Dharmakīrti’s criticism of external realism and the sliding scale of analysis
Birgit Kellner, Vienna
Several studies have recently brought into focus the copresence of conflicting theories about reality and its cognition within the Buddhist logico-epistemological tradition, a phenomenon that is also often discussed in terms of that tradition’s doctrinal affiliation with the (realist) Sautrāntika and the (arguably idealist) Yogācāra systems. George Dreyfus (Dreyfus 1997), Sara McClintock (McClintock 2003) and John Dunne (Dunne 2004) have formed and elaborated the heuristic metaphor that Dignāga, Dharmakīrti and their followers employ a sliding, or ascending, scale of analysis where theories that at first sight seem to be contradictory are in fact located on hierarchically arranged levels of discourse: more intuitive and less accurate low-level
analyses are gradually replaced by less intuitive and more accurate higher-level accounts. Thus, for instance, the idea that external objects cause a perception that appears in their form is replaced by the view that cognition is an entirely mind-internal phenomenon. While Dreyfus and McClintock deal with later representatives of the tradition in South Asia and Tibet, Dunne is concerned with the historical Dharmakīrti and presents the sliding scale in a study of the Pramāṇavārttika (henceforth PV), which he analyses on the basis of Devendrabuddhi’s (ca. 630–690 CE) and Śākyabuddhi’s (ca. 660–720)
commentaries. Whereas Dreyfus and McClintock place higher emphasis on epistemological factors in distinguishing the levels on the scale, Dunne introduces the scale to account for seeming conflicts between multiple ontologies. His altogether four levels differ in terms of the type of entities that are considered as ultimately real. These levels also represent steps in a continuous analytical process that leads from the first all the way up to the fourth. On every level, entities that were considered as ultimately real on the previous are relegated to a merely conventional status, with the help of arguments that I shall here refer to as “transition arguments.” These transition arguments point out problems in the low-level theory which cannot be resolved
without abandoning its fundamental principles. In a next step, and as a consequence of these arguments, these principles are abandoned and logically contradictory ones that form part of the higher-level theory take their place. Dunne’s study suggests the following arrangement of the four levels, here presented together with the transition arguments that link them:1 1. Beliefs of ordinary persons about reality: Spatio-temporal composites exist in distinction of their parts; in philosophical debate this view is exemplified with the whole
Birgit Kellner 292 (avayavin) of Nyāya. Transition argument to level 2: the whole is neither identical with nor different from its parts, and therefore not ultimately real. 2. Abhidharmic typology: Only irreducible constituents (dharma) are real, which to some extent retain conceptual, spatial and temporal extension. Transition argument to level 3: the universal is neither identical with nor different from particulars and therefore does not truly exist, and a temporally or spatially distributed entity is neither identical with nor different from momentary phases or atoms and is not ultimately real either. 3. External realism: Only discrete perceived particulars, spatial atoms and momentary phases are ultimately real. Transition argument(s?) to level 4: extra-mental matter is not ultimately real/extra-mental particles do not cause the objective appearance in perception because the subject-object-duality in cognition is a product of ignorance and therefore erroneous, and/or (?) because the perceptual image of the object is neither singular nor plural.
