On one sentence of the Kaccāna Sutta

Tags: #<Tag:0x00007fc7ba088b60>


One of the most famous suttas, and justly so, is the Kaccāna (or Kaccāyana ) sutta of the Samyutta. It concerns a discussion between the Buddha and monk Kaccāna on right view. The Buddha defines this in terms of avoiding the extremes of “existence” and “non-existence”. The sutta, while short, has many profound and evocative aspects. It’s one of my favorites, and I memorized the Pali many years ago.

Evidently it was well regarded from an early time, for it is quoted by Ānanda in a teaching to another monk, Channa. But its historical significance wasn’t fully felt until much later, when it became the core textual basis for Nagarjuna’s Mulamadhyamikakarika, and hence the whole of the Madhyamika school, which dominates the Buddhism of Central Asia, and is influential in East Asian Buddhism too.

Much has been written on this text, and alas I haven’t caught up on all this. But while translating I found that, while most of the sutta is pretty straightforward, one sentence is very knotty. Fortunately there are several versions of the text, and while they don’t solve all the problems, they do point to some solutions.

The basic text appears at SN 12.15, with a Sanskrit version from Turfan, numbered SF 168 in SuttaCentral’s system. There is a Chinese translation of this at SA 301. This teaching is quoted by Ānanda at SN 22.90, which in turn has a Chinese parallel at SA 262. The Pali passage is quoted verbatim, so it merely repeats SN 12.15 and can be disregarded. SA 262, however, has a number of differences from SA 301, so should be considered as an independent parallel. These differences may have been present in the Sanskrit original, or be a result of the translation.

Of these variants, the Sanskrit and Chinese stem from the same or very closely related textual lineages, so they form one group, while the Pali is somewhat distinct.

All these texts are available in English translation on SuttaCentral. I won’t discuss these translations, as it is all too complicated already! But a simple survey of them would show that they all approach the relevant passage in different ways.

Here’s the difficult passage in Pali.

Tañcāyaṃ upayupādānaṃ cetaso adhiṭṭhānaṃ abhini­ve­sā­nusa­yaṃ na upeti na upādiyati nādhiṭṭhāti: ‘attā me’ti

There are many problems associated with this, so I will follow my usual method of working out what we can know with a reasonable degree of confidence first, then gradually stepping into less clear waters.

First of all, each version does in fact have a different take on this passage, and none of them are obviously correct. In fact it seems likely that all versions have suffered some degree of corruption. This means that any translation will have to take this into account, and cannot be too tightly bound to any one reading.

A more significant point is that the corruption has quite likely emerged from, or at least been influenced by, the shift in meaning of two of the terms here.

  • Upaya: This is a term for the defilements. However it is often confused with the famous term upāya, which in later Buddhism means “skillful means”. There are a few passages where it has a similar meaning in the EBTs, but it has yet to become a clearly defined or prominent term.
  • Adhiṭṭhāna: In later Buddhism this means “resolution”, and is one of the ten paramis. However in the suttas it occurs rarely, and is a neutral term . It does occur in the sense of “determining” the ownership of robes and such in the Vinaya. It also appears as a group of four adhiṭṭhānas in DN 33, although the exact sense there is unclear.

I think it’s also relevant to note that the Sanskritic texts all stem from the northern schools associated with the Sarvastivada. Now, this school either originated in, or was closely associated with, the school of Upagupta at Mathura. And, while the idea of upāya is nowadays associated with Mahayana, in fact it is very likely to have originated with Upagupta. For this reason, it is understandable that the Sanskritic texts should replace upāya when used in a negative sense, either from confusion or from a desire for doctrinal clarity.

Thus where the Pali has upaya the Sanskrit has upadhi. Now, upadhi is another term for defilements, and is not obviously out of place here. However, these terms appear as both noun and verb. And since upadhi doesn’t have a convenient verb form, the Sanskrit leaves the verb as upaiti. Oops! Looks like the Sanskrit has been caught with its pants down. It tried to make a doctrinally meaningful change, but did it sloppily.

So this is a second thing we can say with fair confidence: upaya is preferred over upadhi.

