A Problem for Ganeri’s Buddhaghosa

Jonardon Ganeri's Attention, Not Self is a terrific book. It aims to defend a picture of the mind that emerges from the work of Theravāda Abhidhamma Buddhists writing in Pāli, especially Buddhagosa (5th century CE). The Buddhists defend the theory of "no-self", i.e., the view that there is no ultimately existent referent of the first person pronoun "I", which plays the role of a subject with respect to conscious experiences and thoughts and the role of an agent with respect to mental, linguistic and physical acts. This raises a challenge. The theory of "no-self" is hard to reconcile with the fact that our conscious experiences and actions seemingly unfold within a perspective occupied by a subject who undergoes those experiences and performs actions. Ganeri brilliantly reconstructs Buddhaghosa's response to this challenge. Attention plays a central role in this response: it explains those perspectival features of ordinary conscious experiences and intentional actions which create the illusion of an ultimately existent self. Classical Representationalism. In conscious perceptual experiences, we are non-inferentially aware of mental images (that may represent mind-independent particulars), but never of mind-independent particulars.1 Content Non-Conceptualism. Conscious perceptual experiences do not have the same kind of content as beliefs, thoughts, etc. Ganeri argues that Buddhaghosa rejects both theses. This clearly is something that should lead us to favour Buddhaghosa's views. Classical Representationalism cuts us off from the external world, while Content Non-Conceptualism makes it difficult to explain how conscious perceptual experiences can justify beliefs or can be described using propositional attitude ascriptions. My worry: it's unclear how these attractive features of Buddhaghosa's view can be reconciled with his metaphysical commitments, especially those that allow him to preserve a robust form of the "no-self" thesis. The defenders of Theravāda Abhidhamma think: if there is an ultimately existent self, it must be an inner constituent of a person over and above the psycho-physical elements (the aggregates or khandha) that constitute the person. This inner self is both a subject and an agent: it undergoes conscious experiences and thoughts at different times, and exercises executive control in intentional action. The aim of the Theravādins is to defend an insight offered by the Buddha in his discourses: namely, when we analyse the person, we find no constituent that plays the roles attributed to this inner self (Visuddhimagga XVIII.28-31 in Kosambi 1940, p. 419-20). All we find are bundles of short-lived psycho-physical elements (dhamma, or tropes in Ganeri's interpretation) that come into existence and quickly disappear. The Primacy Thesis. Our best metaphysical theories should constrain our theories in epistemology and philosophy of mind, i.e., whenever there is a conflict between one of our best metaphysical theories and a proposed theory in epistemology or philosophy of mind, we should reject the latter. At least sometimes, it seems that the Theravādins reject the Primacy Thesis. But if Buddhaghosa takes that option, then his rejection of Classical Representationalism and Content Non-Conceptualism will begin to look arbitrary. For, if the Sautrāntika and Yogācāra arguments are valid, then Buddhaghosa can only preserve the attractive features of his account by treating mind and matter differently and by positing common characteristics or universals that can be instantiated by the psycho-physical elements that constitute the world. Both these options seem costly from Buddhaghosa's perspective. Thus, there is a tension between Buddhaghosa's theory of conscious experience and the rest of his metaphysics. Begin with Buddhaghosa's rejection of Classical Representationalism. This is laid out in what Ganeri calls his philosophy of cognition, i.e., his explanation of how conscious perceptual experiences arise (Ganeri 2017, chapters 7-10). For simplicity, focus on the case of vision. Visual processing, on this picture, is divided into three stages: the reception of an object (sampaṭicchana), the investigation of that object (santīraṇa), and the determination (votthapana) of that object. In all this, attention plays the role of a gatekeeper, at first determining what gets perceptually acknowledged and later determining what gets investigated and categorised (ibid., p. 181ff). To borrow Buddhaghosa's example, suppose I am looking at a coin. The second stage involves investigating the stimulus so as to distinguish further properties—e.g., spatial properties like flatness or squareness—of the coin. The final stage involves categorising or determining the object, e.g., as a coin. After this, there is a final stage of post-perceptual processing—what Buddhaghosa calls running (javana)—where outputs of perceptual processing are made available for use in reasoning, verbal reports and physical action. It is at this stage that the agent undergoes a phenomenally conscious experience as of a coin before her. On this picture, the object that is thus registered, investigated and determined is not a mental image, but rather the material object itself. The Sautrāntika philosophers, like Vasubandhu, deny this.2 For them, perceptual processing analogously begins with the contact of the agent's sense-faculty with a material object, which produces an initial visual awareness (cakṣur-vijñāṇa) where a mental image that resembles the relevant material object appears to the agent. Then, there is a further stage of higher-level perceptual processing, where the agent retrieves the same (or similar) mental image (ākāra) from memory, infers