhyamaka right view” as the core of self-understanding, and that fixation is itself a variant of self-view (ātma-dṛṣṭi).

Can Nāgārjuna's system recognize this problem? In principle, yes: the “I” is empty too, and identification as a “Madhyamaka practitioner” has no intrinsic nature. Reductio can break this grasping as well. The difficulty is precisely that applying the instrument of refutation to itself is extremely hard in practice. Using reductio to break grasping at reductio requires using it without treating it as an entity that can be grasped. Psychologically, this is a reflexive operation almost impossible to sustain stably.

One insight about the activity's place must be made explicit: the act of reductio itself is a “formation.” Views are products of formations. An act of reductio, a discrimination, or a pursuit of direct realization is a mental fabrication, within the aggregate of formations. Recall what Madhyamaka refutes: the results of formations, the views that have solidified. But the fabricating activity that produces views sits precisely at a boundary its method may well be unable to break. A tool can negate every conclusion it produces, but cannot negate the act of using it while it is being used. This has much the same structure as the findings of the two earlier chapters. The chapter on the truth of suffering found suffering relocated from formations into view; the chapter on the truth of origin found the diagnosis at the reflective level. Here we find the same matter manifested in the method itself. Formations come first, and suffering is already there. Striking at the results of formations—views—can at most change the direction in which formations flow. Formations continue to fabricate, and suffering may remain. Functional clinging is hardest to notice in oneself precisely because it resides in no proposition open to reductio, but in the act of reductio itself.

Nāgārjuna's analogy of illusory persons and its gap

Nāgārjuna was aware of this tension. Verse 23 of the Vigrahavyāvartanī uses the analogy of one illusory person hindering another to answer a sharp objection: “Your own speech is empty too; how, then, can it effectively refute other claims?” His answer is that one illusory person can hinder another: both being illusory does not prevent the hindering from actually taking place. Likewise, empty speech can refute empty claims; speech need not possess intrinsic nature to be effective. Nāgārjuna actually gives two examples here, not one: one artificial being hinders another, and an illusory person hinders another brought about by his own illusionistic power. The latter adds a layer: what refutes and what is refuted come from the same source. This layer directly addresses the question below of what entitles one illusion to judge another.⁴

The analogy serves a purpose: it shows that causal relations between empty things actually occur, at the conventional level, without presupposing a foundation possessing intrinsic nature. But it leaves a question that Nāgārjuna did not adequately answer. When illusory person one knocks down illusory person two, we need some standard for saying “the striking force of illusory person one is, in some sense, greater than the defensive force of illusory person two.” Otherwise, “knocking down” is merely a phrase, not an actual relation of difference. Nāgārjuna has not explained why his kind of “illusion”—Madhyamaka reductio—can serve as the standard for judging another kind of “illusion”—propositions about intrinsic nature.

If “all speech is empty,” the validity of Madhyamaka reductio is empty too. If Madhyamaka reductio has a special methodological validity—it can expose internal contradictions in the opponent's presuppositions, whereas those presuppositions cannot expose internal contradictions in the Madhyamaka method—then this special validity is itself a position that must be accounted for, a position Nāgārjuna says he does not have.

This fissure must be recalibrated in light of Volume 1, Part II, Essay 18. After recalibration, it narrows and changes location. First, the dilemma's second horn must be scaled back. It tacitly assumes that only what really exists can function, whereas functioning has always proceeded unhindered at both the level of quantification and the level of designation: armies without intrinsic nature still win battles. The Vigrahavyāvartanī's reply exposed this assumption on the spot, and the assumption itself falls within the theorem's target. As for the “position that must be accounted for,” Essay 18 has already accounted for it as an inspectable list: definitions, logic, and one relational premise, with no fourth category. No item on the list is a thesis asserting that something really has intrinsic nature. Accounting for it therefore does not violate no thesis, because what no thesis lacks is a thesis at the object level, not tools at the metalevel. The fissure accordingly no longer lies in “What entitles this kind of illusion to judge another?” It moves to the list's last item: is the relational premise true solely by virtue of word meanings? Essay 18 leaves a downgrading clause and three exits for this question; it has not sealed them off.