Next, we can, relying in part on this, say something about the structure of the list as a whole. In a previous post I mentioned that such lists in Pali are usually of two sorts: either progressive lists, where one factor leads on to the other, or identical lists, where each factor is a mere synonym. I believe that this list is a subtle variation on this rule: it’s both. The terms are, on the one hand, all terms for the defilements and as such can be regarded as synonyms. At the same time, they express a gradually increasing level of engagement, an ongoing process of deepening attachment. But this structure can only be discerned by looking closely at the terms and rejecting later meanings.

  • upaya: Has the basic meaning of “approaching”, “drawing near”.
  • upādāna: Means “taking hold”, “grasping”. It is often translated as “clinging”, but I believe this is incorrect, as clinging means a refusal to let go of what one already has, whereas upādāna normally (for example in dependent origination) means grabbing hold of something new.
  • adhiṭṭhāna: Has a sense of “resolve”, “fix”, “make sure”. In the Vinaya we find it used in an external sense as “determining” by body or speech the ownership of a robe, etc (Pvr 7). Perhaps it was to clarify that it is used here in a psychological sense that the adjective cetaso is used here. But in the suttas the most common sense is that of “fixing” the mind on a place to be reborn: taṃ cittaṃ dahati, taṃ cittaṃ adhiṭṭhāti, taṃ cittaṃ bhāveti (One places, fixes, and develops that thought).
  • abhinivesa: This is used most commonly in the context of the “insistence that this alone is the truth”, “Only this is true, all else is dumb”.
  • anusaya: Translated as “underlying tendencies”, this is when the defilements have become entrenched at a fundamental level.

So we can make out a clear progressive sense here, with the help of our favorite tormented ex-hobbit. Gollum draws near the Ring (upaya). Fascinated, he picks it up (upādāna). Then he clenches it in his fist, determining that it is his and committing to that (adhiṭṭhāna). He develops justifications for why he alone owns it, and fiercely fights off any challenges, insisting that it is “My precious!” (abhinivesa). Finally, it becomes part of who he is; even when he’s not wearing it or thinking about it, it’s gnawing away in the back of his mind, shaping his choices and actions without his awareness (anusaya).

In the context of this pattern, our former conclusion is reinforced, since upaya makes much better sense as an initial factor in this sequence than upadhi. So this is the third thing we can say with some confidence.

Next, the passage is clearly structured around the echoing of verbs and nouns. Such structures are very common in Pali. It regularly provides reinforcement by pairing nouns and verbs that are synonyms (saddham adhimuccati) or even using the same word in both forms (paññāya pajānāti). We do the same thing in English when we tell a tale or sing a song or weave a web. It’s not so common to find the nouns and verbs separated as they are here, but the parallels are too close to ignore.

This leads to our next confident conclusion: we shouldn’t read too much meaning into how these nouns and verbs relate to each other. It’s difficult to parse out the exact grammar of such a complex sentence, and tempting to read into the passage a way of dealing with defilements that is not found elsewhere. But very likely the repetition is just that, a repetition, and is intended to reinforce, not complicate.

Given that, a translator has a choice. In some cases, it is justifiable to simply ignore such repetitions, if they merely clog up the text. If they are to be included, however, they must be set in some kind of oblique case that doesn’t invite forced interpretations.

It is still not really clear exactly how to deal with this passage. This is because the terms vary in the different versions. But we are once again, I think, justified in reaching at least some conclusions. The verb forms in the different editions are easiest to deal with, as they are separated and preceded with a “na”, so let’s list them:

SA 301 			不取		不住				不計我
SA 262	 若不受		不取		不住				不計於我
SF 168	 nopaiti 	nopādatte 	nādhitiṣṭhati 	nābhiniviśaty	ātmā me'ti
SN 12.15 na upeti 	na upādiyati 	nādhiṭṭhāti			attā me'ti

As you can see, the terms generally match up, and clearly stem from a similar root. SA 301 omits the first term, which appears to be a mere accident. The verb form for abhinivesa is only attested in the Sanskrit.

The final term is the most troublesome. It’s not immediately obvious how to parse out the Chinese forms, which moreover differ between each other, although that difference is probably negligible. And it’s not clear whether they are meant to stand for nābhiniviśaty or ātmā me’ti, or perhaps both combined.

However, in SA we find such forms regularly used to stand for the reflexive iti when applied to contemplation of the self. At SA 261 = SN 22.83, asmīti = 計是我, and similarly at SA 63 = SN 22.47 asmīti = 計我. The DDB, therefore, appears to be correct in that it does not list abhinivesa as a source for 計, but it does have iti.

