I’m posting a term paper that I wrote for a graduate philosophy seminar. The topic is the debate between Jayanta and Kumārila on the theory of moral knowledge. It draws heavily from the section of Jayanta’s Nyāyamañjarī I translated here: I See What I Should Do: Jayanta Bhaṭṭa on Mysticism and Moral Knowledge. There are a few differences in translation choices between this paper and the earler post. I also draw extensively from the Ślōkavārttikam (mainly chapter 4) and more broadly from the Nyāyamañjarī as well. All translations from sanskrit are my own, unless otherwise noted.
In the 2nd chapter of The Flowers of Reason1, the Kaśmīrī philosopher Jayanta Bhaṭṭa (~9th ce2) develops and responds to a contraversy in the philosophy of ritual regarding how we know whether or not a given ritual action is correct. The conception of ritual operative here is broad: ritual refers to any pattern of behaviour that can be brought under the purview of rules3. Our concern, in particular, regards forms of ritualized behaviour where explicit rules play only an evidential role in judgments of correctness, but are not what make that action correct or incorrect. Let us call such rituals, ‘natural rituals’. Natural rituals include actions whose correctness may depend on, e.g., ethical, religious, or aesthetic considerations. Understood in this way, the epistemology of ritual has broad implications for the theory of normativity as it impacts the domains of metaethics, philosophy of religion, and philosophy of art.
Within the discipline of Mīmāṁsā, the field of study specializing in the philosophical investigation of Vēdic ritual4, the consensus view is that we cannot have knowledge of the correctness or incorrectness of a ritual action independently of the testimony of an authority. When applied to natural rituals, it is easy to see why such a view could be troubling. Insofar as testimony plays a merely evidental role in beliefs about correctness, appeals to authority are parasitic on belief in the trustworthiness of that authority. In addition, the authoritative status of the testimony itself depends on the testifier’s having prior knowledge of the facts in virtue of which an action is correct or incorrect. In both regards, there is a threatened regress if epistemic access to facts about correctness always depends on testimony.
Nonetheless, according to Jayanta’s predecessor and principle interlocutor in this debate—the influential Mīmāṁsā thinker, Kumārila Bhaṭṭa (~660 ce5)—the alternative would be that we can have no knowledge, at all, about the correctness or incorrectness of ritual actions. This is because inferring that a certain action is correct requires preexisting knowledge of the norms governing that action. However, perception cannot explain how we acquire this background knowledge, in the first place. Since, the correctness of individual actions, on the basis of which general principles could be abstracted, is not the sort of thing that can be directly percieved.
Drawing on Kumārila’s comments in his Annotations in Verse6 as well as Jayanta’s reading of this same text, I argue that the disagreement between Kumārila and Jayanta on the role of perception in the epistemology of ritual originates from a basic difference in how they understand the relationship between conceptualization and sensation. And, moreover, this difference arises out of a more fundemental disagreement about how we judge that a particular observation counts as evidence for a belief. Specifically, while both Kumārila and Jayanta agree that some forms of conceptual operation on preconceptual sense experience are a part of the perceptual process and do not disqualify a particular perceptual experience from counting as perceptual evidence; Kumārila contra Jayanta discounts as properly perceptual those cases where a concept is imposed on sense experience which does not instantiate that concept.
Kumārila gives the example of a person seeing a fire at a distance and judging it to be hot. Often such judgments proceed reflexively and without the involvement of explicit reasoning, due to the automatic imposition by memory of the sense-concept ‘hot’ onto visual experience. In such a case, Kumārila argues, while the resulting belief that ‘the fire is hot’ is not justified by an explicit inference, it cannot be considered perceptually evidenced either. Since, the imposition of the sense-concept ‘hot’ is not justified by the presence of a corresponding ‘hot’ sensation. In the case of ritual knowledge, when people judge that an action is wrong reflexively, without the involvement of an explicit inference, they impose an ‘ought’ concept onto sensation. However, Kumārila contends, there are no ‘ought’ sensations. Therefore, ritual knowledge can never be perceptual.
Jayanta’s response to this revolves around his understanding of how “insight” (pratibhā) can act as evidence for belief. Kumārila derides insight as a mysterious, even mystical, concept that has no place in a rational conception of human knowledge. In defending the epistemic value of insight, Jayanta draws on a core idea from Kumārila’s own epistemological approach—namely, the question of the “evidentiality” (prāmāṇyam) of a mental state. Kumārila uses the concept of evidentiality to explore why we take certain kinds of mental states (e.g. perceptions and inferences) to have evidential value in grounding beliefs, while denying any epistemic role to other forms of cognition (e.g. hopes and desires).
In discussing Kumārila’s theory of evidentiality, Jayanta distinguishes between an epistemic problem (how we know that a mental state is evidential) and a metaphysical problem (what makes a mental state evidential). For Kumārila, the two problems are inextricably intertwined: the epistemic question constrains what kinds of answers we can give to the metaphysical question; and, the answer to the metaphysical question determines the answer to the epistemic one. The key move Jayanta makes in the debate is severing this codependence between the epistemic and metaphysical questions of evidentiality. So, whereas Jayanta accepts—even embraces—the mystical nature of insight, at the metaphysical level; he emphasizes that there is nothing at all mysterious about its epistemology. Insight is just perception. Its epistemology is nothing but the epistemology of mundane perceptual experience.
