21 December 2018

Dependent Arising: Nidānas

In Part I, I began with a detailed grammatical analysis of the traditional Pāli paṭicca-samuppāda formula. I showed that it tells us that the presence of the condition is required for the duration of the effect. The only traditional Buddhist view compatible with this criterion was Sarvāstivāda. And, to be clear, this means that modern teachings on dependent arising are inconsistent with the formula. Also, this entailed a particular view of time: it has to be linear and infinite in the past. Now, in Part II, I will continue by looking at the traditional connection of paṭicca-samuppāda to the concept of nidāna "basis". Although this is not as easy as it sounds.

Given that all Buddhists these days use the 12 link nidāna model, there are a surprising number and range of variations in the early Buddhist texts and disagreements between the various recensions. See the accompanying diagram for a visual representation of the main variations.

click to embiggen
Not included in the diagram are the numerous versions of the standard links that leave out the early links. A number of suttas have standard links but begin at the six senses (e.g., SN 12:24) or even at clinging (e.g., SN 22.80). The existence of these shorter links led Austrian Indologist (and Nazi), Erich Frauwallner to the conclusion that the sequence must have originally been two shorter sequences that got mashed together (cited in Bucknell 1999). The list might also have started off short and expanded with time.

Meanwhile, Polish Indologist, Joanna Jurewicz (2000), has proposed that the nidānas might have emerged as a parody or even polemic of Vedic cosmogony, an idea that Richard Gombrich (2009: 133 ff) has enthusiastically supported.

In addition to these obvious major variations, Rod Bucknell (1999) has noted many minor variations that are not visible to the casual reader. The different lineages of Pāli texts (i.e., Sri Lanka, Myanmar, Thailand, etc) are often at odds, with some leaving links out and some adding them. Also, there are several examples where two variant suttas in Pāli are represented by a single text in the Chinese Āgama translations, which Bucknell says this indicates a variation emerging in Pāli. However, we cannot rule out that either the Sarvāstivādin redactor or the Chinese Translator tidied up the Āgama versions by removing minor variants, just as a modern editor might do (see, e.g., Bodhi 2000 p.586).

The much vaunted unity of the early Buddhist Canon has been imposed on a riot of different teachings by one or more anonymous systematizers. I don't see this as a problem, per se; it is true of all Canons that they are constructed from less systematic raw material. Of course, one might argue that these are all variations on the theme and that the theme itself constitutes the unity. In response I would point to the inconsistencies between the themes: karma does not gel with dependent arising. If they are the product of one mind, then that person was, at best, an unsystematic thinker.

In the standard view, the texts were passed on orally with high fidelity until being written down. In fact, the fidelity must have been very low, judging by the written canon. Given the internal suggestions of the texts themselves it is likely that they were passed on orally in a wide range of dialects and only standardised at the time of being written down.

If anything, the Pāli Canon feels more like the work of a committee of people. The idea that the different nikāyas form a single collection must be relatively late and specific to the Theravādins. There is no evidence that other sects viewed them as such. The Āgama equivalents were separately translated into Chinese. The Pāli Canon is a compromise. Where conflicts could not be resolved, two or more versions of texts were included. The Sarvāstivādins, as evidenced in their Canon of early texts that survive in Chinese, were both less divided and more systematic, despite working from similar raw material.

This is all worth keeping in mind when we reflect on the nature of the early Buddhist doctrines. However, for the porpoises of this essay I will look at just two of the many variations from this mess. The familiar twelve nidānas and another model that will be familiar to some, the upanisās also known as the Spiral Path.


In this section, I'm going to outline the implications of obligatory presence for our understanding of dependent arising applied to the nidānas.

The twelve nidānas are usually seen as an application of the principle set out in paṭicca-samuppāda. Eviatar Shulman has argued that, actually, the formula applies specifically, and only, to the nidānas. In 2010, I outlined his argument in an essay called A General Theory of Conditionality? At that time, I tended to disagree, but I have come around to his way of thinking. As far as the Pāli suttas are concerned, paṭicca-samuppāda is the nidānas, though the nidānas frequently occur without the formula. It does not appear to be the case that the former is a general principle and the latter one application amongst many. However, nidāna models also exist that appear unrelated to paṭicca-samuppāda, per se. So we may be looking at an uneven composite that developed at the same time as the texts.