But if this is the case, why is the negative included, thought it’s absent from both Pali and Sanskrit? In Pali, we actually have a variant, attā na me’ti, which is the accepted reading in the PTS edition (though typeset incorrectly). This includes the negative, and hence aligns it with the two Chinese variants. Whether this is the original form, however, is not clear; it seems to me a bit odd to include the negative in the middle of the phrase like this. It’s standard to put it at the beginning: neso me attā. Given the liberal sprinkling of negatives throughout this passage, it’s probably unwise to make too much of this, as the sense is clear enough.

We are still left with a question mark, because the verbs and nouns don’t match exactly. The list of nouns ends with abhinivesa and anusaya, but as we’ve seen, abhinivesa is attested only in one source as a verb, and poor old anuseti is not found anywhere. That this represents a decay in the text is, I think, possible, but impossible to say with any certainty. On it’s own, I thought it more likely that nābhiniviśaty was original, but since it is absent from both Chinese versions, I have come to think it is more likely a later standardization.

Next, we deal with another problem, the initial tañcāyaṃ. The commentary says that tam refers back to the “noble disciple” who has already become a stream enterer, while ayam points forward to the list of defilements. Ven Bodhi follows this interpretation, but I find it very implausible. There’s no clear referent for the tam to be pointing back to, for a start, so Ven Bodhi has to insert the commentarial gloss into the text. More to the point, such constructions are found constantly in Pali, and they are invariably simple redundancies. There’s no reason to think otherwise here. The Sanskrit text etāni ced confirms this, as it has just one pronoun, and it is plural so must agree with the following list of defilements (which are declined in plural in Sanskrit, but in singular in Pali).

More importantly, the Sanskrit introduces the particle ce, meaning “if”. In fact the Sanskrit has it twice, at the beginning of this passage, and before the next passage on not doubting: “if one doesn’t doubt …” The if is also included in the text of SA 262. In SA 301 the phrase that contains the if is missing, so it may well have been there as well.

But the Pali may well also have “if”. The “c” in tañcāyaṃ is most obviously read as a trunctated ca, but could just as well stand for ce. Given it seems to be a constant in the Sanskritic passages, I think we should accept it here.

The dropping of the ce and the commentaries subsequent forced explanation of this opening—something that should really be too trivial to warrant attention—is indicative of a wider problem. There is no real organic sense of progress in the text. Normally a sutta might begin with a statement of the problem, propose a solution, then implement it. The internal logic is easy to follow. But here it seems to jump from being enmeshed in suffering to being a stream enterer. The important passage on the origination and cessation of the world is also found in a different position in the Pali and Sanskrit texts, so however you resolve this there is clearly a problem with sequence.

Adopting the reading “if” suggests a solution to this: here’s the problem; if you do this, the problem is solved. The overall structure of the text flows as expected.

Finally, we a problem with the application of the series of verbs. The Pali commentary, which as usual is followed by Ven Bodhi, says that nādhiṭṭhāti, the final verb, is distinct from the others in the list, in that it refers directly to the statement attā me, whereas the other verbs presumably refer back to the noun structure from earlier. Once again, I find this implausible. In all versions, the sequence of verbs is clearly cumulative, and shouldn’t be broken up. Doing so tortures the sytnax. Worse, it creates a fatal doctrinal flaw. Here is Ven Bodhi’s translation:

But this one with right view does not become engaged and cling through that engagement and clinging, mental standpoint, adherence, underlying tendency; he does not take a stand about ‘my self.’

Note that “one with right view” does not occur in the text and has been inserted from the commentary; see my previous discussion of the opening phrase. The sentence as a whole has become so convoluted that it’s hard to parse in English. But one thing is clear. When we separate the final phrase off from the rest, we are saying, or at least suggesting, that the stream enterer has no grasping or underlying tendencies. This is of course not the case. They specifically have no defilements related to a theory of self, and the passage must mean that. But in order to do so, all the verbs, not just nādhiṭṭhāti, must indicate the final phrase, ‘my self’.