the existence of a mind-independent particular, and discriminates or categorises it. It is at this stage that what we take to be conscious perceptual experience—here, called mental consciousness (mano-vijñāna)—actually arises. While the two processes—described by Buddhaghosa and the Sautrāntika philosophers—are roughly isomorphic, the second depends importantly on the idea that what appears in the conscious perceptual experience is a mental image. The Doctrine of Momentariness. There are no (ultimately existent) temporally extended entities. The Thesis about the Object of Awareness. Any intentional object (ālambana) of perceptual awareness must both cause the awareness and resemble what appears in it. The Thesis about Causation. If an intentional object of perceptual awareness causes that awareness, then it must be present prior to the perceptual awareness. So, whatever item appears to the subject at these stages can only be a mental image left by the initial sense-object contact. The agent can become aware of, or know about, the relevant mind-independent particular only by inference. If Buddhaghosa rejects Classical Representationalism, which step of this argument will he reject? A plausible answer: he will reject the Doctrine of Momentariness. Defenders of Theravāda Abhidhamma typically think that mental occurrences are just bundles of ontologically simple, impartite, extremely short-lived elements. Interestingly, this is not how the Theravādins think of the material world.3 In The Points of Discussion (Kathāvatthu) XXII.8, we see the Theravādins defending the claim that material objects must persist for longer than mental occurrences on the ground that, if that weren't the case, it would be impossible to account for the perception of the external world (Taylor 1897, p. 620-1). Karunadasa (2015) glosses the passage as follows: "What it seems to imply is that since the Buddhist theory of perception involves a succession of mental events, if a momentary material object impinges on a momentary sense-organ, both will have disappeared by the time the perceptual process is expected to culminate in full perception" (Karunadasa 2015, p. 247-8). This means that the Theravādins reject the Doctrine of Momentariness (as formulated here). This is puzzling. On the most natural way of understanding the dispute, the Theravādins are rejecting a version of the Primacy Thesis: they are saying that if there is a conflict between a full-blown theory of momentariness and the plausible commitment that we are non-inferentially aware of or know about mind-independent particulars, we should reject the former. But if this is right, then the Theravadins' oddly restricted doctrine of momentariness doesn't cohere very well with their other commitments. In fact, as Vasubandhu shows us, the Doctrine of Momentariness just follows from two premises (Pradhan 1975, p. 193). The first premise is that all causally conditioned things, e.g. material objects, eventually perish. The second premise is that we can only explain how an object could be destroyed after persisting for a while by appealing to some internal changes in it, e.g., a change in the configuration of its parts or its other intrinsic characteristics.4 The first premise is common ground amongst all Buddhists, and the second seems plausible. But if the basic constituents of reality are ontologically simple and impartite, they aren't bearers of properties and they don't have parts. So, their destruction couldn't be explained in terms of changes in their intrinsic characteristics or the configuration of their parts. The only options available to a Theravādin like Buddhaghosa are: (a) to deny that the basic constituents of reality are ontologically simple and impartite, (b) to reject the Thesis about the Object of Awareness, and (c) to reject the Thesis about Causation. None of these moves should be appealing from the Theravādin's perspective. The upshot: it's hard to see how Buddhaghosa's rejection of Classical Representationalism can be reconciled with his broader metaphysical commitments. Let's move on to Content Non-Conceptualism, i.e., the claim that conscious perceptual experiences have contents of a different kind from the contents of beliefs, thoughts, etc. Along with his philosophy of cognition, Buddhaghosa also offers what Ganeri calls a philosophy of consciousness, i.e., an account of the structure of conscious experience. On this picture, any conscious mental state (citta) is also always accompanied by certain functional aspects called the concomitants (cetasika). One of these concomitants is identificatory type (saññā). When I perceptually recognise a coin as flat, as white, etc., my perceptual experience involves identificatory types "flat", "white", "coin", etc. which are attached to the perceived object. On Buddhaghosa's view, even conscious experiences of animals that lack linguistic capacities can involve identificatory types of this sort, so identificatory types may not be concepts (at least if we take linguistic capacities to be a necessary condition for concept-possession) (Ganeri 2017, p. 98). The important point, for our purposes, is this: the ability to represent the world using identificatory types makes it possible for us to use the information made available by conscious perceptual experiences for the purposes of reasoning, guiding action, and verbal reports. This means that the contents of conscious perceptual experiences cannot be of a different kind entirely from the contents of the judgements, beliefs, etc. that are involved in these acts. So, Content Non-Conceptualism is false.5 Later Yogācāra Buddhists, e.g., Dignāga (5th-6th century CE) and Dharmakīrti (7th century CE), embraced Content