First, we may say that the circularity argument from Part I is a blow for the three lifetimes conception of dependent arising. The requirement of presence means that all such arguments are topologically identical to the case where an event is the condition for itself. Any form of cyclic conditionality has only two outcomes: everything always exists or nothing ever exists. As I notated it in Part I, any circularity in which the presence of the condition is a requirement for the presence of the effect logically reduces to: (A if A) and (¬A if ¬A). If we accept that conditions must be present, then we have to accept linear time with an infinite past or an eternally existent condition at the start of time. The universe may have spatial epicycles of creation and destruction, but time itself must be linear. A phenomenon cannot appear in its own past as a condition. A way around this would be to argue that the nidānas are categories of phenomena rather than a specific phenomenon: each instance is unique, but together they form a class of similar phenomena. Whether this rescues the three lifetimes interpretation is moot, but I want to move on because delving into the theory of categories would take me too far from my topic.

The traditional depiction of the nidānas around the outside of the "wheel of time" (kālacakra) is not a workable model of the nidānas. They are not a loop because conditional loops are forbidden by the requirement for presence. Also, death is not the condition for ignorance. If you look at how the teaching is presented in the texts, it's not presented as a loop, either. The cycle is birth and death. The rest of it is an attempt to explain what happens during a life to drive the constant recycling into new lives. Compare the important version of this doctrine in the Mahānidāna Sutta (DN 15), for example, which only has 10 links (I leave this as an exercise for the reader). Arguably, for the unawakened, ignorance is a constant presence rather than something newly arising from moment to moment (I will look at the implications of this shortly).

Presence is problematic for the nidāna model. Take cognition (viññāna) as a nidāna. The imasmim sati formula says that cognition must be present all the time for the rest of the string of conditions to occur. And not only does this not work, it is evidently not the intended meaning of the nidāna. There are obviously times, when we are asleep, for example, when we do not cognise anything.

If we work through the logic, the condition for cognition is karmic volitions (saṅkhārā) which are arguably present, and the condition for karmic volitions, i.e., ignorance (aviijā) definitely is present. Ignorance is a constant for the unawakened; it must be, or they would be awakened. and it must persist for many lifetimes. If ignorance is present then, according to the formula, karmic volitions must also be present, and cognition must follow. Except that we know there are times when cognition is not present even when ignorance is. So the nidāna chain is constructed on a model which does not have a requirement for presence, or indeed which has a requirement for non-presence some of the time.

Things get even worse at the other end of the chain. If birth (jāti) is the condition for aging and death (jarā-maraṇa) then it cannot make any sense at all to require presence. Birth is an event, and not a short lived event, but one which takes place over an appreciable and often considerable time. Labour can last for a day or more. But if the presence of birth is the condition for the presence of aging and death, the death would be instantaneous for every new born. Again, presence here is counter-indicated and some other form of conditionality is required. If birth is the condition for death then we hope to delay it by as long as possible, but the average in the developed world is around 75. Also surely birth is not that significant compared to conception. Iron Age India only had vague and inaccurate conceptions of conception so they could not have come to this conclusion.

The consequence of this is that, whatever tradition and scholars tell us, the paṭicca-samuppāda formula emphatically does not logically apply to the nidānas. Indeed, a further analysis would reveal that more than one kind of conditionality is required to make sense of the nidānas and several schemes involving multiple types of condition were concocted by the different Abhidharmakāras. For example, Theravādins teach about the twenty-four different kinds of conditionality that their Abhidhamma speculates have to exist to make it all happen. Sarvāstivādins got by with just four. But most of the time the nature of the conditionality is unspecified, which allows us to have a more intuitive but less precise account of conditionality.

