Sambandha as a ‘Śakti-of-Śaktis’: Bhartṛhari’s Influence on the Relational Realism of Pratyabhijñā

Abstract

Contemporary scholarship has significantly advanced our understanding of the grammarian Bhartṛhari’s influence on the Pratyabhijñā Śaivism of Utpaladeva and Abhinavagupta. One area that has been somewhat neglected, however, is the subject of relation (sambandha). Here, I examine the influence of Bhartṛhari’s sambandha-vāda on the Pratyabhijñā school. As I see it, Bhartṛhari’s understanding of the holistic movement of sphoṭa—the practical process of ‘encoding’ and ‘decoding’ linguistic information—leads to a necessary reevaluation of the general logical form of sambandha, i.e., ‘relationality-as-such.’ On this account, Bhartṛhari articulates a basically transcendental conception of sambandha as a ‘śakti-of-śaktis’ in his ‘Exposition of Relation’ (Sambandhasamuddeśa [SSam]). This effectively means that one cannot designate the general logical form of sambandha in linguistic terms without also thereby changing its essential nature as such (cf. Houben: 170–4). I maintain that Utpaladeva’s ‘Proof of Relation’ (Sambandhasiddhi [SS]) leverages this insight into a series of pragmatic arguments to demonstrate that vimarśa, or recognitive judgment, is the true locus of relational action—i.e., unity-in-diversity (bhedābheda). In doing so, he effectively salvages a coherent understanding of relation as necessarily real (satya) from the deconstructive agenda of the Buddhist eliminativist, even though the referent may indeed appear paradoxical from the perspective of theoretical reason alone.

1. Introduction

Bhartṛhari’s influence on the Pratyabhijñā thinkers Utpaladeva and Abhinavagupta is now well-established.1 Many of the grammarian’s central concepts are appropriated and developed in their own absolute idealist system of philosophy. Among these are the intrinsic self-consciousness of each cognition (svasaṃvedana) (which entails that any single cognition is unable to be the object of another cognition) and his keystone doctrine that all knowledge, including perceptual cognition, is fundamentally linguistic.2 While there has been great progress with respect to these important topics, the subject of sambandha has received relatively little attention from scholars of Pratyabhijñā.3 To be sure, this is partly because the Sambandhasiddhi [SS], Utpaladeva’s main treatise on relations, had not been fully translated until very recently.4 Now, we can better appreciate the ways in which Bhartṛhari’s influence also extends to the Pratyabhijñā theory of sambandha, the central topic of Utpaladeva’s SS, and a recurrent theme in his primary work, the Īśvarapratyabhijñākārikā [ĪPK].5 Although not explicitly mentioned in Utpala’s SS (or, for that matter, in the relevant portions of the ĪPK), I will show that the Pratyabhijñā thinkers basically defend and refine Bhartṛhari’s analysis of relation in the third portion of his Vākyapadīya [VP], the Sambandhasamuddeśa [SSam].6

The structure of this paper is as follows: Since I contend that Bhartṛhari’s theory of sambandha results from his monistic metaphysics of divine speech (para-vāc), in the second section, I analyze Bhartṛhari’s theory of sphoṭa, which depicts a progressive, triadic transformation of the holistic essence of undifferentiated vāc into determinate sequences of particular word-sounds (dhvani-śabda). In the third section, I show that this ‘vertical’ realization of practical speech-acts motivates and informs Bhartṛhari’s novel interpretation of the general form of sambandha as a śakti-of-śaktis—i.e., a transcendental capacity of speech that unifies a diversity in the abstract designation of determinate linguistic signs but cannot be expressed or designated in those very terms. Before discussing Utpaladeva’s appropriation of this idea in the SS, in the fourth section, I briefly outline the Dharmakīrtian position as presented by Utpala—to wit, that ‘relations’ do not exist, because nothing ultimately real can instantiate a dyadic form that is simultaneously unified and diverse. On my reading, Utpala emphasizes that the nominalist axioms of Dharmakīrti’s ontology necessarily entail that all forms of ‘relation’ are no different from universal ‘conceptual constructions’ (kalpanā)—that is, a false, conventional superimposition (āropa) of distributed/continuous (anvaya) properties onto the ‘ultimate’ reality of momentary, non-relational particulars. The fifth and sixth sections turn to the Pratyabhijñā position that leverages Bhartṛhari’s coextension of relation with ‘a subtle form of śakti’ (cf. MacCracken 2017) to overcome the eliminativism of Buddhist nominalism. Utpala first argues that relation cannot be a form of conceptual ‘error,’ because we presuppose a form of relation in the practical judgment thereof. He then offers a set of pragmatic arguments that leverage the coextension of sambandha and śakti to advance his own Śaivite form of absolute idealism: namely, the true locus of relational action—viz., the śakti that manifests (prakāśa) all determinate appearances—is nothing other than the general unity-in-diversity of recognitive judgment (vimarśa) itself.7

