Re. the nature of ahampratyaya: It is clear that scholars after Kumārila have been having the same debates we are having and have concluded that ahampratyaya must be a form of mānasapratyakṣa. They are much more explicit than Kumārila about it, which seems to show that they sensed the problem and addressed it.

In the 2010 article (the one on Rāmakaṇṭha), Watson wonders whether the ahampratyayas of everyone among us would be the same. He mentions (and excludes) the cases of “I am thin”, which Kumārila explicitly refutes. I think that Kumārila favours a “thick” view of the subject, so that ahampratyayas would be distinguishable, even though not through the characteristics of the bodies attached to them.

In the same article Watson also states that in ahampratyaya the self figures as the object, quoting sources later than Kumārila (this might be relevant, because it seems that Kumārila’s commentators have more definite opinions on ahampratyaya being a form of manas-pratyakṣa, see also here). 

In the 2010 article (pp. 307–308, point 5) Watson imagines ahampratyaya to work like the Naiyāyika anuvyavasāya, namely as a temporally subsequent moment, e.g.:
I know a pot—>I know that *I* knew a pot (=>I know that I must have *known* the pot).
The last step is clearly not needed, Kumārila says that we only occasionally perform the last arthāpatti. I am also not sure about its specific chronology, especially because I am not sure about the chronological separation of ahampratyaya. Do we have any evidence that Kumārila thought of it as occurring later? I have to admit that so far I thought that ahampratyaya was the I’s recognition of itself qua knower while it knows. If it were to occur after the cognition, it would have a viṣaya which is no longer available and thus violate the satsamprayoga ‘connection with something present’ requirement of PMS 1.1.4, which is meant to exclude yogic perception, but also Buddhist types of mānasapratyakṣa. Or at least so I thought. (Buddhists allow for that, given that they believe in momentariness and hence stricto sensu for them every cognition is always about a previous moment; Naiyāyikas don’t have this problem because cognitions are qualities of the self, and hence they are perceptible like other qualities) But how could “my” version work? I can imagine two possibilities:

1. We would need to have two cognitions happening simultaneously, namely that of the pot and that of the aham. This would be impossible for Naiyāyikas, since the manas cannot work simultaneously for both cognitions. Mīmāṃsā authors are divided among the ones who claim that manas is atomic and can therefore only join the ātman to one sense-faculty at a time and the ones who claim that it is vibhu and thus allow for simultaneous perceptions (yugapajjnānutpattir iṣṭaiva, Gāgābhaṭṭa p. 16). I wonder whether this would be similar to the case of apprehending at the same time the piṇḍa, its jāti, its qualities etc. I also need more homework to understand which of the two views is Kumārila’s.
2. Alternatively, we could imagine that perception is a temporally extended process for Kumārila (see his discussion of the move from nirvikalpa to savikalpa pratyakṣa). If this is the case, while I look at the pot I could first know it indistinctly (nirvikalpa stage), then as a pot etc. (savikalpa stage). Perhaps the acknowledgement that it’s me knowing it could take place within this temporal extension? The only difference with the Naiyāyika-anuvyavasāya-like hypothesis would be that the object would not be a preceding cognition (which violates PMS 1.1.4) but still the same I that is currently cognising the pot.

Sucarita’s commentary in Watson 2020, fn. 28 suggests that the ātman grasps itself through a dharma of itself, being cognition, hence there is not the same fault of double use of the same thing as in the Buddhists’ svasaṃvedana, because the grasper is only the cognition and the grasped is only the self (whereas for the Buddhists the same awareness is grasper and grasped). It is also noteworthy here that Kumārila explicitly denies any form of self-illumination by the cognition.

By the way, one may wonder whether this temporal synchronicity between perception and its object would not be violated also in the case of recognition. Mīmāṃsā authors explicitly say that recognition (e.g. “This person is Devadatta!”) is made of perception (“This” person I am seeing) and memory (the “Devadatta” I saw in the past and am now remembering). But I have already discussed that ahampratyaya is not always a case of recognition.