We do not teach the nidānas with the requirement for presence despite the clear implication of the paṭicca-samuppāda formula. There is nothing wrong with teaching a story about conditionality that we think works, if the alternative is one that we know doesn't work. But it seems disingenuous to promote this as something the Buddha taught. The whole concept is rather dubious, anyway, as we cannot even say if the Buddha was a historical figure. Even if we stipulate this much, however, it is clear that the early versions of the doctrine are incoherent and later Buddhists disagreed on how to fix this. What we now teach has no connection with the Buddha, it is "Buddhist" pragmatically because it is something that Buddhists say, not because it was something the Buddha said.

This is all bizarre enough, but there is a related, though largely neglected early Buddhist doctrine which is a much better fit for paṭicca-samuppāda, i.e., that also requires the presence of the condition.

The Spiral Path

click to embiggen
The lesser known model of conditionality was rediscovered by Carolyn Rhys Davids while she was editing and helping to translate the Saṃyutta Nikāya for the Pali Text Society. The idea was taken up by Sangharakshita but has never really gained much traction outside of the Triratna Buddhist Order, where it is an important doctrine. Both Ayya Khema (1991) and Bodhi (1980) wrote about it but just the once each, and some time ago now. Sangharakshita called it the "spiral path" (though in his teaching it is not a spiral, but a helix). The Nettiparakaraṇa, an early Pāli commentary included in the Canon, refers to it as lokuttara paṭicca-samuppāda which Bodhi translated as "Transcendental Dependent Arising". Other terms such as "progressive nidānas" have been used. As nidāna "basis" is used to refer to the paṭicca-samuppāda links, I have suggested that the spiral path be referred to as the upanisā "preconditions".

The PED is unsure about whether to derive the word upanisā from upa-ni-√sad, in which case it would be the same as Vedic upaniṣad, or from upa-ni-√śri (P. upanissayati), in which case we would have to see the word as an abbreviated form of the gerund upanissāya "depending on, by means of". Given how the word is used in the upanisā doctrine, the latter is far more likely. Compare BHS upaniśritya. In the Chinese Samyuktāgama (SA 495) the word is translated 所依 (T 1: 2.129a11), which is commonly used for Sanskrit terms such as āśraya and niśraya (ā√śri and ni√śri respectively) both meaning a "basis" or "foundation".

The upanisā doctrine occurs in about 40 suttas, whereas the imasmim sati formula occurs just 13 times, and the two are never related in Pāli. The scattered references were collected into one section of the Sarvāstivāda Madhyama Āgama, which is preserved in Chinese (see my draft translations of MĀ 42-55, Taishō 1: 26 §5).

In the spiral path, the upanisās are cumulative, with earlier conditions needing to be present, and thus  it is consistent with the paṭicca-samuppāda formula. Although existing explorations of the doctrine have focused on the Upanisā Sutta (SN 12:23 ≈ MĀ 55), my view is that the suttas at the beginning of the Chapter of Tens/Elevens in the Aṅguttara Nikāya are the template.
  • Kimatthiya Sutta (AN 10.1 = AN 11.1 ≈ MĀ 42)
  • Cetanākaraṇīya Sutta (AN 10.2 = AN 11.2 ≈ MĀ 43)
  • Paṭhama-upanisā Sutta (AN 10.3 = AN 11.3 ≈ MĀ 44, 47)
  • Dutiya-upanisā Sutta (AN 10.4 = AN 11.4 ≈ MĀ 48)
  • Tatiya-upanisā Sutta (AN 10.5 = AN 11.5 ≈ MĀ 43)
Seen in the light of the broad range of other spiral path texts, the Upanisā Sutta is an oddball, albeit an interesting one since it is the only text that attempts to join the nidānas with the upanisā. The usual nidāna sequence is listed to birth, which is followed by dukkha, then the upanisā sequence beginning with faith (saddha). However, this is a very problematic conjunction. Linking faith to suffering as a condition is prima facie strange but more so when one realises that the usual condition for faith is hearing the Buddha preach. Explanations for this conjunction typically add three extra steps in the process of linking the two. Even in the Triratna Order, where we actively teach the upanisā doctrine, we do not utilise the text as it stands. Instead, we use Sangharakshita's revision. Whether it even makes sense to combine the two models at all is doubtful.