2. Sphoṭa: Bhartṛhari and the Metaphysics of Language

To understand Bhartṛhari’s conception of sambandha and its impact on these later Śaiva idealists, we must first briefly look at his theory of meaning, or sphoṭa. For although the Pratyabhijñā tradition does not embrace sphoṭa semantics explicitly (and Utpala does not seem to elaborate another alternative in the extant works) the Pratyabhijñā tradition does adopt Bhartṛhari’s theory that vāk, or divine speech, manifests through progressive phases of realization.8 This, I claim, generates a certain theory of the general form of sambandha. Yet, in Bhartṛhari’s system, these two ideas go hand in hand—it is impossible to talk about the gradations of vāk without at least briefly touching upon the nature of sphoṭa. Both doctrines, in turn, stand against the backdrop of Bhartṛhari’s linguistic monism; everything that exists is constituted by, and proceeds from, a single infinite reality, what he calls ‘śabda-brahman’ (literally ‘Brahman [in the form of] word/sound’).9 This śabda-brahman represents the absolute and unified ground of all worldly phenomena and knowledge.10 However, quite unlike the ‘idealism’ of, say, Advaita Vedānta or Yogācāra Buddhism (insofar as we may describe them as such), where the determinate and differentiated conceptual trappings of language reflect the quintessence of saṃsāric ignorance, Bhartṛhari believed that salvation comes with the realization that the linguistic nature of thought is identical with the transcendent activity that generates and sustains the conventional world (Matilal 1990, p. 95). Thus, the embodied capacity to express reality linguistically; the conceptual structures of language itself; and the subjective apprehension thereof; all are nothing other than the primordial reality of divine speech (para-vāc).

Sphoṭa is a concept that Bhartṛhari develops from the Vedic grammarians to designate the holistic process of this divine language.11 While it literally means ‘bursting,’ ‘expansion’ or ‘disclosure,’ for the ancient grammarians, it comes to represent the inner principle of all semantic phenomena, which is manifest in the fact that diverse externalized verbalizations (śabda) convey a universal unit of meaning or reference (artha).12 We might call it something like a holistic intuition of pure semantic unity that both generates, and yet is also expressed by, sequential and distinct word-sounds (śabda).13 Patañjali defines it as ‘the eternal and imperceptible element of sounds and words and the real vehicle of the idea which bursts or flashes on the mind when a sound is uttered.’14 Sphoṭa is therefore the part-less, non-sequential, indivisible ‘semanticality’ of any particular speech-act.15 In this respect, words and sentences both have sphoṭa, but the sphoṭa of sentences is not reducible to the sphoṭa of words.16

It is therefore important to note that sphoṭa is not merely the universal referent (artha) of a sentence or word (since both are technically sphoṭa) but the holistic ‘bursting forth’ of a unified idea that occurs in the mind of a particular speaker and hearer, for the phenomenon of language means the disclosure of an innate capacity to recognize that verbalized sounds (śabda) are related to a meaning (artha). Bhartṛhari writes that ‘the wise discern two [aspects] in the expression of words: one [i.e., sphoṭa] is the cause of the word [and] the other [dhvani] is used to convey meaning.’17 The term dhvani here refers to what we might call the tokenized expressions of language. They are the spoken and heard ‘articulated sounds’ that practically manifest the concealed and implicate unity of sphoṭa.18 As Matilal puts it, ‘Language and meaning are not two separate realities such that one conveys the other. They are in essence the two sides of the same coin. The sphoṭa is the unitary principle where the symbol and what is signified are one’ (95). And while Matilal is technically correct, his talk of strict identity of sign and signified occludes, I think, the triadic dimension of sphoṭa. That is to say, sphoṭa actually entails that (i) the vibrations of physical sound, (ii) the inner activity of thinking, and (iii) the determinate appearance of linguistic signs, are all internally related concepts.19

To explain the gradual unfolding of this continuous structure of sphoṭa into the diversity of dhvani, Bhartṛhari postulated three stages of speech: (i) ‘articulate speech’ (vaikharī vāc), or the external, public, tokenized utterances of individual speakers; (ii) ‘intermediate speech’ (madhyamā vāc), which reflects a pre-verbalized form of mental representation;20 and (iii) ‘seeing speech’ (paśyanti vāc), the innermost realm of indivisible semantic unity, which is non-sequential (akramā), undivided (abheda) and concealed (pratisaṃhṛta).21 Although paśyanti vāc exhibits no determinate content, it serves as a universal plenum for all semantic potentials. Accordingly, every self or mind consists of an infinite storehouse of sphoṭa-potentials, the universal types of semantic knowledge withdrawn from tokenized linguistic expression. This is the primordial, eternal realm of language, where universal possibilities are ‘seen’ (‘paśya’) or conceptualized in the mind without being explicitly verbalized. When a speaker has the practical desire to communicate, they are aware of the corresponding vākya-sphoṭa and initiate a formal procedure for its encoding into vaikharī vāc. Thereupon, at the level of public discourse, another person hears the individual pada-sphoṭas and these trigger a decoding process that results in the hearer’s holistic recognition of the very same vākya-sphoṭa within the mind of the speaker (Coward 1997, p. 77). Cognitive activity therefore resembles a spontaneous procedure of encoding and decoding the holistic essence of universal language into determinate and particular speech-acts (Seneviratne 2015, p. 140).22