Non-Conceptualism. In the Buddhist epistemological school that began with Dignāga, there are two sources of knowledge: perception and inference. However, the contents of conscious perceptual experiences are importantly different from the content of inferential judgements: unlike contents of inferential judgements, they cannot be characterised by means of coarse-grained representational devices like concepts, which correspond to the common nouns of a natural language. On this picture, the content of conscious perceptual experiences isn't directly available for the purposes of reasoning, guiding action or verbal reports, but only indirectly. For Dharmakīrti and his successors, it is only after we form an initial judgement (adhyavasāya) whereby we categorise what we are perceiving, that we can reason and act on the basis of our perceptual experiences. So, this view drives a wedge between the contents of conscious perceptual experiences and the contents of beliefs, judgements, thoughts, etc. Once again, this idea is motivated by a metaphysical commitment. Before Dignāga, the Abhidharma philosophers thought that the basic constituents of reality could have two kinds of characteristics: own-characteristics (sva-lakṣaṇa) and common-characteristics (sāmānya-lakṣaṇa).6 Since Ganeri (2001, chapter 4) likes thinking of these constituents of reality as ontologically simple, impartite tropes, let's work with a trope-theoretic ontology. Suppose a leaf on a tree is constituted by a particular greenness-trope. The own-characteristic of that trope is the trope's own nature, i.e., something that no other trope has. But a common characteristic of the greenness-trope is something it shares with other tropes, e.g., the characteristic of being a particular shade of green that it shares with other greenness-tropes. Now, implicitly, Buddhaghosa and other Theravāda philosophers may also be working with this same picture: the identificatory types by which we perceptually recognise particulars track common characteristics shared by certain tropes. So, on one way of understanding the view of knowledge that Ganeri attributes to Buddhaghosa, when we non-accidentally represent something using the right identificatory type by attending expertly to its common characteristics, we gain knowledge about properties of the things that we perceive (Ganeri 2017, p. 146ff). In Dignāga and Dharmakīrti, there is a shift away from this picture. For both these philosophers, the basic constituents of reality don't have common characteristics. But inferential judgements necessarily ascribe common characteristics to things; in that respect, they cannot give us access to the structure of reality. If we want to leave open the possibility that there is at least one source of knowledge that gives us access to the structure of reality, then we should want to say that perception (at least sometimes) doesn't involve ascribing common characteristics to perceived particulars. Thus, there is a natural line of reasoning from the claim that there are no genuine common characteristics to the claim that at least some perceptual experiences—which are non-misleading with respect to the structure of reality—don't have content that can be characterised using concepts. The only way for someone like Buddhaghosa to resist this line of argument is to reject the claim that the things we perceive don't have common characteristics. Perhaps, Buddhaghosa could begin by rejecting the Primacy Thesis again: he could say that we don't have to be guided in epistemology or philosophy of mind by our best metaphysical theories. Perhaps, our best theories of intentional content should constrain our judgements about what the structure of reality is like. So, if it seems difficult to deny that perceptual experiences can justify beliefs or can be verbally described like other propositional attitudes like beliefs, then we should just accept the claim that perceptual experiences involve ascribing common characteristics that our concepts can latch on to. Buddhaghosa and other Theravādins will face a challenge here. The challenge is most clearly articulated by Dharmakīrti in his Commentary on Epistemology (Pramāṇavārttika) III.41-47 (Franco and Notake 2014, p. 116-126). Once again, assume that the basic constituents of reality are ontologically simple, impartite elements. How can such elements have common characteristics? There are two options: for any common characteristic that an element of reality may have, we can either say that it is identical to its own-characteristic, or that it is distinct from its own-characteristic. Now, if the common characteristic is identical to the element's own-characteristic, then it cannot be something that element shares with other elements. So, consider the second option. There are two worries. First, if an element of reality had both its own-characterisitic and a common characteristic, it would have two characteristics, and could no longer be ontologically simple or impartite. On a trope-theoretic ontology, the point is just that if the common characterstic of a trope is just a further trope that it contains, then the relevant trope cannot be impartite. Second, if we say that common characteristics that we track are ultimately real, then the challenge will be to explain how a common characteristic, which is shared by multiple elements existing at different times, can nevertheless be related to all of them. The greenness-trope that exists at a time t1 could be of the same shade of green as the greenness-trope that exists at another time t2. That can only happen if the common characteristic (perhaps, a relation-trope) that these greenness-tropes share persists through time. But that contradicts the Doctrine