The pattern from AN 10.2 begins like this:
The virtuous one, monks, endowed with virtue, need not form an intention ‘may my conscience be clear.’ It is natural (dhammatā) for the virtuous one endowed with virtue to have a clear conscience. Having a clear conscience, there is no need for an act of will ‘may I feel joy.’ Joy naturally arises in those who have a clear conscience.
The dynamic is illustrated in the Upanisā Sutta and several of the Madhyama-Āgama texts with a simile:
"Just as, when the gods pour down rain over the mountains, water flows down the mountainside, filling up the branches of the crevices and gullies; having filled the crevices and gullies, small lakes and the great lakes are filled; the great lakes being filled, the small rivers fill up; the small rivers fill up the large rivers, and the large rivers fill up the great ocean."
 Again, this is consistent with the continued presence of the condition giving rise to the effect and not with the sequences of the nidāna doctrine.

As I showed in my first published article on the subject (2013), the upanisā doctrine is a more elaborate version of the threefold path of conduct, concentration, and insight (sīla, samādhi, and paññā). The factors leading up to pāmojja constitute what we usually call "morality", but which are more about creating the conditions for meditation, rather than being a good person or following group norms. Hence, I would now say that "conduct" is a better translation than "morality" or "ethics", both of which are too broad in this context. The stage of conduct involves accustoming oneself to reduced sensory stimulation through restricting one's exposure to sense objects. It is characterised by restraint and renunciation. This is supposed to lead to a general mood of uplift and happiness (pāmojja) which is the prerequisite for meditation.

Having crossed the threshold of pāmojja as one becomes concentrated in meditation (and here we can infer that jhāna meditation is intended) one passes through a series of phases of increasingly refined experience, with less reference to external sense objects. This is not the place to argue about the general applicability of the model (I have my doubts) but we can say that at least this doctrine is consistent with the doctrine of paṭicca-samuppāda, although the Buddhist tradition seems to have kept apart the two doctrines that do make sense together, and to have largely forgotten the upanisā doctrine (or at best to have let it fall out of use).

What's more, it is apparent that the upanisā list aligns with the bojjhaṅga or "factors of awakening" list. For a graphic representation of this see my big dependent arising diagram.

The conclusion of this section is that the dynamic in the paṭicca-samuppāda, which requires the presence of the condition, is not the dynamic of the nidānas. In fact, the requirement for continued presence is incompatible with the nidāna model. And this despite Shulman's observation that the wording suggests that Buddhist authors saw the nidānas as spelling out in detail the principle of the formula. By a strange twist, the requirement of presence is exactly what we find in the neglected and sidelined upanisā doctrine. The two are not linked in surviving texts (Theravāda or Sarvāstivāda) except the Upanisā Sutta, which cobbles the two together end to end and this could not possibly work (and is not used in practice).

This shows, I think, that the received texts have been quite heavily and ineptly edited to favour an interpretation of doctrine that does not make sense, despite there being a combination that did make sense. This suggests that religious ideology has trumped common sense. It also shows that what appears to be unity to some people, is not apparent on closer inspection. 


There is still one issue with the Spiral Path model to deal with. As before, the problem is explaining cessation. The conditions for rebirth are present and are the reason that we are continually reborn. In order to eliminate rebirth we have to eliminate the conditions for rebirth. But if presence is the norm, then this would involve an infinite regress: conditions must be present and the past must be infinite. In order to wholly eliminate rebirth we would have to go back to the base condition for rebirth, which is the existence of the universe that allows for rebirth. With infinite time we could never get to the beginning of the sequence to remove that ultimate cause. And thus rebirth could never cease because the preconditions for rebirth would always be present. This paradox tells us there is something wrong with the model.