Importantly for our purposes, the inclusion of an intermediate level of speech processing (madhyamā vāc) suggests that the movement from the indeterminate unity of paśyanti vāc into the determinate expressions of vaikharī vāc is a graded realization. In other words, the linguistic activity that manifests sphoṭa cannot be an instantaneous transition. To the extent that we view ‘intentionality’ as broadly coextensive with discursive activity, this suggests that the capacity to utilize and understand the sequential terms of public language rests on a deeper awareness of their progressive manifestation. The general three-fold procedure for ‘encoding’ and ‘decoding’ sphoṭa thus chiefly reflects stages in a phenomenological process whereby determinate linguistic signs emerge from an inner awareness of deeper, more indeterminate forms of semantic potential. The theory of sphoṭa therefore already implies that the practice of ‘languaging’ requires a certain intentional form of continuity, one that the Pratyabhijñā will come to identify with the unity of vimarśa (‘reflexive consciousness’) and prakāśa (‘manifestation’).

3. Sambandha as a ‘Śakti-of-Śaktis’

As should be clear, Bhartṛhari’s three-fold progression of linguistic self-disclosure suggests a certain conception of the way relationality functions in an intentional/linguistic context. In particular, the ideas in the VP and the SSam prefigure the important distinction fleshed out in Utpala’s SS between the relational tokens of conventional speech-acts (dhvani-śabdha) and the universal type (i.e., ‘relationality-as-such’) that characterizes a holistic sphoṭa’s active transformation into a particular sequence of sounds.23 This means that the way in which relational forms appear in language (‘horizontal relations’) must be logically distinguished from the general form that characterizes the intentional process of ‘languaging’ itself (‘vertical relation’). In other words, sambandha comes to denote a transcendental principle, one whose activity (śakti) informs the ‘syntactical’ realization (prakāśa) of sphoṭa at an intentional level of description.24

Bhartṛhari comes to this theory of sambandha-qua-śakti in the SSam through investigating the question: ‘how can ‘relation’ be expressed in words?’ (Houben: 98) In the opening two verses of the SSam, he proclaims that there is a ‘well-established’ relation between words and the object of understanding (i.e., the vācyavācakabhāva).25 In verse 3, he proceeds to pinpoint the genitive relation as the means by which we know that a referent (artha) is signified by a word (śabda) (e.g., as expressed in the propositions ‘this is the signifier of this, and this is the signified of this’).26 Since we cannot practically deny that we have knowledge of the designation of a thing while also using and understanding the signification function (viz., semiotics) of language, Bhartṛhari suggests that some general relational capacity is presupposed in the intentional and linguistic structure of all qualified cognitions.27

In the next two verses (4–5), however, he identifies a perennial problem that ensues when we attempt to identify sambandha itself as a substantial term through an ascription of some genitive or predicative properties. In brief, we cannot exhaustively characterize ‘relation’ in linguistic terms because it will always be presupposed in the capacity that enables speakers to identify things through the conceptual ascription of linguistic properties. Bhartṛhari interprets this to mean that the ‘extremely dependent’ form of sambandha cannot be designated through words:

There is no word that signifies the relation according to its specific property. Because it is extremely dependent (atyanta-paratantra), its form cannot be pointed out. Where this [relation] is, because some service is rendered [from one thing to another, or: from signifier to signified and vice versa], there one arrives at a property (viz. dependence) [but not at the relation itself]. It is even a capacity (i.e., something dependent) of capacities [which are themselves dependent upon the entity which possesses the capacity]; it is even a quality (i.e., something dependent) of qualities [which are themselves dependent upon the entity which possesses the quality] [so it is extremely dependent].28

Bhartṛhari describes relation, in its general form, as ‘extremely dependent,’ in the sense that it cannot itself be signified in words. When we cognize some limitative ‘service rendered’ (upakāra) between phenomena, we ‘infer’ (anugamyate) that a property such as ‘dependence’ has occurred.29 But this inferential ‘property’ is always necessarily one step removed from the śakti of relation in and of itself, because the capacity to designate this relational property must already be in place for the designation itself to occur.30 Thus, while a relation incontrovertibly exists between signifier and signified, no terms can designate or characterize what the specific property of this relation is. Houben summarizes: ‘Therefore, it is never possible to isolate the relation itself from the things that are connected through the relation, and, consequently, it is not possible to attribute specific attributes or qualities to the relation itself, without transforming it into something which it is not… In other words, if one reifies the relation it is not a relation anymore’ (174).31