of Momentariness.7 It is reasoning of this sort that leads Dharmakīrti, like his predecessor Dignāga, to the conclusion that, when we group things using a concept, the concept in question doesn't track any objective common characteristics. It only tracks the cognitive tendencies that guide us in our interactions with the world. The upshot: it's unclear how Buddhaghosa's metaphysical commitments can be reconciled with his rejection of Content Non-Conceptualism. I have been arguing that the attractive features of Buddaghosa's account of conscious experience, which Ganeri highlights, are in tension with his other metaphysical commitments. This, I think, raises a challenge that any historical interpreter of Buddhaghosa ought to address: Is there any way of making Buddhaghosa seem more coherent? There is one salient strategy. One could insist that I am misreading Buddhaghosa: when Buddhaghosa analyses the person or conscious mental phenomena into various elements, he is merely offering the way in which the world is presented in conscious experience and thought, without defending any claims about the structure of reality (Ronkin 2005, p. 50). So, it's just a mistake to ask whether Buddaghosa's theory of conscious experience is compatible with his metaphysical commitments. First of all, even if this interpretation of Buddhaghosa is right, it doesn't fit well with Ganeri's interpretation of him. In what Ganeri calls Buddhaghosa's philosophy of cognition, Buddhaghosa's aim is not just to analyse what shows up in conscious experience, but also to lay down an explanation of how conscious experience arises. A part of that story concerns stages of lower-level perceptual processing that we aren't conscious of. Yet, Buddhaghosa wants to claim that we are engaging with the same mind-independent particular through all these stages, orienting our attention towards it, seeking out its spatial boundaries, categorising it. This account, at least, seems to presuppose the persistence of the particular throughout the process. In order to reconcile this story with a background commitment to the momentariness of mental phenomena, Buddhaghosa must defend a seemingly arbitrary view on which only mental entities (or events), but not material ones, are momentary. Second, the "no-self" thesis (as Buddhaghosa explains it in Visuddhimagga XVIII.28-31) is not just a claim about how things appearing in conscious experience should be analysed or understood, but a claim about whether compositionally irreducible selves or persons exist. The challenge facing the Buddhists was one of reconciling this "no-self" thesis with the fact that our conscious experiences and our actions unfold within a perspective occupied by a subject who undergoes those experiences and performs actions. In their attempts to do so, Vasubandhu, Dignāga and Dharmakīrti embraced controversial views like Classical Representationalism and Content Non-Conceptualism. If Ganeri's aim in this book was to show that it is possible to answer this challenge without accepting such views, it remains unclear whether we have a satisfactory answer yet. Here's a fairly general way of putting the point. If we defend the "no-self" thesis by denying the existence of compositionally irreducible selves or persons, then a similar line of reasoning (plausibly) leads us to the conclusion that ordinary material objects that persist through time, or their repeatable properties, are not the basic furniture of the world. If we want to allow the true structure of reality to be epistemically accessible to us through conscious perceptual (or some kind of epistemically direct) experience, then our conscious experiences cannot always be directed at ordinary material objects or their properties. So, the dilemma is this: if we want to reject Classical Representationalism and Content Non-Conceptualism, either we must reject the "no-self" theory or must concede that we don't have any epistemic access to the true structure of reality through conscious epistemically direct experience. Neither horn of this dilemma, I suspect, is something that any Buddhist would be willing to accept. Perhaps, what all this shows is that the attractive bits of Buddhaghosa's picture are things that we can only preserve by jettisoning the background metaphysical commitments he shares with other Buddhists. This, if true, would be an important discovery. For it would raise some questions about what Ganeri describes as cross-cultural philosophy: "A cross-cultural philosophy claims that it is methodologically essential to consider theories from a plurality of cultural locations if one's ambition is to discover a fundamental theory true of the human mind as such, for theories of mind developed exclusively within individual scholarly communities will inevitably be prone to narrowness and provincialism, freighted with vested interests" (Ganeri 2017, p. 341). How much of Buddhaghosa's own views can and should we preserve as we are engaged in such a cross-cultural project as Ganeri's? How far should we be guided by our own preferences—shaped at least partly by recent arguments against Classical Representationalism and Content Non-Conceptualism—in isolating the attractive parts of Buddhaghosa's theory? If we are guided by such preferences, doesn't that undermine the very aim of cross-cultural philosophy, i.e., of overcoming the preferences thrust upon us by our cultural background? Ganeri's conception of philosophy of mind as a cross-cultural project is inspiring, but I am not sure how exactly one can go about successfully pursuing it. I am hoping he will set me straight

Emotions and Moral Behavior · Indian and Buddhist Studies · Phenomenology and Existential Philosophy · Philosophy