4. Epistemic idealism: All entities are mental (?). In Dunne’s account, the movement up the scale may be considered to exhibit soteriological dimensions. Its philosophical significance, however, lies in that it manifests an underlying uniform method, consisting in a mereological strategy that takes into consideration how (what appear to be) wholes can be broken up into, and reduced to, individual parts – more specifically, Dharmakīrti is held to perform a reductive analysis, in the course of which he reduces a successively higher number of entities to infinitesimal particles through a chain of “neither-norarguments.” The focus of this paper is the transition from level 3 (external realism) to level 4 (epistemic idealism). As the question-marks in the
above presentation already indicate, Dunne’s account of this transition, based on PV 3.194–224, is not entirely clear. He identifies the fundamental error of external realism, and the principle that is abandoned in the move to epistemic idealism, as the belief in the existence of extra-mental matter, and claims that this belief is eliminated through the realisation that “the subject/object-duality apparent in awareness is actually due to the influence of ignorance (avidyā).” Because this duality is erroneous, any belief that is based on it, such as the notion that extra-mental particles cause the objective appearance in perception, is erroneous as well.2 Two pages later, Dunne claims that the primary argument against the existence of extra-mental objects is the “inability to specify whether the image in perception is singular or multiple.”3 It is not clear to me how Dunne conceives of the relationship between these two arguments, and how exactly he derives them from PV 3.194–224.4 His onto
4 The problem more specifically lies in the connection between the premise that perceptual images are neither one nor many and the conclusion that extra-mental matter does not exist/does not cause perceptions – this conclusion seems to me not to be (at least) explicitly drawn in PV 3.194–224. The section does contain the claim that the duality of grasper and grasped is a product of ignorance, but the only
consequence that seems to be drawn from this as far as objects are concerned is that, when differences within cognition are fundamentally erroneous, the difference between objects becomes erroneous as well because objects are established as different from each other on the basis of differences within cognition, cf. PV 3.214: tadbhedāśrayiṇī ceyaṃ bhāvānāṃ bhedasaṃsthitiḥ | tadupaplavabhāve ca teṣāṃ bhedo ’py upaplavaḥ || One could argue that this erroneous
Dharmakīrti’s criticism of external realism 293 logical reading of the transition, furthermore, seems to be due to the overall importance of ontology in the construction of the scale, and due to the need to have the transition from level 3 to 4 take off precisely where the transition from level 2 to level 3 ended. But this ontological reading also makes the two arguments sound rather unconvincing and raises the suspicion that something here is not quite right. How could the realisation that the subject/object-duality of cognition is due to ignorance eliminate the belief that extra-mental matter as such exists?
Or, for that matter, how could this belief be eliminated through the realisation that perceptual images are neither one nor many? There is (at least) one further section in the same chapter of the Pramāṇavārttika, with close parallels in the Pramāṇaviniścaya (henceforth PVin), where Dharmakīrti articulates a transition from what Dunne refers to as external realism to epistemic idealism: the exposition of the means of valid cognition (pramāṇa) and its result (pramāṇaphala) in PV 3.301–366 (PVin 1.34–57).5 In what follows, I am going to sketch a transition argument (actually, a complex of arguments) that is given there and argue that it yields a rather different picture of this transition than Dunne’s model implies.
character somehow implicitly extends to extra-mental matter as such, but would still have to take into consideration that Dharmakīrti does not include the refutation of external objects among the more or less explicit goals of proof of this section. Since Dunne bases his reading of Dharmakīrti on Devendrabuddhi or Śākyabuddhi, it is conceivable that the arguments he attributes to Dharmakīrti are in this form articulated in their commentaries – in fact, he suggested as much in the discussion that followed the presentation of this paper at the conference. However, one should then be alert to the
possibility that commentators draw arguments to their logical conclusion by exploiting their implications, without necessarily following closely the goals and methods of proof in the basic text. It is risky to rely on commentaries for the construction of a scale of analysis that involves such transitions because commentators, being already convinced that the higher-level theory is preferable, might feel tempted to involve higher-level principles to a higher degree in the criticism of lower-level theories than the author of the basic text. 5 The numbering of the stanzas from PVin 1.41 onward presents problems. Vetter’s edition of the Tibetan translation contains no equivalents for 41cd and 42a. Lindtner conjectured PVint 1 88,23f. 'dod pa daṅ mi 'dod par snaṅ ba ni rtog pa yin gyi dbaṅ po'i blo ni ma yin no źe na as 41cd, parallel to PV 3.344ab iṣṭāniṣṭāvabhāsinyaḥ kalpanā nākṣadhīr yadi (Lindtner 1984: 164). However, Steinkellner’s edition of the Sanskrit text shows that this is a prose sentence, see PVin 1 37,1: iṣṭāniṣṭāvabhāsinyaḥ kalpanā
nendriyabuddhaya iti cet. Vetter’s PVin 1.42b, raṅ bźin sems pa yin na ni, metrical in Tibetan, is also a part of the prose in the Sanskrit text, cf. PVin 1 37,7 svabhāvacintāyāṃ (cf. PV 3.350). Vetter’s PVin 1.42cd don rig de yi bdag ñid phyir raṅ ñid rig pa ’bras bur bśad, a half-stanza in Tibetan, is considered as prose in Steinkellner’s edition of the Sanskrit: tādātmyād arthasaṃvidaḥ svasaṃvit phalam ucyate. While this sentence (also paralleled in PV 3.350) has the necessary 16 syllables for a half-stanza in anuṣṭubh metre, it is metrically incorrect, for the seventh syllable of the third pāda (°vi°) has to be long in either recognized variant of the anuṣṭubh metre (cf. the table in Steiner 1996, English translation in MacDonald 2007). This can be achieved if °vidaḥ is conjectured to °vitteḥ, against all three manuscripts – a conjecture which allows to recover the otherwise mysteriously missing PVin 1.41cd. Steinkellner omitted 41cd; his stanza numbering is not affected by whether or not one makes the conjecture and recovers it. Vetter’s numbering is in any case one number higher than Steinkellner’s from 1.43 onward.