Could it be that we have it all wrong and that the formula expresses a different dynamic? Is is more like a cue hitting a snooker ball? Or a line of dominoes? A seed growing into a tree?

Let's take the first idea. As a general approach it suggests that the condition gives the effect a "push" that enables the effect to be present in the absence of the condition. But if the effect can be present in the absence of its condition, then it is the condition for its own presence and we have explored this scenario in detail already. Alternatively, we may argue that the condition bequeaths a temporary quality of momentum to the effect. But what is the condition for the continued presence of this momentum? Nothing can be unconditioned except nirvāṇa (and perhaps space, ākāśa). Clearly, at the outset, the presence of momentum is due to the condition itself. But the momentum has to be an effect of the condition and dependent arising says that when the condition ceases the effect ceases. So there is no way for the condition to pass on anything to the effect by way of momentum if the condition is not present. The requirement for presence cannot be subverted by adding extra steps, because if any effect can outlive its conditions, under any circumstances, then it has become a condition for its own presence.

In the line of dominoes analogy, the suggestion is that the condition changes its state in such a way as to force the effect to change its state in the same way, creating a cascade. The problem is that at the outset a line of dominoes are all present and are simply knocked over. They don't cease to be present once they have fallen over. So at the beginning and the end we have dominoes. There is no help from this analogy, although this is similar to the Sarvāstivāda view of constant presence combined with a changing state of activity. 

A traditional image that is sometimes use to try to illustrate dependent arising is the seed sprouting and turning into a tree - continuous change. This is an argument from analogy which is not an explanation. But let's consider it. What is the condition for the seed to sprout? We may say that it must grow in soil (a complex mix of organic and inorganic elements) and at a minimum be watered. What happens to the soil once the seed sprouts? Nothing much. It has to continue to be present. Similarly, water must be continually present in some form or the plant dries out and dies. One might argue that this analogy is considerably worse that others we have considered.

I cannot think of any other situation in which we take dependent arising at face value and arrive at a workable solution. The only logical conclusion is the sarva-asti-vāda. Conditions all exist all of the time because if they did not everything would cease. Indeed, if any past condition ceased then all of the downstream effects related to it would instantaneously also cease. Of course, this is not what we see when we look at the world. Rather, we see phenomena arising and ceasing.

Usually Buddhists take what they see to be what they are supposed to see and assert that this is what dependent arising shows. Clearly, this is not what the formula of dependent arising says. Rather, it says a condition must be present for the entire duration of an effect. We do not teach dependent arising according to canonical accounts, but substitute what seems intuitively correct to us. And if we are allowed to do this with our supposedly central doctrine, then what changes are we not licenced to make?

Contrarily, if we insist that the canonical account is the Buddha's verbatim teaching, then we have to admit that the Buddha screwed up his most distinctive teaching or those who attempted to preserve his teachings were hopelessly confused by the time they came to be written down. Either way it seriously undermines the case for the supposed authenticity of the early Buddhist teachings.


It seems to me that the people who promote dependent arising as a metaphysical truth have really not thought deeply enough about it. Even taking dependent arising on its own terms we end up at the sarvāstivāda position that all past conditions must still exist. This is not a position I advocate, it is simply the inescapable logical conclusion of dependent arising as stated in the imasmiṃ sati formula. This was, in fact, a hugely popular approach in classical Buddhism in India and in China. The Sarvāstivāda Abhidharma books are all preserved in Chinese. While every Buddhist (presumably) accepts some form of conditionality doctrine, no living Buddhist actually holds this view.

The imasmim sati formula does not describe the conditionality of the nidāna doctrine, which is not, in any case, a single kind of conditionality. On the other hand, the formula does describe the upanisā doctrine, which can be seen as extending the threefold path (conduct, meditation, wisdom) and as consistent with the bojjhaṅga doctrine as well. It's just that the formula and upanisā doctrine are never linked in early Buddhist texts. No traditional school continues to promote the upanisā doctrine and its importance has largely been lost sight of.