Watson 2020 is also very relevant for the identification of the forth view, attributed to Prabhākara, and its phenomenological character (with the ātman being neither pratyakṣa nor parokṣa, fn. 44) and has helpful footnotes on Śālikanātha’s understanding of Kumārila’s view.

Watson 2010, pp. 303—310, is key on ahampratyaya vs. svasaṃvedana, and how we might be aware of the ‘I’ without being aware of the cognition it is undertaking when we are aware that “*I* know”.

(Corrections on March 25 and April 10 2026, thanks to Alex Watson!)

One of them is adhikāra. Possible differences:

1. adhikāra is generally a vox media (you have the adhikāra to do X, which does not necessarily mean that X is a good thing), unlike “right” (where generally having the right to do X is a good thing).
2. adhikāra might imply duty, whereas rights don’t (you may have the right to remain silent, and this does not imply that you ought to remain silent). For instance, for Prabhākara if you have the adhikāra to perform a given sacrifice, you also have the responsibility to carry it out. However, it is not so in Kumārila or Maṇḍana, where the additional obligation to perform fixed rituals descends from their fixedness (nityatva), not from the adhikāra, as proven by the fact that no such obligation follows in the case of elective rituals (kāmya).
3. adhikāra is connected to ability (sāmarthya), whereas this does not apply to rights, which can instead ground the need of ability being ensured. For instance, if you have the right to go to school but cannot physically move, (in an ideal case) your government will provide you with the devices needed to let you attend school etc. By contrast, adhikāra presuppose ability, in the sense that unless there is ability to do X, there is no adhikāra to do it. Since adhikāra is a vox media, this might be a good thing after all. For instance, if you don’t have the adhikāra to do something difficult to get A, you will be allowed to do something easier instead. Please notice also that Śabara helpfully distinguishes the lack of an external (bahirbhūta) ability (sāmarthya), which is temporary and does not affect the adhikāra (for instance, you don’t lose your adhikāra if you temporarily run out of ghee or are too poor to perform a given sacrifice), and intrinsic (ātmavṛtti) ability (śakti), in the absence of which there is no adhikāra. Much to my disappointment, this distinction is not kept by later authors.

In fact, Mīmāṃsā authors were not just describing a natural scenario. They engineered a highly sophisticated system, with bridges connecting different actions and sewage systems to get rid of unwanted left-overs.

That’s why even though new Mīmāṃsā authors might change the flag on the top of the hill (as Maṇḍana did) or some particular aspect here and there, they were cautious not to jeopardise such a carefully engineered system.

For instance, when it comes to subordination, the only real options are Kumārila’s viniyoga system and Prabhākara’s upādāna. Other authors substantially follow the one or the other.

Bṛhatī ad 6.1.1 says na pravṛttapravartane prayogaḥ āmantraṇādiṣu vyabhicārāt, literally: “[Exhortative endings] are not used to promote people who are already active, because of the deviant case (vyabhicāra) of invitations”. 

In fact, Prabhākara appears to believe that commands are always imparted on someone who is not yet active and who becomes active upon hearing them. The addressee of a command  desires something already and recognises themselves as the addressee through such desires (see Freschi 2012), but they are not active until they become enjoined.

Now, if Prabhākara means that active people are never promoted to act, why are āmantraṇas a good example? Orders (ājñā) are not used as a standard example because their being a clear case of promoting someone not already active had been attacked by the opponent in the previous line.

As for āmantraṇa, the situation with āmantraṇa is ambiguous, but Śālikanātha takes it as connected to a command uttered among peers (thus, it is not a clear case of a command uttered for people who are not already active). Moreover, what is the role of the vyabhicāra argument? How can a vyabhicāra argument be used to convince someone that something is *always* the case? A vyabhicāra can be used to show that it is incorrect to claim that “All As are B”, since there is at least one A that is not a B, but it can’t be used to corroborate “All As are B” by showing a further case of an A that is indeed B. In other words, why speaking of “vyabhicāra” and not of “yathāmantraṇeṣu” or the like if this was what he meant? The only possible explanation seems to be to think of āmantraṇādiṣu vyabhicārāt as “because the case of invitations, etc., deviates from the [opponent’s] claim”.  