The transition from externalism to internalism in Dharmakīrti’s exposition of the means of valid cognition (pramāṇa) and its result (pramāṇaphala) The section on means and result in PV and PVin deals with what, in the case of perception, serves as the means of valid cognition, and what exactly is its result. In PV, which is structured as a commentary to Dignāga’s Pramāṇasamuccaya and -vṛtti, this section corresponds to PS(V) 1.8cd-10. The section in PVin is shorter, differs in wording as well as in the arrangement of arguments; certain arguments are also omitted there. The following summary is based on PVin, though the same kind of argument with respect to Dunne’s scale could also be made on the basis of PV, perhaps with slight variations. Like Dignāga, Dharmakīrti basically advances the theory that means and result are non-different (anarthāntara) in that both are aspects of cognition, distinguished in
terms of how one conceptually determines them as properties, but not existing as separate real entities.6 In expounding means and result, Dharmakīrti pits two alternative theories of perception against each other that can be characterised as externalist and internalist respectively. The externalist theory assumes that some extra-mental, material object causes a perception that has its form. By contrast, the internalist theory assumes that perception, as well as all other cognitive activity, takes place solely within the mind, and that nothing else is to be experienced by cognition. These characterisations are functionally equivalent to what Dunne refers to as external realism and epistemic idealism, but their definitions differ because they concentrate on those features which are of immediate relevance in the exposition of means and result, which places epistemological aspects of these two levels of analysis in
the foreground where Dunne concentrates on ontology. It is to be emphasized, furthermore, that when I speak of an “externalist” theory in the following I am not referring to any philosophical theory that posits external objects of perception, but specifically to the theory that Dharmakīrti first adopts and then abandons: that external objects produce a perception which has their form (ākāra), or which resembles them. Dharmakīrti begins with an externalist account of means and result (PV 3.301–319, PVin 1 30,9–33,11). The means of valid cognition is required to differentiate valid cognition according to
individual objects. In the case of perception, the means must thus be capable of explaining why, for instance, a visual perception is one of blue, and not of yellow. The result (pramāṇaphala) is the understanding of the individual external object (arthādhigama). The means is nothing other than the similarity of perception to its external object (arthasārūpya),7 or, in other words, the fact that it has the form of the validly cognised object (meyarūpatā).8 After refuting other candidates for means of valid cognition, such as the sense-faculty or the senseobject-contact (indriyārthasannikarṣa), Dharmakīrti undertakes a critique of the externalist theory that leads him to abandon it, as well as its conception of means and result, in favour of an internalist view – a clear transition from one level to another within the framework of a sliding scale.
See PVin 1 33,4–6 (with the first part parallel to PV 3.315): jñānāṃśayos tu sādhyasādhanabhāvo vyavasthāśrayatvāt. vastvabhedāt kriyākaraṇayor aikyavirodha iti cet, na, dharmabhedābhyupagamāt. 7 PVin 1 31,10f.: na ceyam arthaghaṭanārthasārūpyād anyato jñānasya sambhavati, PVin 1.34ab=PV 3.305ab: arthena ghaṭayaty enāṃ na hi muktvārtharūpatām | 8 PV 3.306ab: tasmāt prameyādhigateḥ sādhanaṃ meyarūpatā | Parallel in PVin 1.34cd, which has pramāṇam for sādhanam.