Modern Buddhists may wish to argue that conditionality is not linear but that everything conditions everything else. Certainly this is a common Buddhist assertion. It has the disadvantage of gaining no support whatever from early Buddhist texts. However, many innovations in doctrine emerged over time to fix exactly this kind of problem, so this disadvantage is not fatal. Does it get us out of the bind? It does not, since, if everything is a condition for everything else, then this reduces to events or entities being the condition for their own existence. And again, this is not what we see when we observe conditionality: we see arising and ceasing. Any explanation we pose must at the very least allow for arising and ceasing. We see multiple conditions, but not infinite conditions. 

Should we be surprised that an Iron Age religious doctrine doesn't stand scrutiny? Hardly. I cannot think of a contemporary idea from anywhere in the world which has survived contact with modernity. Outside of academia no one gets excited by Plato and claims that he discovered the nature of reality. Platonic Idealism is trotted out to confuse philosophy undergraduates, but plays no great role in how modern intellectuals think about the world. Equally, no one in Britain thinks that our lives are controlled by the whims of Tiu, Woden, Thor, and Frigga any more. It's not credible to keep arguing that outdated theories tell us about the nature of reality.

What is really worrying is how much of academia is involved in trying, for example, to reconcile karma with modernity. I'd like to do some research on this, at some point. As an interim measure I think all believers in Buddhism should declare a conflict of interest with writing about Buddhism so that readers can be alert to unexamined assumptions. Editors should make it clear when someone is writing as an apologist or theologian rather than a scholar. I think I'll start adding a disclaimer to all my articles from now on.

So far, I have tried to avoid any argument rooted in modernity; even the logic I refer to is of a similar age to the doctrines themselves, if from another culture. I've tried to show that Buddhism is fundamentally incoherent on its own terms: the logical conclusions are avoided in favour of intuitive conclusions. Of course, it is much worse when we consider Buddhism in the light of modern philosophy and science. The painful thing is that we have a very useful practical approach that seems to benefit people; we just don't have a coherent explanation of what we are doing or why.

In the final instalment, I revisit the argument that our problem is treating doctrine as metaphysics and  that the solution is to shift the discussion to epistemics. Buddhism has a veneer of respectability as metaphysics but it is thin and getting thinner. The modern world is going to figure us out at some point, so we Buddhists need to get ahead of the curve in order to survive. We have nothing coherent to say about the nature of reality, but the same statements read as commentary on our paradigm make more sense and are sometimes compatible with modern knowledge. For the long term survival of Buddhism as a meaningful cultural movement, we need to come to terms with this issue.



Bodhi. 1980 Transcendental Dependent Arising: a Translation and Exposition of the Upanisa Sutta.(The Wheel Publication no.277/278.) Buddhist Publication Society. Online: http://www.accesstoinsight.org/lib/authors/bodhi/wheel277.html

Bodhi. 2000. The Connected Discourses of the Buddha : a Translation of the Saṃyutta Nikāya. Boston : Wisdom Publications.

Bucknell, Roderick S. 1999. "Conditioned Arising Evolves: Variation and Change in the Textual Accounts of the Paṭicca-samuppāda Doctrine." JIABS 22 (2) 1999: 311-342.

Gombrich, Richard. 2009. What The Buddha Thought. London: Equinox.

Jayarava. 2012. Chinese Spiral Path Texts from the Madhyāgama. Draft Translations. Sept 2012.

Jayarava. 2013. "The Spiral Path or Lokuttara Paṭiccasamuppāda." Western Buddhist Review 6: 1–34.

Jayatilleke, K. N. 1963. Early Buddhist Theory of Knowledge. Motilal Banarsidass, 2010.

Jurewicz, Joanna. 2000. "Playing with Fire: The pratītyasamutpāda from the perspective of Vedic thought." Journal of the Pali Text Society 26 (2000) pp. 77 – 103.

Khema, Ayya. 1991 When the Iron Eagle Flies. Penguin.
Related Posts with Thumbnails