This conception of relation as a form of śakti inspires Bhartṛhari’s novel reinterpretation of samavāya, the ‘inherence’ relation, in verses 8–12 of the SSam. Recall that for the Vaiśeṣika relational realists, samavāya is an independently existent category, often described as the ‘glue’ that binds together the other independent categories of existence.32 Kaṇāda’s canonical definition of samavāya is ‘that which produces the cognition ‘this is here’ [or ‘this subsists in this’] with respect to a cause and its effect.’33 In this respect, it is that objective category which localizes, or restricts, universals to particulars. In an earlier verse (6), Bhartṛhari concurs that samavāya is often called ‘relation’ (saṃbandha) because it seems similar in virtue of its dependent status. However, he adds that this does not indicate the general capacity of sambandha, because samavāya’s function is restricted (niyata) to certain domains (i.e., primarily between whole and part). Thus, Bhartṛhari reasons that this referential restriction must occur in virtue of some higher-order ‘service rendered’—i.e., an even more general form of sambandha that restricts the domains of relational subsets such as samavāya.

That capacity (śakti), called samavāya, rendering service to capacities, beyond difference and identity (bhedābheda), being established otherwise, is assisted by saṃbandha (relation), which is beyond the attribute of all categories or objects (padārtha) and which is characterized by everything. This is the tradition from the ancients. By others it is taken for granted that saṃbandha be always made into a category or object. [But] with this, word meaning cannot be sorted out.34

This description of samavāya itself as a śakti completely departs from the Vaiśeṣika realists; no sources describe it as such. It is an innovation of Bhartṛhari’s system (at least insofar as the traditional authority of ‘the ancients’ referred to above remains obscure). According to Bhartṛhari, a śakti is something neither identical with, nor different from, its substratum (Houben: 183–4). Thus, in identifying the samavāya relation with śakti, ‘beyond difference and identity’ (bhedābheda), Bhartṛhari not only de-reifies it from its categorical status, but also renders it referentially ambiguous or vague in terms of bhedābheda. This is quite in opposition to the Vaiśeṣika, whose ‘radical’ approach to realism invariably consisted of pinpointing the exact character and properties of a category (padārtha), and how precisely it differs from all others. So, the delimited relation of samavāya already corresponds to a capacity (śakti) of linguistic reality that establishes the bond between universal and particular.

And yet, despite this redefinition of samavāya as a form of śakti, it is still not a ‘capacity’ general enough for Bhartṛhari’s understanding of sambandha. In kārikā 11 above, he stipulates that saṃbandha itself actually ‘assists’ (anugṛhṇāti) samavāya. Saṃbandha accordingly becomes a capacity of an order higher than mere inherence, and thus of all qualified cognitions; it is ‘sarvalakṣaṇa’ which means, following Helarājā, that each and every object characterizes and defines saṃbandha (Houben: 184.). Houben remarks, ‘[t]o say that saṃbandha ‘relation’ is ‘defined by everything’ amounts…to saying that it is not defined, not delimitated [sic], not characterized by a specific, restricted domain in which it occurs. It is therefore just another way of expressing the complete absence of a specific own character of relation’ (185). When we attempt to reify relation, we are left with a set of abstractions that correspond to a gradation of capacities that ‘render service’ to other capacities, but we never ‘bottom-out’ at a determinate, self-standing referent that is ‘sambandha.’ The abstract status of these phenomenal capacities corresponds to the restriction of semantic domains associated with the relata in that domain: saṃbandha (śakti-of-śaktis) → samavāya (śakti) → other śaktis (Houben: ibid.). The upshot is that as the definition of sambandha becomes more and more abstract (viz., moving from tokens of ‘relation’ to its general form), rather ironically, it eventually becomes coextensive with the concrete capacity that enables meaning and reference to occur in the first place.

In this concrete capacity of ‘languaging’, sambandha is a peculiar form of intentional activity that interprets signs through the designation of other signs (i.e., ‘semiosis’); yet, for that very reason, its nature cannot be represented thereby.35 The ultra-dependent form of relation, in other words, entails that it is not an autonomously intelligible ‘critter’ (cf. Phillips: 23; Dunne: 44–5); viz., it cannot be isolated from its relata and remain itself, for any attempt to reify ‘relation’ as an independent term in judgment will presuppose what it tries to reify in the very act of reification. Houben concludes that we ‘end up with a relation which cannot be expressed or signified as it is, which is indeterminate and to all intents and purposes without independent character’ (190). In direct contrast with the Buddhist nominalist and Vaiśeṣika naïve categorical realists, then, Bhartṛhari does not believe that ‘relation,’ in its most general form, can be identified with any specifiable relational ‘property’ of dependence (pāratantrya), or inherence (samavāya), etc. Rather, saṃbandha is intentionally disclosed as an all-pervasive and continuous ‘śakti’—the transcendental capacity of śabda-brahman to manifest the diversity of the world in and through the practical use of language.