In the following sections, I will first discuss Kumārila’s critique of inferentialism in the epistemology of ritual. Following this, I will discuss the common core theory of conceptualized perception, shared by Kumārila and Jayanta. I will then explain Jayanta’s notion of insight. Next, I will develop Kumārila critique of insight, drawing on his account of conceptualization. Finally, I will show how Jayanta is able to answer these criticisms by drawing on comments he makes both in his explicit response to Kumārila as well as aspects of his general epistemology and theory of perception.
The locus classicus for Kumārila’s argument against the inferrability of ritual correctness occurs in the 4th chapter of the Annotations in Verse. The argument is short and worth quoting in full:
Inference is possible only after the reason, etc, have already been grasped through perception. This does not obtain in the case of the property of correctness. Nor can we ground inference in inference. For, in this case, we could not account for the necessary prior knowledge of reason, target, and relation7
The idea is as follows. Kumārila understands inference as involving a target proposition being inferred on the basis of a reason. In his account, for someone to infer the target from the reason, they must have prior knowledge of a general relationship connecting the predicates of the reason [liṅgam] and target [liṅgi] propositions, respectively8. In the context of ritual knowledge, this means knowing that certain properties of an action entail that the action is correct or incorrect. This entailment relation cannot be perceived, assuming perceptualism is false. And, it cannot be inferred because this would launch a regress.
Kumārila continues:
Inference does not disclose what something is. So, if we were to take correctness [known only by inference] to predicate of something, then the target of the inference would have no definite character9.
Inferences track relations between concepts and allow for movement from one concept to another. However, inference cannot, itself, establish that a given concept represents anything real or reflects any matters of fact. Take the well known Kantian argument against the permissability of false promises.
[W]ould I be able to say to myself: everyone may make an untruthful promise when he finds himself in a predicament from which he can extricate himself in no other way? Then I soon become aware that I could indeed will the lie, but by no means a universal law to lie; for according to such a law there would actually be no promise at all since it would be futile to pretend my will to others with regard to my future actions10
Here, we move from a certain conception of what a promise is, as something that engenders confidence in others regarding one’s stated intent, to a conclusion about what kinds of acts of promising are correct or incorrect. If the ritual of promising were such that lying were permissible; there would be no expectation that a promise would be kept—regardless of one’s belief in the correctness of the promising act or the virtue of the promiser. Such a ritual form, where lying is not incorrect, would fail to be a ritual of promising, since it would fail to satisfy its defining purpose.
Kumārila’s point, however, is that this sort of argumentation tells us nothing about what promising is, in definite terms11. This is because the concept of promising includes, at the onset, notions about what it means for that action to be correct or incorrect. So, confronted with an act of verbalization, gesticulation, etc.; we cannot judge that this is an act of promising unless we can know what it would be for that act to be correct or incorrect. Purely sensory aspects of the behaviour—the verbalized sounds, the gesticulated shapes—do not, themselves, necessitate that this be a form of promising.
The point of Kumārila’s argument against inferentialism is not to claim that inference plays no role in ritual knowledge. Rather, what we have is a bootstrapping problem. Once certain core concepts and conceptual relations are available, further knowledge can be obtained through reason. But how do we acquire concepts, in the first place? If it cannot come from inference, perhaps it comes from perception. That is, we can acquire concepts and basic conceptual relations by abstracting from objects which we encounter perceptually. As Kumārila puts it:
As for the claim that reasons, etc, are not grasped perceptually, since the latter is nonconceptual: we reject this claim. For, we hold that conceptual cognitions, too, may be perceptual insofar as they are necessary for grasping the nature of an object12…First, we have a form of awareness that is mere experiencing and is nonconceptual. Like the cognitions of infants and the nonverbal, it relates to the pure object. Here, there is no awareness either of properties or of universals, but only of the concrete entity which is their basis13…Subsequently, when we grasp this same object in terms of properties and universals, this cognitive state must also be considered perceptual. For, it too depends on the sense organs14.
This process of “grasping the object in terms of properties and universals” involves memory and a faculty of linguistic abstraction. Nonetheless, because of its capacity to introduce new information, it cannot simply be identified with recollection or considered as something constructed by the mind. As Kumārila argues:
Inference and testimony would also become nonevidential, if [conceptualization] were the mere imposition [of a concept on an object]. Even in cases of objects known only through language, although, in its absence, there would be no awareness of the object; the nature of the object would not vanish. Without vision, color could not be grasped. But, we would not, thereby, conclude that the basis for color would cease to exist15.
And he continues, a bit further down in the passage:
When a person recalls the relationship between an object and a word [when seeing a cow], the resulting cognition does not fail to be perceptual, even though it involves the synthesis of prior sensory encounters. It is true that since the word and relation are recollected, their awareness could not be considered perceptual. However, this does not preclude the cognition of the object from being perceptual. For, even though our grasping of its ‘cowness’ is touched with memory, since it appears as distinct from prior cognitions due to its being marked by the object’s particular identity and temporality, it must be considered evidential16.