Dharmakīrti’s criticism of external realism 295 Dharmakīrti begins his critique of externalism by questioning the notion of the “cognition of the external object” (arthasaṃvit) that the externalist posits as the result is questioned. Speaking as an internalist, Dharmakīrti grants that there is an object-specific cognition. But why should it be, of all things, a cognition of an external object? According the externalist, this is the case because the cognition arises from from the external object and resembles it (utpattisārūpyābhyām). This view, however, has an unwanted consequence: “Then, an immediately preceding cognition with the same object would turn out to be the object.”9 Assume a situation where two perceptions arise in immediate succession from the same kind of external object, e.g. a blue patch. The first serves as the immediately preceding and homologous condition
(samanantarapratyaya) of the second and is therefore its cause. The second perception has the same form as the first, i.e. blue, and thus resembles it. Both conditions for the object of a cognition are consequently fulfilled, and the first perception would therefore have to be considered the object of the second, which an externalist clearly cannot accept. Interpreted in this fashion, Dharmakīrti’s argument points out that there is at least one situation where the externalist must admit that something is an object which by his own standards is not; his definition turns out to be over-extensive.10 This argument will subsequently be referred to as “the samanantarapratyaya-argument.” The externalist replies with a counter question: But of what, then, is experience – if it is not of an external object?11 Dharmakīrti first entertains the idea that a perception is linked with the specific external object that one subsequently determines in the form “this was heard, this was seen.” But this is putting the cart before the horse, for precisely the close connection (pratyāsatti) between perception and its object is under scrutiny. A perceiver can only determine on the basis of that connection that a particular object was seen, so the primary question remains just what this connection might be.12 Having argued that a pre-reflective connection between perception and its
alleged external object has to be accounted for, Dharmakīrti now sets out to prove that precisely this is impossible. Inspired no doubt by arguments that Dignāga put forward in his Ālambanaparīkṣā and -vṛtti, he claims that the coarse or spatially extended form which appears in a perception (sthūlākāra) does not exist in each of the individual atoms that are held to cause it. Moreover, the coarse form enters perception as one. The many atoms can therefore not be said to appear in a coarse form in perception when they are aggregated (sañcita), for what has one form in perception cannot in external reality be many (bāhulya). Furthermore, the many atoms do not have such a coarse form. Neither is there one single coarse object of perception that appears as coarse, such as the Nyāya-Vaiśeṣika’s whole (avayavin),13 which is refuted at some length.14 This complex of arguments will subsequently
9 PVin 1 33,12–34,1: tad arthavedanaṃ kena? utpattisārūpyābhyām. anantaraṃ tarhi vijñānaṃ tulyaviṣayaṃ viṣayaḥ prāpnoti. PV 3.323: tatsārūpyatadutpattī yadi saṃvedyalakṣaṇam | saṃvedyaṃ syāt samānārthaṃ vijñānaṃ samanantaram || See Kobayashi 2005 for an account of Prajñākaragupta’s interpretation of PV 3.323. 10 My reconstruction of this argument follows Tosaki (1985: 7). 11 PVin 1 34,1: kva tarhīdānīm ayam anubhavaḥ? 12 PVin 1 34,2–5 (parallel in PV 3.324–325): idaṃ dṛṣṭaṃ śrutaṃ veti darśanaśravaṇābhyāṃ yatrāvasayapratyayaḥ, te tasyānubhava iti cet, nanu saiva tayoḥ pratyāsattir atra vicāryate – kathaṃ tat tasya darśanam iti. tayor hi sambandham āśritya draṣṭur eva viniścayaḥ. sa tadabhāve na syāt. 13 PVin 1 34,5–35,6: na ca viṣayasārūpyaṃ vijñānasya, (a) tatpratibhāsinaḥ sthūlākārasya pratyekaṃ paramāṇuṣv abhāvāt. (b) ekaś cāyaṃ jñānasanniveśīti na ca bahūni rūpāṇi sañcitāni tathā pratibhāntīti yuktam, ekarū
Birgit Kellner 296 be referred to as “arguments from incongruence” because they essentially raise as a problem that there is a fundamental incongruence between what might serve as external object of perception and what appears as its content. In terms of a proposed sliding scale of analysis, Dharmakīrti has so far provided arguments that identify problems in the low-level theory which cannot be resolved without abandoning its fundamental principles. The next step, replacing these principles with conflicting ones that form part of the higher-level theory, follows promptly, and answers the externalist’s