Still, if it is impossible to command someone who is already inclined to act, how can Prabhākara make sense of permissions? When we are already about to light a cigarette and look around and ask whether it is OK and someone tells us “No problem, go ahead!”, isn’t the “go ahead” a command directed to someone who is already active? Could Prabhākara perhaps say that a command cannot enjoin someone who is already active because it would miss the apūrva element?

In his monograph on prescriptions, he implemented a revolutionary approach to deontics, interpreting prescriptions as sheer descriptions of state of affairs, namely the instrumentality relation holding between the seemingly enjoined action and the agent’s desired result. “Do X if you desire Y” is therefore nothing but “X is an instrument to Y”.
Similarly “Don’t do X” is nothing but “X is an instrument for an undesirable outcome”. However, in the case of prohibitions, the undesirable outcome is incommensurably greater than any possibly reachable desired output.

In this way, he could achieve multiple goals (in increasing order of importance for Maṇḍana),
1. treat prescriptive statements as if they were descriptive, thus allowing for a smooth interaction among them,
2. make space for non-prescriptive statements in the Vedas (wheras Kumārila and Prabhākara claimed that the Vedas contain only prescriptive statements),
3. allow for a general theory of human motivation as based on rationally conceived goals.

At the same time, Maṇḍana did not want to break up with the previous Mīmāṃsā tradition and its distinction among three types of duties. His reduction might have led to the conclusion that all duties are just based on instrumentality and have therefore no different degree of obligatoriness, but Maṇḍana’s attempt was based on the idea that instrumentality was the real core of the notion of obligatoriness, not an alternative notion. Thus, it should have been able to make sense of the entire Mīmāṃsā deontic building.
Consequently, Maṇḍana kept fixed duties as fixed, but needed therefore something to be fixedly desirable as their coveted result. What could this have been? Maṇḍana discussed and eliminated various options. Happiness, for instance, would not be enough, because though people generally desire happiness, they might be too lazy to undertake cumbersome actions to achieve it. By contrast, reduction of pain is considered to be a universally desirable output and one which would motivate even lazy agents. Thus, Maṇḍana settles for reduction of bad karman as the output of fixed duties, because of the present unpleasant outcomes of bad karman (such as present suffering) and because bad karman will hinder one’s future enterprises. In this way, Maṇḍana could distinguish between elective duties (bound to specific desires and outputs) and fixed and conditional ones (bound to the reduction of bad karman, which is generally desirable).

The core of his solution for the Śyena problem centers on the rationality of the agents involved: No one would want to obtain a small benefit in exchange for the negative output (the always undesirable bad karman) of transgressing a prohibition.

Even though later Mīmāṃsā authors generally disagreed with Maṇḍana’s solution, they often incorporated some elements of it within their own schemes (e.g., the Bhāṭṭa Mīmāṃsā Pārthasārathi incorporated the awareness of instrumentality in his teaching of prescription, see Freschi 2012). Several Vedāntic schools also took advantage of his solution for justifying their reading of the Vedas as (also or chiefly) descriptive.

—Śālikanātha (around the 9th c.), a thinker of the Prābhākara subschool of Mīmāṃsā, has been incredibly influential for almost all Sanskrit philosophy

—a typical methodology of Sanskrit philosophers for finding the truth is examining definitions and refining them by eliminating what is wrong or superfluous and trying to find the perfect definition (see here for Śālikanātha’s refutation of the Buddhist definitions)

—concerning the issue of knowledge, Śālikanātha’s perfect definition of it sounds pretty minimalist: knowledge is experience. This is enough, he thinks, because it excludes memory (which is not experience) and because there is no need to exclude doubt or error. In fact, doubt is not a single cognition but a sum of two correct cognitions. Erroneous cognitions, in turn, do not exist as such, but are only incomplete cognitions. For instance, we might believe that something shiny we see on the beach is silver, whereas it is in fact mother-of-pearl. This is not a real mistake, since the “this” part of the cognition “This is silver” is correct. The “silver” part is just a memory, since it corresponds to the silver we have seen elsewhere and we have been reminded of because of the equally shiny mother-of-pearl. Thus, experience are always correct and “knowledge is experience” is a sufficient definition of knowledge.