Candrakīrti's refutation of the Svātantrikas and his treatment of “conventional valid cognition alone” are the most important response to this fissure. His strategy assigns methodological validity entirely to conventional valid cognition at the conventional level, so that no privilege at the ultimate level is needed. Yet, as discussed in Part I, Chapter 3 and Part II, Chapter 5, this response bears its own internal tension: at the conventional level, it is not obvious how conventional valid cognition can distinguish effective use of Madhyamaka reductio from effective use of arguments for intrinsic nature.

Traditional warnings and their blind spot

The tradition warned of this danger, and the location of its warnings precisely marks its blind spot. Nāgārjuna himself warns in the “Examination of the Four Noble Truths” that those who contemplate emptiness poorly are harmed by it, like those who handle a venomous snake or a spell poorly.⁵ Commenting on Chapter 13 of the Mūlamadhyamakakārikā, Candrakīrti carries the warning to its limit with two famous similes from the Kāśyapa chapter of the Ratnakūṭa Sūtra. Emptiness is a purgative: if the medicine is taken but not expelled, the disease worsens. Thus, “Better to arouse self-view heaped as high as Sumeru than to arouse overweening conceit through a view of emptiness.” These warnings show that the tradition knew the danger of grasping emptiness well and used its strongest rhetoric against it. But examine their coverage. Both the snake and medicine similes concern the danger of grasping emptiness as a view, as a conclusion—the very level covered by “emptiness too is empty.” For grasping at the level of tools and abilities—clinging to techniques for catching snakes, the skill of compounding medicine, or even the identity of “one who contemplates emptiness well”—the warning system of the classical treatises has no corresponding entry. On the contrary, “skillfulness” remains a term of unqualified praise in the treatise tradition. This book itself later supplies an entry. Volume 1, Part II, Essay 18, Chapter 9 judges the theorem to be a raft, whose entire efficacy lies in crossing. It is not itself furniture on the far shore; those who walk with the raft on their heads are referred to the clause on levels. Chapter 10 of that essay also contains a first-person record identifying the object grasped as precisely the technique for canceling views. The added entry does not change this chapter's diagnosis, only corrects its tense. The point is not that no one thought of it, but that the classical tradition had no such entry and this book has supplied one. The functional clinging identified here grows precisely in this area beyond the warnings' coverage.

The warning canopy includes two phrases omitted from the English image: the snake simile concerns grasping a venomous snake unskillfully (MMK 24.11), and the medicine simile says, “The medicine remains in the abdomen, and the illness grows worse.” The chapter and its note use the version in which medicine has been taken but is not expelled; they separately identify a contrary retelling in which it is not taken. The figure follows the former. “Identity” in the English lower heading gathers up the identity as a skillful observer of emptiness already listed in the Chinese panel.

Conclusion: the gap's philosophical significance

Together, functional clinging and the gap in the analogy of illusory persons reveal a deep structural problem in Madhyamaka methodology: any philosophical system that seeks to remove objects of grasping completely will itself produce grasping at its method in the process of using that philosophical method, and this grasping is the part the system finds hardest to dissolve from within. All deconstructive philosophical traditions share this problem; it is not unique to Madhyamaka. But for a tradition that locates the core mechanism of liberation in “breaking grasping,” the gap carries particular soteriological weight.

Madhyamaka's two mechanisms of self-correction—“emptiness too is empty” and the analogy of illusory persons—both touch this problem, but neither resolves it fully at the methodological level. Recognizing this is both a fair assessment of Madhyamaka's philosophical achievement—it is more clearly aware of the problem than any other tradition—and an examination of its soteriological project on its actual merits.

We return to the preface's judgment: Madhyamaka's attachment is attachment to intellectual activity itself. This chapter provides the mechanism of that attachment. After “emptiness too is empty” closes off grasping at conclusions, grasping shifts to tools and abilities. This level operates precisely in the name of “breaking grasping” and is hardest to notice in oneself. This judgment at the functional level will be combined with the preceding chapters' conclusions for an overall assessment in the next chapter.

References¶

The passage in Kumārajīva's translation is at T30, no. 1564, p. 18c16–20. For the *Dazhidu lun*'s explanation of the eighteen emptinesses, see T25, no. 1509, pp. 287c25–288a11; the definition of “emptiness of emptiness” is at p. 288a2–10.