Finally, I’d like to present my translation. I’ll include a little bit before and after the passage we’ve been discussing, so the context is clear. (Note there here, as a stylistic choice, I use the generic you, as I cordially loathe the cold and archaic overuse of “one” in Buddhist translations. The Pali text, of course, has neither, it just has the verbs.)

The world is for the most part shackled to approaching, grasping, and insisting. But if—when it comes to this approaching, grasping, mental commitment, insistence, and underlying tendency—you don’t approach, grasp, and commit to the notion of ‘my self’; and if you have no doubt or uncertainty that what arises is just suffering arising, and what ceases is just suffering ceasing, your knowledge regarding this is independent of others.

See the discussion for further modification of this.

If anyone is interested in a further discussion of this, there’s a long discussion on Dhamma Wheel, which has valuable links to many references. I haven’t read all these, so please correct me if I have made any mistakes. I have read Jayarava’s essays on this sutta here and here. I learned a lot from both these essays, but I wanted to mention that in a couple of details I had arrived at the same conclusion independently. Namely, that upaya should be preferred over upadhi, and that the sequence of nouns and verbs should be considered together. Just yesterday I mentioned a case where independent inquiry came up with similar results from different directions, and this is another instance. It’s useful to bear this in mind, as it’s easy to think that scholars disagree about everything, so their methods must be a waste of time!

MN 2, why is "no self exists for me" a wrong view?
Why it is called Rupa?

Fascinating! Thanks!

Going by SC I think it’s actually 12.15, but thanks also for 12.25 - it’s awesome!


Oops, let me fix that …


Bhante, did you mean one of those to 262s to be 301?


This is so cool, I just don’t have to worry about accuracy or proofreading, and you guys will fix it for me!


Crowd-sourced research. Yes!


Thanks, that’s a very interesting read. I’d love to see such an approach more in our field. Sometimes it feels like there’s so few, doing such hard work, in such a large area, that we just want to keep our heads down and get on with it. Anyway, I hope that with my translation, the initial version will be seen as a rough draft, and over the years it can be refined by many hands and made into something beautiful.


I’d be interested in any insights on the effect this change might have on the statements about notions of existence and nonexistence.

One of the points that Sylvester made, and I filled in in detail on DhammaWheel was that the Northern versions (SA 301, etc) had the passage Bhante Sujato has examined between the statements on the notions of existence and non-existence in the Pali SN 12.15.

With this re-ordering, and Bhante Sujato’s translation, it reads:

“This world, Kaccana, for the most part depends upon a duality—upon the notion of existence and the notion of nonexistence.

The world is for the most part shackled to approaching, grasping, and insisting. But if—when it comes to this approaching, grasping, mental commitment, insistence, and underlying tendency—you don’t approach, grasp, and commit to the notion of ‘my self’; and if you have no doubt or uncertainty that what arises is just suffering arising, and what ceases is just suffering ceasing, your knowledge regarding this is independent of others.

But for one who sees the origin of the world as it really is with correct wisdom, there is no notion of nonexistence in regard to the world. And for one who sees the cessation of the world as it really is with correct wisdom, there is no notion of existence in regard to the world.

Given Bhante’s clarification that the passage he addresses is about stream entry, perhaps the meaning of the latter could be clarified. The SA 301 translation reads:

Why is this? One who rightly sees and knows, as it really is, the arising of the world, does not hold to the non-existence of the world. One who rightly sees and knows, as it really is, the cessation (passing away) of the world, does not hold to the existence of the world.

This seems to be in agreement with the interpretation of the notions of existence and non-existence in the commentary in terms of eternalism and annihilationism, but perhaps, given Bhante’s interpretation in terms of stream entry, it instead, or also, refers to understanding dependent origination.


Yes, i tried to avoid the philosophical issues, as they are less amenable to a confident solution!

i think the order you’ve put it in makes sense. Seeing the “arising and cessation” " as they are" (yathābhūta), these are unambiguous keywords indicating stream entry. In the Pali text we go from that, then back to a statement of the nature of the problem, then back to stream entry again with the statements on having “no doubt”. Even lacking the Northern texts it should be obvious this is mixed up.

Regarding the presence or absence of sabbam atthi, of this I am not so sure. It makes sense that the Sarvastivadins would remove it, but it would make equal sense that the Theravadins would add it. I’m not aware of any decisive arguments on either side.