question of what the experience might then be (if not of an external object) in the negative: nothing is to be experienced through cognition. The defining characteristic of the external object (viṣayalakṣaṇa), meaning that it causes a perception with its form, has been refuted.15 Dharmakīrti goes on to explain means and result within an internalist framework. Because there is no cognition of an external object, which had been proposed as the pramāṇaphala by the externalist, self-awareness comes to be assumed as the result.16 The object of valid cognition is then the form of the grasped (grāhyākāra), whereas the means of valid cognition is the form of the grasper (grāhakākāra). These two characteristics of cognition are beyond reproach only when their distinction is recognised as corresponding to the way they appear to beings plagued by error (i.e. ignorance), while cognition in reality is without the forms of knower and known.17 What would happen if we were to replace Dunne’s arguments for the transition from level 3 to 4 with the transition argument proposed in the section on means and result? The heuristic value of the scale would be diminished because one could no longer claim that levels 3 and 4 are connected with mereological arguments that reduce entities to particles with the help of ontological neither-nor-arguments, for the simple reason that neither the samanantarapratyaya-argument nor the arguments from incongruence can be characterised in this way. But since this method is the single uniform criterion that links the individual levels and defines them as points on a scale, it then becomes questionable whether epistemic idealism can be meaningfully considered a further point along this scale at all. To rehabilitate Dunne’s version of the sliding scale, one would first of all have to make plausible how his transition arguments can prove the inexistence of extra-mental matter – I
pasya bāhulyavirodhāt. (c) bahuṣu ca tathāvidho nāstīti na tayoḥ sārūpyam. (d) nāpi sthūla eko viṣayas tathāvabhāsi … tasmān naikaḥ kaścid arthaḥ, yo vijñānaṃ sarūpayati. PV 3.321cd: sarūpayanti tat kena sthūlābhāsaṃ ca te ’navaḥ || Sentence (c) is open to diverse interpretations: (1) a reiteration of (a) in slightly different form as a closure of the argumentation about atoms, (2) an additional claim about the many atoms: in (b) it was presupposed that atoms appear as coarse when gathered together, so Dharmakīrti might here add that many atoms do not appear as coarse in some other form of combination which is not identical with an aggregation (sañcaya). 14 PVin 1 34,10–53,5, see Funayama 1990 for a study (in Japanese) on the basis of PVint. 15 PVin 1.38a: nānyo ’nubhāvyo buddhyāsti, PV 3.327a: nānyo ’nubhāvyas tenāsti. See Kellner 2010, for a discussion of transmission variants of the PV stanza. Cf. also PVin 1 35,10: vyastaṃ hi viṣayalakṣaṇam iti na kvacid anubhavo … 16 PVin 1.41ab=PV 3.332cd: tadānyasaṃvido ’bhāvāt svasaṃvit phalam iṣyate | 17 PVin 1.39–40=PV3.330c–332b: avedyavedakākārā yathā bhrāntair nirīkṣyate | vibhaktalakṣaṇagrāhyagrāhakākāraviplavā || tathā kṛtavyavastheyaṃ keśādijñānabhedavat | yadā tadā na sañcodyagrāhyagrāhakalakṣaṇā || Note that Steinkellner in PVin reads tathā kṛtavyavastheyaṃ as one word.
Dharmakīrti’s criticism of external realism 297 have argued above that this is far from clear. Secondly, one would have to find a way to regard these arguments, if indeed they can be derived from PV 3.194–224, as more important than the samanantarapratyaya-argument and the arguments from incongruence – which would inevitably raise the methodological question on what grounds these arguments should be weighted. These are, I think, the most important challenges for Dunne’s version of the scale. One fundamental challenge for any version of the sliding scale, however, is the apparent multiplicity of arguments that Dharmakīrti has on offer against externalism and for internalism, and their equally apparent heterogenous character. Clearly, the samanantarapratyaya-argument and the arguments from incongruence are of different caliber when it comes to refuting externalism – and the chapter on perception in PV holds a number of further intriguing arguments that scholars have so far not taken into consideration, or at least not in a comprehensive way. Instead of trying to hold on to the heuristic metaphor of a sliding scale that proves problematic in the face of textual evidence, it may in the end be more fruitful to take stock of the entire range of arguments that Dharmakīrti provides, and to pay close attention to the multiplicity of perspectives that they represent.