When a person observes some particular creature at a distance and thinks, “it’s a cow!” their judgment involves the deployment of a concept ‘cow’ onto the percieved particular by way of a synthesis of prior particular encounters and the abstraction of these encounters into a (proto)linguistic sign. Nonetheless, if the application of a concept to an object were nothing but a mental imposition or psychosocial construct, then any subsequent reasoning or expression involving these concepts would also end up as mere conceptual constructions.
The analogy to dependence on the visual faculty for awareness of color is illustrative here: that our linguistic faculty plays a necessary role in the conceptual determination of an object is not itself problematic. The fact that we need symbols to conceptualize does not imply that the basis for conceptualization is something unreal. Jayanta goes further, arguing that there must be some sort of implicit structuring even at the nonconceptual level as a condition of possibility for conceptualization:
Where only an isolated particular is grasped nonconceptually, how would we explain the spontaneous elaboration of concepts? Conceptualization can only happen in dependence on what is present nonconceptually17.
Conceptualization and abstraction are pervasive in our epistemic and cognitive lives. Moreover, this abstractive process cannot be, itself, a faculty of reason. Since, reasons, as such, play no role. As Kumārila argues:
This mental state is not inferential since it does not involve having reasons. And, insofar as there are no defeaters present, there is no basis for considering it to be nonevidential. And since it is not the reproduction of a prior observation, it is not recollection. So, it must be perception18.
This is the so-called ‘cognitive perception’ (mānasapratyakṣam) which Jayanta identifies with our capacity for ‘insight’ (pratibhā):
When we form a truthful judgment about some object without the explicit involvement of language, this is called cognitive perception. Examples include judgments of the form “kētakī flowers are fragrant,” and, “sugar is sweet.” There is no basis for complaining that insight is a form of cognition lacking causal regularity, since it obeys the same causal conditions as perception19.
Let us consider where things stand, at this point. Both Kumārila and Jayanta agree that inference depends on conceptualization as a condition of possibility. And both agree that cognitive perception, where concepts are imposed on experience via abstraction from individual sensory encounters, plays this role as a distinct faculty for the formation of judgments and the provision of evidence.
Once we accept conceptualization and cognitive operations on sense experience into the realm of perceptual evidence, what grounds do we have for restricting the scope of perceptibility to just the sensible properties of actions? What is the basis for Kumārila’s allergic response to the idea of insight?
Jayanta summarizes Kumārila’s objection to the sensibility of ritual correctness in the following terms:
Ritual has the form of an ‘ought’ and is utterly unconnected to the three temporal divisions. Surely it’s outrageous to say it could ever be a visual object20?
Whatever it is in virtue of which ‘ought’ statements are true, they cannot be sensible properties such as what are disclosed in our visual experience of an action. We see bodily movements as they ‘are’ not as they ‘ought to be’. However, once we aknowledge that the deployment of concepts over a percieved object or event through the intermediation of our cognitive faculty also counts as perceptual evidence for the beliefs formed thereby, then it is less clear what the force of the above objection is.
As Jayanta puts it:
There is still the case of awareness derived from the cognitive faculty, such as we see arising from meditative practice, which could apprehend ritual correctness in moments of intense focus, like a lover thinking of their beloved. And the mind can take anything as its object. There is nothing at all that lies outside its scope. Through practice, even insensible objects can appear as lucid presentations. As [Dharmakirti21] once said: “Tormented by insane nightmares, filled with sadness and longing, people see things that never existed as if standing before their eyes22.”
The kind of intuition Jayanta appears to have in mind here is unlike that of, .e.g, the basic geometric intuitions through which the Euclidian axioms are seen to be self-evident. Rather, it is the intuitions of “experts” who can detect facts or objects, not on the basis of special features of those facts or those objects, but due to special features of their trained cognitive-perceptual faculties. In the case of ritual facts these would be experts in ritual matters, i.e. “sages” (yōginaḥ):
(Objection) Then, answer this: what evidence do you have for the existence of this “sage perception” that is supposed to have access to normative properties?
(Response) Here’s what we say: the evidence lies in the fact that perceptual faculties can be improved upon23.
…(Objection) Surely, even with practice, you do not see endless or unprecedented improvement. Consider jumping practice: when someone spends every day single mindedly practicing jumping, they would, still, only be able to jump a few extra feet above the ground; not leap over mountains or oceans.
(Response) To this, we say:
Jumping does not attain such excellence because the body is made up of material constituents like phlegm, etc. But, what is stopping the mind? If, in the preceding days, someone practices jumping and the like; in the following days, they do not experience a fundamental change in their body. Rather, through the reduction of fat and phlegm, the body becomes leaner and they can jump around as they please. In the case of the mind, however, the cause of cognitive excellence is in the gradual accumulation of the sorts of psychological dispositions that are conducive to the arising of insight24.