The passage is certainly about eternalism/annihilationism, which is one of the aspects of wrong view that the stream enterer sees through. I’m not sure why this is problematic; sorry, but I couldn’t make out Sylvester’s argument on the Dhamma Wheel thread, I’m afraid. I’m not aware that Nagarjuna was disputing this. The whole point, it seems to me, of his argument is that any notion of “inherent existence” commits you to a hidden eternalism. If it exists “in and of itself” then it is not conditioned, and hence not impermanent. Thus the sabhava doctrine ends up with eternalism. Obviously the Kaccana Sutta doesn’t overtly address the notion of sabhava, it’s an inference. But it seems to me a perfectly reasonable one.


This passage has been the subject of numerous discussions on Dhamma Wheel:

And for one who sees the cessation of the world as it really is with correct wisdom, there is no notion of existence in regard to the world.

Some hold that any interpretation of the Dhamma that admits the possibility of an objective reality of external objects is wrong view, because that would be “a notion of existence”. Personally, I don’t think that the Buddha cared about the issue of the reality (or not) of external objects such as rocks or planets, and such philosophical issues are simply not relevant.


Oh, okay, yes. Well, as I understand that was a controversy even within Nagarjuna’s followers. I heard a Tibetan lama speak on this once, and he said that the Buddha’s teachings concern what is knowable, and that is always a relation. It seemed reasonable to me.


Perhaps you could expand on “It makes sense that the Sarvastivadins would remove it…”

Sylvester’s argument was as follows:

… the loss of the "The All exists"and “The All does not exist” passages proved particularly acute for the Northern Buddhists, as their sutra became dislocated from the pre-Buddhist context. It does not help that SN 12.47 and SN 12.48 do not seem to have survived in the Agamas either. Even if echoes of the Upanisadic “sarvaṃ asti” survived in the memory of the Northern schools, it did not help that the Sarvastivadins adopted a similar-sounding but unrelated motif, which gave rise to so much spilled ink over the issue of svabhāva that detains Ven Nanananda today.

Let it be said - the Buddha was not in the least bit interested in the question of own-being or essence of dhammas.

If we accept the reordering of SN 12.15 above, then the Pali version would read:

… But for one who sees the origin of the world as it really is with correct wisdom, there is no notion of nonexistence in regard to the world. And for one who sees the cessation of the world as it really is with correct wisdom, there is no notion of existence in regard to the world.

‘All exists’: Kaccana, this is one extreme. ‘All does not exist’: this is the second extreme.

Without veering towards either of these extremes, the Tathagata teaches the Dhamma by the middle: ‘With ignorance as condition, volitional formations come to be; …

The Northern versions omit the bolded part. Admittedly, this is not a huge change but if it is omitted then the extremes are the notions of existence or non-existence, rather than “All exists/All does not exist”.

I agree it may be too obscure a point, and perhaps Sylvester could clarify.


How could Sylvester possibly improve on your exposition. It’s crystal clear!

I checked Nagarjuna’s MMK 14.7 and it cites only that portion of the Katyayana dealing with existence and non-existence. There is no mention of the Sarvam Asti passage, suggesting the possibility that it had been lost by his time.

As for Bhante Sujato’s suggestion that the Sarvam Asti could have been inserted by the Theravadins in response to the Sarvastivadin’s “Sarvam Asti”, i think that is unlikely given SN 12.48. That sutta’s issue with Sarvam Asti etc seems to point backwards to the Upanisadic debate of existence (Sat) and non-existence (Asat) that it is unlikely to have been a reaction to the unrelated Sarvastivadin issue.


What about the logical pair to a list that “in the darkness binds them”… a list of unbinding? Not sure if there are variations but I think it’s a stock phrase in the suttas; here’s an example from SN47.32:

ekanta­nibbidāya virāgāya nirodhāya upasamāya abhiññāya sambodhāya nibbānāya

[lead to] utter revulsion, to dispassion, to cessation, to peace, to direct knowledge, to enlightenment, to Nibbāna.

Bhante, just wondering what category of list this might be?


… the loss of the "The All exists"and “The All does not exist” passages proved particularly acute for the Northern Buddhists, as their sutra became dislocated from the pre-Buddhist context.

Maybe… possibly… but that’s really just one potential narrative. Also, all Buddhist traditions were somewhat disconnected from much older contexts. Sri Lanka too did not have the religious culture of northern India from the time of the Buddha.