What we are going to see today:
Śālikanātha’s definition of sense-perception

Why should we care?
1. Because Śālikanātha is a great thinker, who influenced almost all later Sanskrit philosophers.
2. Because Śālikanātha is an elegant writer, a pleasure to read.
3. Because Śālikanātha has interesting arguments to offer on the topic at stake.

On 3: One might think that the definition of “sense-perception” is easy and that there is no need to discuss it at length. We would probably all agree that it depends on sense-faculties and much of the disagreement has already been dealt with under the heading of the definition of knowledge (such as the issue of infallibilism).
However, this is not the case. First of all, Śālikanātha needs to discuss the (wrong, in his opinion) definitions coming from Dharmakīrti’s school, Nyāya and Mīmāṃsā itself. Much of these definitions is wrong because of reasons already discussed in connection with the definition of pramāṇa.

Accordingly, his definition of sense-perception will be minimalist: A direct experience (sākṣātpratīti) generated from the connection of the sense-faculties with the knowable items, i.e., substances, universals and qualities, with or without concepts:

sākṣātpratītiḥ pratyakṣaṃ meyamātṛpramāsu sā |

meyeṣv indriyayogotthā dravyajātiguṇeṣu sā ||

savikalpāvikalpā ca pratyakṣā buddhir iṣyate |

(Amṛtakālā v. 4–5ab)

Since the definition needs to contain nothing redundant, let us analyse each element of it:

• sākṣātpratītiḥ (“it is a direct cognition”): to exclude inference etc. (agreed upon by all schools)
• meyamātṛpramāsu sā (“it is about object, knower and knowledge”): sense-perception can grasp, besides substances, also qualities and even universals (agreed upon by Bhāṭṭa Mīmāṃsā, against Dharmakīrti’s school). Śālikanātha specifies on p. 142 of the pramāṇapārāyaṇa that if one were not to accept that, there would not be anything left to cognise via sense-perception, since one never grasps substances on their own.
• meyeṣu indriyayogotthā (“it is generated by a contact of the sense-faculties with the objects”): against the idea of intellectual intuition (agreed upon by Bhāṭṭa Mīmāṃsā, against Dharmakīrti’s school and Nyāya)
• dravyajātiguṇeṣu sā (“it is about substances, universals or qualities”): sense-perception can grasp, besides substances, also qualities and even universals (agreed upon by Bhāṭṭa Mīmāṃsā, against Dharmakīrti’s school). Śālikanātha specifies on p. 142 of the pramāṇapārāyaṇa that if one were not to accept that, there would not be anything left to cognise via sense-perception, since one never grasps substances on their own.
• savikalpāvikalpā (“it is conceptual or non-conceptual”): it can be conceptual and non-conceptual (agreed upon by Bhāṭṭa Mīmāṃsā, against Dharmakīrti’s school)

Next, Śālikanātha has to define what counts as perception. Possible candidates are (according to Dharmakīrti’s school):
1. sense-perception
2. mental perception (mānasapratyakṣa)
3. self-awareness (svasaṃvedana)
4. intellectual intuition (yogipratyakṣa)

1. is clearly accepted.

2. is, surprisingly and against Bhāṭṭa Mīmāṃsā, accepted (see the discussion about meyamātṛpramāsu sā above). All cognitions are therefore immediately accessible to self-awareness.