4. The Buddhist Denial of Real Relations

Before explaining Utpala’s form of relational realism—and Bhartṛhari’s influence thereupon—it would help to briefly outline the Buddhist position, Utpala’s pūrvapakṣin in the SS.36 In the ĪPK I.2.11, Utpala summarizes the reasons for the Buddhist rejection of a realist conception of relation, alluding to Dharmakīrti’s deconstructive arguments in the SP:37 ‘[There is no such thing as (sambandha)] because its form, [which is based on] a dual-locus, is not unitary, and because what is [already] actualized (siddhasya) cannot ‘require’ another, and dependence, etc., do not make sense. Thus, the agent is also a conceptual construct (kalpita).’38 Essentially, Dharmakīrti is arguing that sambandha is something situated in two places (dviṣṭha).39 However, since, according to Dharmakīrtian nominalism, all real momentary entities are ‘self-contained’ and ‘reside in themselves,’ the ‘distributed,’, or ‘continuous’ (anvaya), nature of the sambandha necessarily undermines its status as an ultimately real entity.40 For once a particular moment comes to exist in the causal procession, instantiating a self-contained essential nature (svabhāva), it ipso facto requires no dependence merely to do so. We should mention that the same argument applies, mutatis mutandis, to notions of continuous action, which unify a diversity of moments into a single continuous form.41 Thus, Dharmakīrti himself views the definitions of relation and action as tightly intertwined, concluding on this basis that the relative notion of an agential self—i.e., one upon whom the realization of distinct actions depends—is ultimately also a conceptual construction.42

In the SS, Utpala expands upon the Buddhist reductionist position he introduces in the ĪPK, such that the simultaneously unified and diverse form of ‘relation’ is simply a flagrant violation of the law of the excluded middle: ‘Furthermore, a unity of many that is coextensive with multiplicity does not make sense, because [it is] incompatible with the nature of existence and non-existence. There would arise (udbhavet) either unity from a manifold material cause (anekasmād upādānād ekam), or a manifold from a unity—insofar as they are only mutually unrelated, they would resolve simply in themselves.’43 The Buddhist therefore sees no difference between the assertions ‘X is both unified and diverse’ and ‘X both exists and does not exist’. If X exists as a simple (ekātva) particular, it definitionally does not participate in the identity of any other real existent, Y.44 Any ascription of these predicates to a phenomenon must therefore be sequential properties of distinct moments, not simultaneous attributes of a substantial (dravyatas) existent.

In effect, then, the Buddhist demonstrates that the axioms of nominalism and its dichotomous ontology—i.e., between ultimately real svalakṣaṇa and merely conventional sāmānyalakṣaṇa—automatically pigeonhole sambandha into the same category as universals, for the distinction between non-relational particulars and conceptual elaboration associated with the determinate language of discursive thought necessarily entails that ‘relations’ only appear in the latter moment. Thus no genuine continuity between the two moments can obtain, for all relations are attributed post hoc to the disconnected sequence.45 As Utpala’s Buddhist concludes, ‘Hence, only at the time of conception (pratītikāla) do we accept a relationship as though universal.’46 Since both are ‘distributed/continuous’ (anvaya) entities (Dunne: 80), they are both not ultimately real—viz., they only appear in the moment of determinate conceptual elaboration (kalpanā), not the initial indeterminate, non-relative moment of perceptual cognition (pratyakṣa).47 The conceptual dichotomy between the ‘unity’ and ‘diversity’ as predicates of a substantial existent therefore cannot characterize the real momentary entity, for each numerical property can only appear sequentially, characterizing distinct phenomena in the causal procession.

I argue below that Utpala replaces the implicit Buddhist identification of the categories of sambandha and sāmānyalakṣaṇa with Bhartṛhari’s identification of sambandha and śakti. That is to say, the general form of relation, or unity-in-diversity, is not a predicative property of a substantial existent, one that we can characterize or describe in terms of other determinate concepts. Rather, it is coextensive with a transcendental form of action whose intrinsic continuity ultimately makes the intentional form of synthetic judgment—and thus practical reference to universal entities—possible in the first place. In terms of the relational axes mentioned earlier, this redescription of sambandha represents a movement from the Buddhist’s ‘horizontal’ assessment of ‘relations’ in terms of various conventional tokens (i.e., those experienced solely in the moment of vikalpa) to an analysis of the general type (i.e., ‘relationality-as-such’) which manifests ‘vertically’ in the continuous action of vimarśa, or recognitive judgment.