Consider, again, the analogy to recognizing that a particular creature is a ‘cow’. Repeated encounters with similar creatures in the past, together with an innate faculty of synthesis and symbolic representation, allows the experienced observer to see the creature as a cow, rather than as some nondistinct entity. In a more elaborate case, the trained eye of a horticulturalist may be able to detect features of, e.g., a plant’s health that are invisible to the untrained layperson. This is what Jayanta means by ‘insight’. In the same way, the ritual expert may come to possess perceptual capacities unavailable to ordinary people, through which they can detect normative features of actions:
Consider: when a student starts doing concentration exercises and reading practice, certain dispositions become impressed upon their psyche and manifest as skill in recitation and recall. Or, when gold is fired in a furnace, it ever so gradually becomes purified and reaches a state of unparalleled refinement. In the same way, through meditative practice, the sage’s mental faculty becomes fit to directly access all objects of insight25.
Nonetheless, there is a problem to this story, already implicit in Jayanta’s example quoted from Dharmakīrti regarding experiences coming from sadness and longing. As Jayanta has an objecter point out in the immediately following sentence:
Come on! These are all examples of erroneous mental states. You cannot use them as analogies for a sage’s perception26.
The problem is as follows. It is true that certain forms of cognitive operation on sensory experience are part of the perceptual process and count as evidence. Nonetheless, it is not the case that all such cognitive processes are evidential. Jayanta wants to draw a distinction between those forms of cognitive processing that arise from insight and those that arise from, e.g., “sadness and longing”. The forelorn lover’s delusions do not count as evidence for their beliefs. Why then should the supposed insight of “sages”?
Kumārila puts the point in more explicit terms:
Some say that there is a form of perception available to sages and enlightened souls by which they can see the past and future, the small, and the far away. And they suggest that the phrase “it grasps only what is present” does not apply, in this case…Cognitions that pertain to non-present objects, as in the case of desire and recollection, cannot be considered perceptual since this is not corroborated by our experience of the world. The same applies to the concept of ‘insight’, which is likewise ruled out by reference to “grasping what is present”. The insight of ordinary people does not depend on sense experience and is, therefore, insufficient for knowledge. The same applies to the insight of sages27
In other words, there is a demarcation problem at the heart of Jayanta’s theory of insight. How do we judge which forms of conceptualization count as evidence, and which do not?
To this end, let us briefly discuss Kumārila’s general account of how we know what forms of cognition count as evidence.
We can understand evidentiality (prāmāṇyam) as a property of a mental state such that a person may be said to have justification for some belief in virtue of posessing that mental state. An analysis of evidentiality must serve two purposes. It must explain what makes a cognition evidential and it must explain how this evidentiality is known. These two explanatory roles co-constrain the form the analysis of evidentiality can take: whatever properties feature in this analysis must be (1) plausibly attributable to cognition and (2) knowable by the cognizer.
The properties that feature in an analysis of evidentiality can be characterized as ‘internal’ (svataḥ) or ‘external’ (parataḥ); where a property is considered internal if it is can be known by introspection alone28. Internal properties include such things as a cognition having propositional content and involving an attitude of belief or doubt. The property of being sensitive to or tracking with facts would be an external property, since it would be grounded in, e.g., causal properties of cognition.
The thesis of evidential internalism (svataḥprāmāṇyavādaḥ) states that:
All properties that feature in the analysis of evidentiality are internal properties.
Kumārila argues, in a widely discussed passage29, that rejecting evidential internalism results in a regress; because, determining that some external property obtains for a cognition requires possessing additional evidence. If the evidentiality of this evidence is not ascertained, then you would need further evidence to determine this, and so on.
Kumārila’s answer is that evidentiality applies as a default status to all cognitions obeying a minimal criteria that can be immediately and indubitably verified by introspection. Namely, that the mental state has an object, predicates a property of that object, and does so with an attitude of commitment rather than doubt. Kumārila then suggests that nonevidentiality is seperately ascertained by reference to some defeat condition, which can involve external properties. In this case, since the observation of nonevidentiality has its evidentiality by default, there is no regress.
What this amounts to is that, for Kumārila, evidential internalism acts as a bootstrapping mechanism to identify certain kinds of evidence as foundational. The phenomenology of certain forms of experience is such that they grip us more strongly and demand our acceptance of them over and above others30. It is these phenomenological properties, rather than some story about self-evident propositions or direct acquanitance with sense data, that explains why, e.g., valid inferences and sense perception end up as evidential primitives.
The upshot of Kumārila’s account of evidentiality is that evidentiality is a default but brittle status. Prior to any knowledge about the origins of some cognitive phenomenon and their relations to their intensional objects, we default to belief in their evidentiality. However, any further knowledge about them threatens to cancel this default status. In the case of conceptualization, the fact that all forms of conceptualization share a common origin in our cognitive faculties threatens to cast the whole lot into disrepute on the basis of the known unreliability of a few. Kumarila argues that a certain class of cognitive processes can be carved out of this and rescued by virtue of their causal and structural dependence on sensation.
When the ears are destroyed, there can be memory of sound; but, present sound cannot be cognized. So, we conclude that there must be some underlying structure. If sensation were undifferentiated, either everything would be perceptible or nothing. But, because we see a difference in perceptibility between objects, we must recognize different sensory capacities. Otherwise, a deaf person could perceive sounds with, e.g., their eyes. Or, the mind would have independent access to external objects31.