3. is refuted. This might be slightly complicated, since manas is not defined as in Nyāya and Bhāṭṭa Mīmāṃsā, namely as the inner sense faculty, which can grasp pleasure and pain. Śālikanātha succintly defines mānasapratyakṣa as “produced by the cognition through the sense-faculties together with the instant immediately following the one the instant (kṣaṇa) which has been previously cognised (by the sense faculties)”. The commentator, Jayapuri Nārāyaṇa Bhaṭṭa, explains that it is therefore produced by the object and the cognition together. The terminology and the topic itself make Śālikanātha steer here in the direction of discussing Dharmakīrti’s school. He then later (p 142 of the pramāṇapārāyaṇa) refutes it, explaining that it is nothing different than continuous perceptions (dhārāvāhikajñāna), like the ones we have while staring for a long time the same object. By contrast, the so-called mental perception that continues also once the contact with the sense faculties has ceased, is just no longer a case of knowledge, Śālikanātha explains.
4 is refuted. Why? Because what we think to be grasping directly but without the senses (e.g., the four noble truths during deep meditation) is nothing but our memory presenting us with some content we were already acquainted with. There is nothing new that derives from meditation itself. Claiming the opposite is something you can do for religious reasons, not on epistemological bases.

What I would like you to remember of this series?
—Go check Śālikanātha’s extensive Prakaraṇapañcikā if you want to make a stroll in philosophical landscapes, almost all topics are covered, and always with interesting ideas.
—”Knowledge is experience”: there is no error.
—Intellectual intuition is nothing but memory.

In some sense, one could even say that he is more important than Prabhākara himself, since he is way more systematic than Prabhākara, and explores through his various thematic essays almost all topics commonly dealt with in Sanskrit philosophy. Moreover, he is certainly more influential than Prabhākara, since his works are regularly read and cited instead of the terse words of Prabhākara’s only extant work, the Bṛhatī.

Chronology
Like in the case of the relation between Kumārila and Prabhākara, Śālikanātha’s position in the history of Sanskrit philosophy needs further investigation. His systematisation of Prābhākara philosophy, answering (or trying to answer) all challenges coming from the Bhāṭṭa field is so thorough that no philosopher after him went back to Prabhākara alone without taking into account his explanations. For instance, no one went back to Prabhākara’s account of arthāpatti, independently of Śālikanātha’s reinterpretaion. All of Kumārila’s interpreters and commentators have been influenced by Śālikanātha and at times mutuated their siddhānta from Śālikanātha’s objections.
However, there is one author referring to Prābhākara ideas and not taking into account Śālikanātha’s points. This is Jayanta, who is also among the few authors whose dates are relatively settled (870–950 ca.). Thus, Śālikanātha either lived after Jayanta, or was not yet known at the time of Jayanta in Kaśmīr.

Agenda
As hinted at above, Śālikanātha tried to systematise Prabhākara by making an all-encompassing Prābhākara philosophy. In other words, he tried to stretch Prabhākara’s views way beyond what was more important to Prabhākara (such as deontic and hermeneutic issues) and to cover also ontology etc. He also tried to raise to the challenge produced by Kumārila by reinterpreting Prabhākara’s theory in a way apt to answer to Kumārila’s objection (for instance, by reconsidering the role of apūrva, by admitting smārita padārtha within the process of signification and by discussing the cognitive aspect of abhāva).

These are just some of the reasons that make it relevant and necessary to read and study Śālikanātha. Other reasons include his being a) philosophically intriguing (as certified even by his opponents, see above concerning Kumārila’s commentators reusing them); b) an enjoyable and elegant author.

(cross-posted on the Indian Philosophy Blog, where you can also read some interesting comments)

First of all, he discusses the criterion of avisaṃvāditva ‘non deviation’ (used by Dharmakīrti and his school) and shows how this is not enough to exclude memory (smṛti). Dharmakīrti could exclude memory because it is conceptual, but this would exclude also inference (anumāna).

Next suggestion (again from Dharmakīrti’s school): using causal efficacy (arthakriyā) as criterion. But in this way memory should again be considered a source of knowledge, since it can be causally efficacious. One could say that, unlike in memory, in the case of inference there is a connection (though indirect) with the object. But this, again, applies to memory as well!

A new attempt is to say that a source of knowledge is identified insofar as it leads to know something unknown (aprāptaprāpaka), which is a criterion typical of Kumārila. A variant thereof is to say that it causes to act people who were previously inactive (pravartakatva), but this would lead to the fact that non-conceptual cognitions (nirvikalpa) would not be sources of knowledge, given that they cannot promote any action.