5. Utpaladeva’s Pragmatic Elaboration of Sambandha-Śakti

Although Bhartṛhari is not explicitly mentioned, we can begin to appreciate his influence on Utpaladeva with reference to the general methodology of the SS. That is, Utpaladeva does not marshal arguments for any subset of relations (e.g., pāratantrya, kāryakāraṇabhāva and samavāya), but instead seeks to articulate, and prove to be real (satya), the existence of a general form of sambandha. According to Utapala, the common feature of every ‘relation’ (e.g., ‘mixture’ (saṃsarga), ‘contact’ (saṃparka), ‘fusion’ (saṃśleṣa), ‘dependence’ (pāratantrya) and ‘requirement’ (apekṣā)48) is nothing other than unity-in-diversity (bhedābheda).49 Like Bhartṛhari, then, Utpala abstracts from all tokens of sambandha to get a handle on its general logical form, and then infers that it is irreducibly characterized as being neither wholly identical with, nor wholly distinct from, its respective relata.

After the Buddhist lambasts this idea as theoretically incoherent (on the basis of the nominalist modes of reasoning just outlined), Utpala cites none other than Dharmakīrti himself to show that the Buddhists already take for granted the existence of relational cognitions that unify a diversity of relata: ‘First of all, this clear apprehension (pratīti) is acknowledged even by you [Buddhists] for the sake of designating a relation that pertains to the form of a unity of many, such that it is said: ‘Thus entities themselves are disjunct; conceptual construction conjoins them’ [SP 5cd]. That its form is a conceptual construction, though (ca), is not a problem. Even the distinction (vikalpa) ‘this is a pot,’ ‘this is a cloth’ is just a conceptual construction (kalpana).’50 This passage articulates Utpala’s movement of the conversation into a ‘transcendental’ register, in the sense that the real issue at hand is not Dharmakīrti’s justified skepticism towards the reality of any intentional content, but an incoherent skepticism toward the synthetic form of cognition itself—i.e., the ‘capacity’ that lends conceptual determinacy to this content. For Utpala, in other words, the ‘problem’ with unity-in-diversity is not that relational cognitions are conceptual constructions, the ‘problem’ is rather in thinking this means they are not real (satya).51 Dharmakīrti purportedly wants to consider relations to be a form of conceptual ‘error’ (bhrānti) or ‘defect’ (doṣa)—that is, a superimposition upon a reality that is constitutively otherwise (i.e., momentary, non-conceptual, impersonal and non-relational). But Dharmakīrti takes for granted that designating perceptual objects as particulars already instantiates a synthetic form of cognition—and, most importantly, he does not consider this form of cognition an error.

Of course, Utpala’s Buddhist acknowledges that he does not regard the synthetic form of these perceptual judgments of particular objects as purely conceptual, due to the causal efficacy of the ultimately real percept: ‘In this case, there is also a perceptual appearance such that (tathā) there is precisely the form of [determinate, individual] things like pots and cloths, etc.; hence it [i.e., the appearance] is not a conceptual construct.’52 This statement reflects the prototypical Buddhist position of the causally privileged nature of perceptual cognitions—even if the relations that characterize the designation are not real, the perceptual ‘data’ still instantiate the self-standing nature of ultimately real objects. Insofar as the content of these perceptual judgments is a particular object, the cognition is not just a synthetic conceptual construction.

But Utpala argues on constitutively practical grounds that Dharmakīrti is not entitled to infer from the conceptual relations that distinguish objective content that the synthetic form of cognition itself is consequently unreal. That is to say, there is no practical difference between the general conceptual form of relations that distinguish particular objects from each other (i.e., an individual ‘pot’ and ‘cloth’) and the general form that happens to refer to a conceptual relation (i.e., ‘the king’s servant’ and ‘rice in a pot’53). In Utpala’s view, this imagined dichotomy is based on a category mistake that conflates the ‘external’ tokens of relation that define the contentful appearance of perceptual objects with the ‘internal’ relations that practically constitute the synthetic form of judgment.54 After all, Utpala insists, even the practical recognition of an empirical error presupposes a synthetic form of judgment, insofar as it requires a cognition of similarity between two different appearances—i.e., the universal idea previously overturned and the percept currently doing the overturning55: ‘For it is not because of its being closely connected with the appearance of an object that it [i.e., the cognition of relation] is [difficult to be uprooted] by a defeater; rather, it is just because of the real existence of similarity among appearances of objects. Even in the case of the illusion of silver, there actually is the real presence of similarity with mother-of-pearl: For all misapprehensions (bhrānti) have similarity as their content—and some of those (cānyā), by excluding similarity, make sense only as being closely connected to the appearance of an object.’56

For Utpala, the Buddhist misconstrues the practical significance of his own observation—i.e., that some perceptual ‘errors’ (such as relation and partite extension) are difficult to practically uproot because they are ‘closely attached’ to the appearance of the object, for this ‘close attachment’ is not an instance of conceptual error, but rather just what it practically means for a particular object to appear as it does in perceptual judgment. In other words, the peculiar phenomenal feature of the appearance of particulars is not that it picks out some ‘causally privileged’ percept that abides wholly unrelated to this synthetic act of conceptualization. Rather, it is that the apprehension of an appearance as an external particular simply consists in the conceptual exclusion of everything that appears as similar thereto under some practical description.57 For Utpala, the phenomenal particularity of any given object is continually ‘pinpointed’ in consciousness through the conceptual process of apoha, which excludes appearances according to their relevance for the determinations necessary to make sense of its own practical activity.58 On pragmatic grounds, therefore, Dharmakīrti cannot appeal to a judgment of some non-relational particular to discount the validity of the perception of relations such as ‘the king’s man’ and ‘rice in a pot,’ even though they are conceptually determined. For, just like the appearances of the particular objects, these relational cognitions are never practically overturned, even though they exhibit a form of unity-in-diversity.59