Primary sources
PV 3 Dharmakīrti’s Pramāṇavārttika, chapter on perception. See Tosaki 1979 (kk. 1–319) and 1985 (kk. 320–539).
PVin 1 Dharmakīrti’s Pramāṇaviniścaya. Chapters 1 and 2. Ed. by Ernst Steinkellner. BeijingVienna: China Tibetology Publishing House – Austrian Academy of Sciences Press 2007. For corrigenda, cf. Wiener Zeitschrift für die Kunde Südasiens 51 (2007–2008) 207–208 as well as http://ikga.oeaw.ac.at/Mat/steinkellner07_corrigenda.pdf (last accessed 14 January 2009). PVint 1 T. Vetter: Dharmakīrti’s Pramāṇaviniścayaḥ. 1. Kapitel: Pratyakṣam. Einleitung, Text der tibetischen Übersetzung, Sanskritfragmente, deutsche Übersetzung. Wien 1966.
PS(V) E. Steinkellner: Dignāga’s Pramāṇasamuccaya, Chapter 1. A hypothetical reconstruction with the help of the two Tibetan translations on the basis of the hitherto known Sanskrit fragments and the linguistic materials gained from Jinendrabuddhi’s Ṭīkā. 2005. http://ikga.oeaw.ac.at/Mat/dignaga_PS_1.pdf, last accessed 24 August 2009.
Secondary sources
Dreyfus 1997 G. Dreyfus: Recognizing Reality: Dharmakīrti’s philosophy and its Tibetan interpretations. Albany, NY 1997.
Dunne 2004 J. Dunne: Foundations of Dharmakīrti’s Philosophy. Somerville, MA 2004.
Funayama 1990 T. Funayama: Bubun to zentai – Indo-Bukkyō-chishikiron ni okeru gaiyō to kōki no mondaiten. {*Part and whole: an outline (of its discussion) in Buddhist epistemology and problems in the late period}. Tōhō gakuhō 62 (1990) 607–635.
Kellner 2010 B. Kellner: Towards a critical edition of Dharmakīrti’s Pramāṇavārttika. In: Text Genealogy, Textual Criticism and Editorial Technique = Wiener Zeitschrift für die Kunde Südasiens 52– Birgit Kellner 298
53 (2009–2010), ed. J. Hanneder and Ph. A. Maas, 161–211.
Kobayashi 2005 H. Kobayashi: Ninshiki no tayōsei to sono konkyo – Prajñākarateki appurōchi {Prajñākaragupta on What Brings about Diversity in Cognition}. Hikaku ronrigaku kenkyū {The Annals of the Research Project Center for the Comparative Study of Logic} 3 (2005) 75–81.
MacDonald 2007 A. MacDonald: Revisiting the Mūlamadhyamakakārikā: Text-Critical Proposals and Problems. Indotetsugaku Bukkyōgaku kenkyū 14 (2007) 25–55.
McClintock 2003 S. McClintock: The Role of the ‘Given’ in the Classification of Śāntarakṣita and Kamalaśīla as Svātantrika-Mādhyamikas. In: The Svātantrika-Prāsaṅgika distinction: what difference does a difference make?, ed. G. Dreyfus, S. McClintock. Somerville, MA 2003, 125–171.
Steiner 1996 R. Steiner: Die Lehre der Anuṣṭubh bei den indischen Metrikern. In: Suhṛllekhāḥ. Festgabe für Helmut Eimer, ed. M. Hahn et al. Swisttal-Odendorf 1996, 227–248.
Tosaki 1979, 1985 H. Tosaki: Bukkyōninshikiron no kenkyū, jōkan, gekan. Tokyo 1979 (jōkan), 1985 (gekan).