Why not using aprāptaprāpaka as criterion? Because this would not apply to the case of continuous cognitions (dhārāvāhikajñāna). These are cognitions like the ones originated out of continuously looking at the same object. These count, according to Śālikanātha, as sources of knowledge, but would not be such if the criterion of aprāptaprāmāṇaka were to be the defining one.

What about dṛḍha ‘sure’ as criterion, then?
Here Śālikanātha can give voice to the Prābhākara theory of knowledge. First of all, he asks, what would dṛḍha exclude? If it excludes doubt, then this is wrong, since there is no doubtful cognition. What we call ‘doubt’ is instead the sum of two distinct cognitions (readers might want to recall the fact that for the Nyāya school, doubt is a cognition in which two alternatives are exactly equally probably).
As for erroneous cognitions (bhrānti), these also don’t need to be excluded from the definition of knowledge, because there are no erroneous cognitions. What looks like an erroneous cognitions, is at most an incomplete one. For instance, mistaking mother-of-pearl for silver means rightly recognising a shining thing on the beach + remembering silver. The latter part is not knowledge, but just because it is memory. Śālikanātha similarly treats the case of jaundice and other perceptual errors.

His conclusion is a minimal definition of knowledge: pramāṇam anubhūtiḥ “knowledge is experience”.

(cross-posted on the Indian Philosophy blog, where you can also read some interesting comments)

In other words, assuming that one learns the Veda due to the prescription to teach it, does this prescription include the duty to teach the meaning? No, says Veṅkaṭanātha in his refutation of the Prābhākara position. Just like the knowledge of the meaning of the Veda is no included in the prescription to learn it by heart, so it is not included in the prescription to teach it. Both stop their function at the Vedic phonemes.

So far so good, but then Veṅkaṭanātha adds an additional reason why the prescription to teach does not reach until the meaning of the Veda, namely:

अबाधितप्रत्ययोत्पत्तावनपेक्षत्वलक्षणप्रामाण्यस्य वक्ष्यमाणत्वाच्च

And because in the case of the coming into being of a cognition which has not been invalidated, we will say that its validity (prāmāṇya) consists in its being independent. (SM ad 1.1.1, 1971 p. 27)

The reference is clearly to PMS 1.1.5, where the Veda is said to be a pramāṇa because it is independent from any other source. That is, once a cognition has indeed come into being and is not sublated, the only thing which could make one doubt about it is its having the wrong source, but if it is independent on any source, no such worry can arise. Why is this said here? Perhaps because a cognition of the meaning does indeed take place upon learning the Veda by heart and unless one can prove that it is wrong, one needs to consider it valid. Hence, the need to study Mīmāṃsā cannot be justified on the basis of the need to understand the meaning of the Veda.

• 1. Each prescription needs an independently desirable purpose and it is not clear what could be the purpose here (I discussed this topic in a previous post, here)
• 2. Did you understand the adhyayanavidhi while reading it? If not, you really need to study Mīmāṃsā, but you will not get there, since you will not even start learning by heart the Vedas, given that you never came to know that you should have learnt them. Did you understand the adhyayanavidhi? Great! But this means that you don’t need to study Mīmāṃsā, since you already understand the meaning of Vedic prescriptions, isn’t it?

Prābhākara Mīmāṃsā authors can solve the problem by saying that one does not have an independent duty to learn the Veda, but one does it nonetheless because teachers have the duty to teach and there cannot be teaching without learning. In other words, they can in this way make sense of the fact that at the time you undertake the learning, you are not mature enough to decide rationally on the basis of a means-goals calculation.

Rāmānuja and Veṅkaṭanātha are overtly against Prabhākara for other reasons. Hence, they need to offer a different solution. Accordingly, Rāmānuja suggests that the prescription to learn does indeed reach until the meaning, but that the meaning it reaches is not a fully satisfactory one. It leads one to an āpātapratīti of it. Veṅkaṭanātha elaborates further the implications of this assumption: Through the prescription to learn, one is led up to a first impression of the meaning of the Veda. Then, out of interest, one continues studying Mīmāṃsā in order to solve one’s doubts and achieve proficiency in the Vedic texts.