There was also not one sūtra for “Northern Buddhists,” but instead many different textual lineages of the Āgamas throughout India and beyond. One of those textual lineages went to Sri Lanka and became Theravada Buddhism.

It does not help that SN 12.47 and SN 12.48 do not seem to have survived in the Agamas either.

There are hundreds of Pali suttas that are not found in Chinese, and vice versa. Texts in the nikāyas that are not in the āgamas were not simply “lost.” There are also many texts in the āgamas that have no parallels in the nikāyas. These collections varied to some extent.

Researchers have also found this variance reflected in the Gandhari texts, which are typically different from the Chinese versions and the Pali versions.


I hadn’t considered this, but I think you’re right. Certainly the terms “sabbam atthi” and so on have an Upanishadic flavor, and the fact that they are directly connected with a brahman reinforces that. (And, incidentally, gives a hint as the meaning of “lokayaita”.)

This being so, it’s likely that sabbam atthi wasn’t invented by the Theravadins, although it’s still possible it was inserted into the Kaccayana Sutta. Nevertheless, the point remains, as, so far as we know, there are three suttas with the phrase, and two of them have parallels, and in the third the parallel is not found. To me this shows that there’s clearly a sectarian bias at work.

Most likely the phrase was, as you say, originally a critique of Upanishadic ontology, but at a later date that became irrelevant, and it was perceived as ammunition for attacking the Sarvastivadins, so was quietly dropped from their recension.

It does bring up a few things. First, the phrase has its negative: sabbam natthi. What was that about? Could this be a reference to Yajnavalkya’s famous neti, neti?

And in the context of Buddhist tradition, there is of course the long term tendency for Buddhism to slip back into the eternalist ideas of the Upanishads. We see this today in the “original mind” school, of course. But the Sarvastivadins may easily be seen as one of the first steps in that direction. If a critique meant for the Upanishads was suppressed by the Sarvastivadins, is this an unconscious admission that their doctrine really was influenced by the Upanishads, or at least was leaning that way?

There’s an interesting idiom in SN 12.17, iti vadaṃ sassataṃ etaṃ pareti. It means something like “This doctrine leans towards (or “amounts to”) eternalism”. It’s this idea that forms of eternalism and so on might not be explicitly stated, but there are underlying assumptions that lead in that direction. This is the kind of connection, I think, that Nagarjuna was making between sabbam atthi and the svabhāva doctrine.


I think here at least “northern” is just an overly vague way of saying “sarvastivadin”, since as it happens all the relevant texts we have are in fact from them (or related schools).

Yes, but it’s not a matter of simply assuming that if it is missing it must be sectarian bias. In this specific case, the vast majority of substantive suttas in the Samyutta do in fact have parallels. Most of the ones that are missing are articificial suttas (in repetition series and the like). To find the same idea in three suttas, and to have all of those passages missing in all the recensions of another school, when the passage in question specifically critiques the phrase that is the very name of that school; I could hardly imagine a more likely scenario for sectarian bias.


My goodness, you’ve just discovered the first cogent argument in support of translating Nibbana as “unbinding”!

Well, it’s certainly progressive, as these items, or at least most of them, frequently feature in a progressive sense. At the same time, most, perhaps all, of the terms can be used as terms for Nibbana, so perhaps we should consider it as “progressive-identical”. In this way it’s similar to the list of defilements.


Thanks Bhante.

Re your query on sarvaṃ nāsti, I think this and its counterpart sarvaṃ asti are tied to the issues of the source of the world being either what was existent (Sat) or non-existent (Asat) in the Chandogya U, 6.2.

By some perverse logic, even the non-existent is in some form/mode of “existence”, going by CU 3.19.

I think for a clearer differention of the Upanisadic sarvaṃ from the Sarvastivadin sarvaṃ , we can follow Olivelle’s suggestion to take up Gonda’s translation of the former as “the Whole”; this is a singular noun. On the other hand, the Sarvastivadin’s sarvaṃ meant “everything”, an indefinite plural pronoun.

The Upanisadic sarvaṃ also pops up in MN 1 and most of its Chinese parallels. Quite amazingly, the EA parallel seems to have translated sarvaṃ as 悉具足 , echoing Gonda’s suggestion.


AKA Indian logic?