Nonetheless, for Kumārila, this procedure does not extend to cases of conceptualization where concepts are imposed on sense experience such that the basis for conceptualization does not feature in the experience as sensation. He takes the example of a person seeing fire in a distance. Such a person may have a natural and immediate grasping of the fire as ‘hot’, without recourse to explicit reasoning. This is possible due to the capacity of memory to impose sense-concepts over sense experiences that, though not instances of that concept, are still related to it in some way. However, Kumārila argues, this form of conceptualization is in fact not evidential:
(Objection) Surely, if this were the case; then, even the conceptualization of a visually presented fire as ‘hot’ would count as perceptual, just like cognitions involving ‘cowness’ and the like?…
(Response) In the case of ‘cowness’, there is no preceeding cognition already classed as perception. Moreover, in the case [of ‘hot’], the concept is explicitly presented as nonvisual. So, the cognition cannot be considered perceptual32.
The same sort of issue, argues Kumārila, extends to the concept of ‘insight’:
Even for sages, perception cannot be something extraordinary. Their perception must, like our own perception, grasp just what is present; since, it originates from sensory operation over a present object.
So far, I have argued that Kumārila’s rejection of insight as a form of evidence for belief traces back to certain of his commitments in the foundations of epistemology regarding the nature of evidentiality. Specifically, Kumārila’s defense of evidential internalism results in a spartan epistemology where there are a small class of foundational evidential primitives and any cognitive activity that fails to fit into the template set out by these primitives is denied status as evidence due to a blanket skepticism applied to human cognition. This yields a solution to the demarcation problem for conceptualization such that only those cases where a sensed property is identified as an instance of a sense-concept count as forms of perceptual evidence. Any other form of conceptualization is denied this status simply on the basis of its being disconnected in a narrow sense from sensation.
The corrolary to this is that Jayanta’s defense of insight as an evidential form of cognition likewise rests on his understanding of the nature of evidentiality, in general. The key move Jayanta makes is to push back against Kumārila’s initial formulation of the question of evidentiality. Recall, that an analysis of evidentiality is supposed to simulaneously serve two purposes: it must explain what makes a cognition evidential and how we know that it is evidential. In other words, the principle underlying evidential internalism is that one and the same property determines whether a mental state is evidential and forms the justification for beliefs regarding the evidentiality of that mental state.
While there is a certain intuitive appeal to this principle, it comes at a cost. Namely, the properties that explain the evidentiality of an evidential mental state are no longer the same as those that explain the nonevidentiality of a nonevidential mental state. An evidential mental state is evidential simply due to its being informative. However, a nonevidential mental state is nonevidential due to faults in its causal origins. It is this very asymmetry that blocks Kumārila’s regress of evidence.
For Jayanta, this cost is too great:
No causal process operates without having success and failure conditions; therefore, there is no third type of effect except ‘success’ and ‘failure’…Medical science supports the successful operation of our sense organs. When a doctor prescribes some medicine to improve health, this involves promoting success conditions, not merely in eliminating failure conditions…The claim that evidentiality is internal and does not depend on external success conditions for its obtaining is unacceptable33.
The sense of the initial argument regarding “no third type of effect” is somewhat obscure, but the idea is as follows34. If, in the operation of some sense faculty, failure could only be explained in terms of the presence of some failure condition and the absence of a success condition plays no role; then, there would be nothing distinctive about ‘successful’ operation as such. Since, the constituents of succesful operation would form the common core shared by both succesful and unsuccesful cases. Then, there would be nothing that explains why the operation is succesful. This would leave open the possibility of sensory operation where nothing specifically connects the output to the target property being sensed; but, due to a lack of any specific failure criteria, the outcome could not be considered failure. Such a result would be nominally considered ‘success’, but practically speaking could not be considered successful, either.
To avoid this, argues Jayanta, it is necessary to explain both success and failure in terms of a combination of both explicit success conditions and explicit failure conditions. So, even when no explicit failure criteria is met, the absence of any success conditions would count as an implicit failure condition. The example of medical treatment is given as further illustration of this idea.
The upshot of this is that retaining the intuition that evidentiality and nonevidentiality must be explainable, jointly, in terms of the same set of conditions requires giving up on the intuition that the same properties serve to fix the set of evidential mental states as serve to justify beliefs regarding evidentiality. This is the basic principle underlying what Kumārila criticizes under evidential externalism (parataḥprāmāṇyam).
Jayanta’s point is that giving up on the latter intuition—that the ontic and epistemic problems of evidentiality submit to the same explanation—also avoids Kumārila’s regress. This is because the regress only occurs at the epistemic level. Namely, when the evidentiality of a mental state is doubted, the doubt can only be lifted with reference to another evidential mental state. But, if the evidentiality of the latter mental state is also in doubt, then it cannot serve the purpose of evidencing the former.