According to Utpaladeva’s system of absolute idealism, then, this general form of relation, unity-in-diversity, can only manifest due to the reflexive nature of self-consciousness that spontaneously (svātantrya) re-cognizes a manifold of perceptual appearances in conceptual terms.60 This is the true significance of relationality for Utpala, as he states in the ĪPK II.2.4: ‘Things that are self-contained and manifested separately possess a unity in the sense of mutual connection within the unitary knowing subject. This is the basis of the ideas of relation.’61 We can say, therefore, that Utpala regards all tokens of sambandha—from grammatical to causal—as necessarily instantiating the general unity-in-diversity of reflexive judgment. This proves the Buddhist cannot appeal to the apparent distinctions between particular objects to establish the causally privileged (read: ultimately real and non-relational) status of perceptual cognitions, because the referents of both grammatical and causal relata are determined through the same synthetic form of cognition.

6. Judgment and the Practical Determination of Relational Tokens

As we have seen, Utpala argues that all relations exhibit a general form of bhedābheda. At some point in the SS, he entertains a possible Buddhist objection that this theory merely recapitulates the theoretical problems of the Vaiśeṣika category of samavāya, for this ‘unity-in-diversity’ supposedly means a single, general form relates everything, which either entails the familiar regress of Dharmakīrti or the inability to account for the diversity of relational tokens. But Utpala retorts that ‘this would be the case if a relation were to consist exclusively of unity; [but] a relation is a unity with parts to the extent (yāvatā) that the parts are distinct. And therefore, the unity that is ‘tinged’ (uparakta) by “the king’s servant” is a different relation from the one implied (upalakṣaṇa) in the ‘color’ (uparāga) of “father and son.”’62 In other words, Utpala avoids the regress of relational reification by upholding Bhartṛhari’s conception of relational tokens as dependent upon the nature of the appearance of relata. No two particular relations possess precisely the same ‘color,’ because the particular form of the relata determines the nature of the relation.

But, for Utpala, the ‘relation’ between the king and servant depends upon the conventional circumstances in which they are investigated, for even the perceptual determination of the relata themselves ultimately depends upon the practical capacity of reflexive judgment (vimarśa) to take any given appearance as either relational or non-relational.63 In other words, the relational content of all qualified cognitions—even relative absence and otherness—spontaneously conform to the practical interests and intent of the agent in virtue of the recognitive capacity of vimarśa. To prove his point, Utpala appeals to the rules of grammar wherein one can interpret distinct referents as a single compound word (dvandva).64 For instance, in the phrase ‘rājñaḥ puruṣa’ (‘the king’s servant’), the king is a distinct word with its own genitive case ending, and thus a dyadic relation obtains between distinct terms. However, in the case of ‘rājapuruṣa’ (i.e., king-servant), the king ‘appears (prathate) as having completely (ekāntena) assumed (āpanna) the identity of the qualified, its own form being altogether relinquished. Thus, in this case, there is no sense in expressing a relation.’65 Likewise, in the case of simple predication, the appearance ‘blue-lotus’ could either be relational (i.e., ‘the lotus is blue’) or non-relational (e.g., ‘the blue-lotus is not white’), depending on whether we wish to take the appearance of ‘blueness’ as distinct from the lotus.

Indeed, Utpala maintains that for any qualified form of relational cognition (he includes those of absence, difference and propositional kārakas66), its individual terms, and the grammatical relations that define their intentional appearance, always already disclose the existence of the capacity for them to be interpreted in either way (i.e., as a relatum or a term). Theoretically, then, the givenness of any appearance(s) necessarily instantiate an infinite capacity for relational determination; viz., any single appearance contains within itself an unbounded potential for, say, predication or relative absence. Which of these conceptual possibilities comes into clear and distinct awareness is precisely that which judgment renders intelligible for the sake of some practical orientation. There is thus an incipient or nascent form of infinite interpretability even in the act of perceiving a single, unitary object.67 In the Pratyabhijñā system, this only makes sense because the reflexive, self-luminous consciousness of the agent (i.e., the ātman) spontaneously enacts relational tokens that render intelligible the diverse, indeterminate factors that enter into its own practical activity.68