In response, Jayanta argues that there is a basic asymmetry between the ontic and epistemic problems of evidentiality because the epistemic question, unlike the ontic one, is a fundementally pragmatic one. In general, we make observations and act on their basis without engaging in second-order reflection about whether or not these observations count as genuine articles of evidence at the ontic level.
When we use a thermometer to check our temperature, we assume that it works unless we have positive reason to think otherwise. This is because rationality is fundementally oriented towards success in action. When we engage in second-order reflection about evidentiality, we do so because we have doubts about whether the actions that stem from these cognitions will lead to success. The corrolary to this is that belief in evidentiality can be vindicated by the subsequent experience of success—what Jayanta refers to as the “satisfaction of intent” (pravr̥ttisāmārthyam):
What exactly is ‘satisfaction of intent’, which students of Nyaya claim as the basis for knowledge of evidentiality? When a latter cognition corroborates or specifies a former cognition, we call this the satisfaction of intent…Satisfaction of intent can also be described as the experience of success in action.35
And our experience of success, unlike instrumental forms of experience, are not the sorts of things regarding which questions about evidentiality arise. As Jayanta explains:
(Objection) But, what is so special about the experience of success? Wouldn’t determining its evidentiality also depend on external evidence; therefore, resulting in a regress?
(Response) Cognition that motivates action, because it does not always result in success, is such that its evidentiality can be investigated. However, cognition of success, insofar as it itself amounts to the attainment of one’s purpose, is not the sort of thing for which evidentiality submits to investigation. This being so, where is the regress36?
Before we discuss how Jayanta ultimately works out his account of the evidentiality of insight, it will helpful to consider a caveat to the above discussion.
As we discussed earlier, the first-person account that we give ourselves when trying to understand why we take certain mental states to count as evidence for beliefs does not require us to take into consideration details of, e.g., how the physiological mechanisms underlying those states operate. But, the third-person account of what makes evidence evidential does need to reference actual states of affairs and causal relations. After all, the purpose of theory, at this level, is explanation as such—not merely a pragmatic concern for how to make decisions in the course of normal business. Nonetheless, the ontic question feeds into the epistemic one because it provides a useful framework for adjudicating questions about the evidentiality of specific mental states or processes.
Again, the example of medical science viz. a viz. visual perception is instructive, here. At bottom, we accept visual perception as evidential because we experience success in action where we follow the guidance our eyes provide. Nonetheless, it is useful to have some understanding of how eyes work and what factors contribute to success or failure when we encounter doubt in particular circumstances. This is because, the causes of success and failure can be more difficult to ascertain in individual cases as compared to a class of mental states as a whole. Having a theory of visual perception involving physiologic explanations of the functioning of the visual organs allows us to make more fine grained decisions about the reliability of an individual’s visual facutlies. In addition, such theoretical knowledge allows us to draw distinctions between, for example, beliefs about color versus shape: Knowledge of differences in color perception across species might cause us to doubt that color beliefs have the same grip on the optical properties of external objects as shape beliefs.
An analogous situation applies to questions about cognitive perception. Recall the core issue at hand: how do we demarcate between cases of conceptualization that count as ‘insight’ from those that do not. In this regard, a detailed theory involving a fully worked out metaphysics of morals is not necessary. All that is necessary is a framework that is vindicated, at a general level, by the criterion of ‘satisfaction of intent’ and which can be leveraged in particular instances to demarcate between insight and non-insight.
Jayanta’s idea is that the cases of, e.g., delusions emerging out of “longing and sadness” or Kumārila’s examples of “desire and recollection” fail to be evidential not merely because they are nonsensory forms of conceptualization but because they face explicit defeat conditions:
(Objection) Come on! These are all examples of erroneous mental states. You cannot use them as analogies for a sage’s perception.
(Response) No, the analogy being drawn here is just that both are experienced in a lucid way…Here, our examples of meditative states arising from sadness and longing fail to be evidential specifically because they are struck down by defeaters. But, since this is not the case with perceiving correctness, the basis for denying its evidentiality does not obtain37.
In other words, part of the success condition for insight is that it has a certain lucidity. In addition, Jayanta insists that insight is connected to its object such that while this connection is not strictly causal it does create the expectation that insight tracks with the features of its object:
What’s more, even people like us can sometimes form judgments about future events such as, “my brother is coming home tomorrow,” on the basis of an evidential process called ‘insight’. And, in so far as it is not (1) unconnected to the object of judgment, (2) doubted, (3) struck down by defeaters, or (4) the product of a faulty causal process; it must be accepted as evidence.
If some defeat condition does obtain; then, sure, it would fail to be evidential. But, if the brother does show up on the following day; what would you say to that? If you say, “it is just a coincidence.” Well, insofar as something is grasped via an evidential process, you cannot dismiss it as coincidence.
(Objection) Oh, come on. That judgment is not causally connected to its object.
(Response)…It’s like this: somehow or other, e.g. because you are acquainted with [your brother’s] pattern of gettting hungry, he suddenly comes to mind and you get a sense that he will come visit tomorrow. So, he is the cause of the insight. And, since it is not the case that the insight is unconnected to its object, it can be evidential with regards to him38.
Note here that Jayanta uses a case where success can be clearly adjudicated in order to establish the prima facie acceptibility of insight as evidence. He can then extend this to cases where success conditions are less clear—such as moral insight—by arguing that they fall under the same category. There is, again, the analogy to visual perception. Here, we establish the reliability of vision with regards to things that can be crosschecked and then apply it to objects, like the moon and stars, where such crosschecking may not be possible.
If these are the success conditions, then what are the failure conditions? Here also, Jayanta offers a general account:
For ordinary people like us, the mind is occluded by psychological impediments like partiality and prejudice; and, so, fails to reach this state of cognitive excellence. But, as these impediments are eroded through the daily practice of meditation, and the mind of the sage is made pure; what, then, could they fail to see? When partiality and prejudice are finally destroyed, the sage is able to fully open their minds and reflect on the highest good. Then, with the elimination of all such cognitive faults, their mind fixed in meditation, the sage is able to clearly grasp all things39.
The problem, in other words, is that cases of uninsightful judgment—such as delusions grounded in longing, or descrimination based on race and gender, etc.—fail to be evidential, despite being lucid or involving reflection on prior experience, because they are tainted by prejudice and partiality. And, insofar as these two are causes of deviation from the truth, they are sources of epistemic failure.
To review, I have argued that Jayanta and Kumārila share a conception of ritual insight as a form of conceptual cognition in which an individual forms judgements involving ‘ought-concepts’ without the explicit involvement of inferences or authoritative testimony as sources of evidence. They differ in that Kumārila, unlike Jayanta, does not consider insight to be a form of evidence. This is because, insofar as ‘ought’-concepts do not correspond to any sensible properties, there is no principled way to separate insight from forms of conceptualization, such as wishful thinking, which are known to be nonevidential.
I have further argued that the crux of the debate lies in a difference between what Jayanta and Kumārila take to be the basis for knowledge about the evidentiality of a cognition. Kumārila insists that knowledge about evidentiality must be grounded in knowledge about those properties of cognition in virtue of which that cognition is or is not evidential. Kumārila draws a surprising conclusion from this intuitive principle: the properties that ground evidentiality are distinct from those that ground nonevidentiality.
Jayanta takes this conclusion to be a reductio on the original principle. In so doing, he defends a pragmatic account of our knowledge of evidentiality. Under his account, knowledge of the evidentiality of a cognition is grounded in the capacity of that cognition to “satisfy the intent” of the cognizer. Jayanta argues that it is rational to accept insight as a form of perceptual evidence just insofar as our taking insight to yield knowledge results in a pattern of action that harmonizes with our intentions and with other cogntions across time.
It is worth spending a few words to consider what is at stake, for Jayanta, in defending his account of insight. For Jayanta, rejecting the evidentiality of insight is tantamount to abdicating all human agency in the acquisition of moral knowledge. Under Kumārila’s account, the only source of knowledge for moral truths is an eternal, unbroken, unchanging tradition; since there is no rational basis for forming ‘ought’ judgements outside of what is contained in the received tradition, excepting merely deductive inferences drawn from propositions contained therein. However, this means there is neither any scope for external critique nor, as Jayanta is especially concerned to point out, is there any way to negotiate differences of opinion in a religously and intellectually diverse society. As Jayanta argues40:
If you claim that the criticism by Br̥haspati regarding the authority of the Vēdic tradition can be rebutted by appealing to the tradition’s acceptance by exemplary people. Well, the same can be said about alternative traditions. After all, there are smart people who accept these traditions as well. Whatever arguments you use to establish the authority of the Vēdic tradition; the same apply to the other traditions as well41.
One possibility is just to take a syncretic, universalist approach:
A tradition may be considered authoritative because it is transmitted by truth-speaking experts. Or, as with the Vēda, intrinscially due to its eternality. Or, as with the Laws of Manu, because it concords with the Vedic tradition. Regardless, the wise hold all traditions to be authoritative42.
But, while Jayanta is sympathetic to this strategy, he sees a problem where there is no space to talk of geniune differences of opnion at all. If there is no distinction, at all, between correctness and incorrectness; Jayanta argues, then “ritual itself would implode”:
Now, some say: But, because such reasoning is underconstrained, permissibility would become indeterminate and ritual itself would implode. After all, with such an “anything goes” approach, are there any practices at all to which this sort of reasoning could not apply43?
The key, for Jayanta, is that it must be possible to evaluate a tradition’s substantive claims as well as its history and sociopolitical implications. A tradition that cannot be evaluated cannot be validated:
When it has a widely known and unbroken history of transmission. When respectible people do not turn away in disgust on hearing it discussed. When the practices it recommends are neither unwholesome nor antithetical to society. When it did not just spring up out of nowhere. When it does not appear to be based in insane ramblings, personal ambition, or something disconnected from lived experience. Only then can a tradition can be validated in the above manner. Not, just anything whatsoever44.
However, in an epistemology of ritual where only testimony can serve as evidence for judgments of correctness, there is no scope for critiquing a tradition on “the practices it recommends”. In such a case, the problem of ritual pluralism and the underdetermination of ritual knowledge by tradition becomes insoluble.