In his effort to rebuff the Buddhist nominalist, Utpala extends this reasoning to the causal relation, for even the simultaneous appearance of a distinct sequence of ‘relata’—i.e., in the ‘metaphorical’ (gauṇavṛtti) sense of ‘assister’ and ‘assisted’69—must indicate an act of conceptual synthesis, a certain form of unity-in-diversity.70 This form of relational judgment presents two temporally distinct moments as simultaneous for the sake of those practical endeavors, the conceptual synthesis of distinct appearances upon which all conventional life ultimately depends. Pertinently, to buttress his point, he invokes an argument initially deployed by Bhartṛhari71: For one running along a path (or, more generally, engaged in any task where a manifold of perceptual data go consciously unrecognized) perceptual features of the surrounding environment (e.g., the grass on the side of the road) remain phenomenally unestablished. However, one can later recall having experienced these things, but only insofar as they appear in a conceptual form expressible in terms of judgment.72

For Utpala, the fact that a subsequent mnemonic cognition can ‘unfold’ the many indeterminate perceptions which were ‘enfolded’ into the determinate conceptual judgment at the time of activity demonstrates that ‘consciousness does not passively record sensory data but appropriates them through some kind of silent expression that can later be elaborated on when (s)he thinks back on what (s)he has done and explains it. This means that discursive thought is nothing but the development of a subtle, condensed expression already present in any act of perception’ (Ratié 2021, p. 96).73 According to Utpaladeva, then, there is no hard and fast distinction between the indeterminate moment of initial perception and its subsequent conceptual determination in the stream of consciousness; the ‘clear and distinct’ quality of perceptual judgments already presupposes the transcendental operation of a continuous form of sambandha-śakti that gives relations in precisely the way they are self-consciously taken. Utpala’s invocation of Bhartṛhari therefore suggests that even though he does not adopt the latter’s theory of sphoṭa wholesale, he concurs that an implicit continuity of discursive activity defines a progressive conceptual determination of perceptual content.

Summarizing the influence of Bhartṛhari on the Pratyabhijñā, Torella comments, ‘in order to undermine the discontinuous universe of the Buddhists, [Utpaladeva] decides to avail himself precisely of the [doctrine of] the language-imbued nature of knowledge, which is meant to demolish its main foundation stone, the unsurpassable gulf between the moment of sensation and that of conceptual elaboration, representing, as it were, the very archetype of the Buddhist segmented reality’ (Torella 2008, pp. 350–51). We can now appreciate not only the significance of Bhartṛhari in this transcendental reformulation of relational realism, but also why the topic of sambandha was so central to Utpaladeva that he decided to revisit the subject on its own in the Siddhitratyī; for the problem of proving the reality of a dynamic, unified self against the Buddhist nominalist coextends with establishing a continuity between the initial moment of bare perception and its subsequent conceptualization—and only an objective, general form of relationality could provide a sufficient condition to establish such a feature of reflexive consciousness.

7. Conclusions

I have argued that Bhartṛhari’s ‘emanationist’ metaphysics of sphoṭa (i.e., the capacity of divine ‘speech’ (paravāk) to realize itself in determinate, delimited expressions) demanded a novel reinterpretation of sambandha as a transcendental form of śakti—a ‘śakti-of-śaktis’. According to Bhartṛhari, while we might designate a particular token of ‘relation’ through a universal term (e.g., pāratantrya, samavāya), all these forms derive from the relational capacity of paravāc that makes the very act of linguistic reference—and thus all qualified knowledge—possible in the first place. We have seen that the Pratyabhijñā non-dualists, in the service of establishing their own absolute idealism, appropriate and develop this identification of sambandha and śakti into a series of practical arguments against the Buddhist relational eliminativist: in the SS, Utpala argues that the apparent unity-in-diversity of relations cannot be a form of conceptual ‘error’ (bhrānti), like universals, because the practical appearance of any conceptual ‘errors’ itself depends upon the synthetic form of practical judgment. He then proceeds to claim, on the basis of grammatical and phenomenological considerations, that this general synthetic form of judgment determines the appearance of any given referent as either relational or a separate term in yet another predicative relation. In other word, no ‘horizontal’ linguistic designation (i.e., the distinct relational tokens of particular judgments) can ever represent the ‘vertical’ capacity to interpret intentional content in those very terms (i.e., as either relational or non-relational).

In the end, the Buddhist nominalist cannot deny the unity-in-diversity of the general form of relation on purely theoretical grounds without thereupon casting asunder the practical capacity to determine a manifold of perceptual appearances through the act of conceptual designation. And, since the Buddhist presumably takes this capacity for granted even in the perceptual judgment of particular objects, he cannot invoke our awareness of these very appearances to justify the constitutively non-relational character of the ultimately ‘real.’ In sum, then, Bhartṛhari’s reformed conception of sambandha contributes greatly to the Pratyabhijñā objective idealist project, which seeks to leverage transcendental arguments to move beyond the skeptical representationalism and epistemic idealism that the dichotomous ontology of Buddhist nominalism invariably engenders.74