Arnold, Daniel, “Kumārila”, The Stanford Encyclopedia of Philosophy (Summer 2024 Edition), Edward N. Zalta & Uri Nodelman (eds.), URL = https://plato.stanford.edu/archives/sum2024/entries/kumaarila/
Arnold, Dan. Buddhists, brahmins, and belief: Epistemology in South Asian philosophy of religion. Columbia University Press, 2008.
Bhaṭṭa, Jayanta. Much ado about religion. NYU Press, 2005.
Bhatta, Jayanta. Nyāyamañjari of Jayantabhatta with Tippaṇi‒Nyāyasaurabha Vol. 1. Ed. Vidvan, KS Varadacharya.(Oriental Research Institute Series No. 116) Mysore: Oriental Research Institute (1969).
Bhatta, Kumarila. The Mimansa-Sloka-Vartika. Edited by Rama Sastri Tailinga. Benares: Chowkhamba Sanskrit Book-Depot, 1898.
Gangesa. (2020). Jewel of Reflection on the Truth about Epistemology (S. Phillips, Trans.). Bloomsbury.
Gillon, Brendan, “Logic in Classical Indian Philosophy”, The Stanford Encyclopedia of Philosophy (Spring 2024 Edition), Edward N. Zalta & Uri Nodelman (eds.), URL = https://plato.stanford.edu/archives/spr2024/entries/logic-india/.
Jaimini. (1979). Purva-Mimamsa-Sutras with an Original English Commentary. Bharatiya Publishing House. https://archive.org/details/dli.ernet.170066/page/n3/mode/2up
Kant, Immanuel, Jens Timmermann, Mary J Gregor, and Ebrary. Groundwork of the Metaphysics of Morals a German-English Edition / Immanuel Kant ; German Text from the Second Original Edition (1786) Edited by Jens Timmermann ; English Translation by Mary Gregor, Revised by Jens Timmermann. German-English edition. Cambridge ; Cambridge University Press, 2011.
Potter, Karl H. “Janaki Vallabha Bhattacharya,” Jayanta Bhatta’s Nyaya-Manjari (The Compendium of Indian Speculative Logic)”(Book Review).” Philosophy East and West 31, no. 2 (1981): 239.
Taber, John. “What Did Kumarila Bhatta Mean by Svatah Pramanya?.” Journal of the American Oriental Society 112, no. 3–4 (1992): 204-221.
Tillemans, Tom, “Dharmakīrti”, The Stanford Encyclopedia of Philosophy (Spring 2021 Edition), Edward N. Zalta (ed.), URL = https://plato.stanford.edu/archives/spr2021/entries/dharmakiirti/.
1 Nyāyamajñarī. see (Potter 1981) for a general introduction
2 ibid.
3 As in formula (1.1.2) from Jaimini’s The Mīmāṁsā Formula: “Ritual correctness is what is made known by a rule” (Jaimini 1979, p. 3)
4 (Arnold 2024, “The Brahmanical tradition of Pūrva Mīmāṃsā”)
5 (Arnold 2024, para. 1)
6 Ślōkavārttikam, a commentary on Jaimini’s The Mīmāṁsā Formula. (Arnold 2024, “The works of Kumārila as commentaries on foundational texts”)
7 Ślōkavārttikam [SV] 4.96-97 (Bhatta 1898)
8 SV 5.5.1-3, see also (Gillon 2024, “Culmination of the Classical Period”)
9 SV 4.98
10 Ak. IV 403
11 Indeed, this seems to be the whole point for Kant
12 SV 4.110
13 SV 4.111-113
14 SV 4.120
15 SV 4.212-214]
16 SV 4.229-233
17 (Bhatta 1969, p. 253)
18 SV 4.138
19 (Bhatta 1969, p. 276)
20 (Bhatta 1969, p. 270)
21 A famous Buddhist philosopher and contemporary of Kumārila. See (Tillemans 2021). The verse is from his Notes on the Theory of Evidence (Pramāṇavārttikam) 3.282 (Bhatta, p. 271)
22 ibid.
23 (Bhatta 1969, p. 268)
24 (Bhatta 1969, p. 272)
25 (Bhatta 1969, p. 273)
26 (Bhatta 1969, p. 272)
27 SV 4.26-32
28 (Gangesa 2020, p. 83)
29 see (Taber 1992)
30 (Arnolds 2008, p. 95-96)
31 SV 4.162-164
32 4.249-251
33 (Bhatta 1969, p. 443-444)
34 Here, we follow Gaṅgēśa’s elaboation a similar argument. See (Gangesa 2020, p. 158)
35 (Bhatta 1969, p. 446)
36 (Bhatta 1969, p. 446-447)
37 (Bhatta 1969, p. 270-271)
38 (Bhatta 1969, p. 274-275)
39 (Bhatta 1969, p. 273)
40 In his satirical stage-play, “A Riot of Religions” (Āgamaḍambaram). cite: (Bhatta 2005)
41 (Bhatta 2005, p. 244)
42 ibid.
43 (Bhatta 2005, p. 244)
44 (Bhatta 2005, p. 246)