15 January 2021

Modern Interpretations of the Khandhas: Saññā

Continuing my series exploring the Pāli khandhas in the early Buddhist texts as they are interpreted in two books published in 2000: Tilmann Vetter's The Khandha Passages in the Vinayapiṭaka and the Four Main Nikāyas and Sue Hamilton's Early Buddhism: A New Approach. In addition to these main sources I have also started consulting Rupert Gethin's article: “The Five Khandhas: Their Treatment In The Nikāyas And Early Abhidhamma” (1986).

So far I have covered rūpa and vedanā. My verdict on each was that both authors' accounts of these terms are problematic. This is partly because of the inordinate weight put, quite uncritically, on the Khajjanīya Sutta (SN 22.79). Because there is a dearth of explanation of the khandhas in Pāli, this one sutta that appears to explain them is given privilege. The definitions in the Khajjanīya Sutta turn out to be untrustworthy, but this has not stopped them being given pride of place by modern commentators. However, explanations of the khandhas do not reflect the general usage in Pāli. If anything, a picture that was vague to begin with becomes more deeply obscure.

I suspect that some of us cannot shake the tendency to think of ancient Indian Buddhists as akin to the Greek philosophers. I think we imagine sitting around trying to reason about the world or just discussing it and coming up with grand narratives explaining life, the universe, and everything. We do have accounts of group discussions so they probably did happen, but Indian religieux also pursued a religious lifestyle and this included the religious exercises peculiar to India that we now refer to with the misnomer meditation. A lot of what we read about in Pāli suttas is an idealised account of this lifestyle: men and women leaving behind all social responsibilities and domestic life to pursue the deathless.

It's a mistake to think of the khandhas as the result of metaphysical speculation. The idea that early Buddhists were even interested in existence is on the wrong track. It's clear that Buddhism (and Indian religion in general) is underpinned by altered states of consciousness experienced in meditation, especially by the complete cessation of sense experience while remaining conscious. If Buddhists were speculating about anything, it was about the significance of cessation, the absence of sense experience, and as a corollary, the nature of sense experience. There is no talk of "reality" in early Buddhism (at least in Pāli) so far as I can see. There is a lot of talk about experience.

Rupert Gethin points out that in these sources: upādānakkhandhā = dukkha = loka = satta = ajjhattika-āyatana = sakkāya

"All these expressions apparently represent different ways of characterising the given data of experience or conditioned existence, and are also seen as drawing attention to the structure and the sustaining forces behind it all" (1986: 42).

The caveat is that there is no term that means "conditioned existence". We can say, categorically, that saṅkhāta is compounded with only one word in the Pāli suttas, saṅkhāta-dhamma: there are conditioned dhammas (saṅkhāta-dhammā) but no Pāḷi term that means conditioned existence. Moreover, dhammas are the object of the manas or mind-sense. As I have argued elsewhere, conditioned dhammas refer to sensory experience and the singular unconditioned dhamma (asaṅkhāta-dhamma) is the state of absence of sense experience in which all the conditions for experience are absent (suñña), i.e. nibbāna. This is the broader context in which we have to think about khandhas. In other words we should be in the overlapping domains of phenomenology and epistemology: accounts of the phenomena and knowledge about phenomena, as opposed to accounts of noumena or existence. And this goes double for terms that clearly refer to some mental capacity or function, such as the next khandha, saññā.

By "conditioned existence" Gethin might have meant saṃsāra the rounds of rebirth. But in early Buddhism this is governed by a different, incompatible process, i.e. karma. Dependent arising asserts that there can be no delayed consequences (something Nāgārajuna picks up on centuries later) and karma is all about delayed consequences. Of course, some tried to explain karma in terms of dependent arising, but providing the necessary continuity for karma was always a problem because it always contravened the limits imposed by dependent arising.

In any case, let us now consider saññā.


In defining saññā, Vetter continues to draw on the problematic Khajjanīya Sutta (SN 22.79) by asserting that "saññā is derived from sañjānāti" (24). It might be better to say that, in this rather odd and unrepresentative sutta, the word saññā is explained by the verb sañjānāti and that the two words are derived from the same affix-verbal root combination: saṃ√jñā. The verb sañjānāti typically refers to the act of recognition and putting a name to something. The Khajjanīya Sutta elaborates on this theme of recognition by naming four colours. And in the light of psychology this invokes the more involved idea of perception (rather than naming) or in Hamilton's case apperception, i.e. "perception in the cognitive rather than in (just) the visual sense" (76). Gethin, too, references the colours as part of the definition.

Note that the colours named are the four "basic" colour terms in Pāḷi: nīla "dark/black", pītaka "yellow", lohita "red/brown", and odata "light/white". And this deserves a digression. Here "basic" is a jargon term with a very specific, if somewhat contested, meaning, the definition of which takes up a whole chapter in C. P. Biggam's The Semantics of Colour: A Historical Approach (2012). Since Biggam identifies nineteen potential headings under which to discuss the concept of basicness we could easily get bogged down. I will give a simple definition and suggest readers consult Biggam for a more comprehensive discussion of this fascinating idea. For our purposes:

A basic colour term is an adjective that is solely used to describe colours, it is at the top level of the taxonomy of colour terms (there is no broader category other than "colour" into which the basic term fits), it is not a recent loan word, it is not a compound word and does not rely on affixes to convey meaning, and in a language-using community basic terms are understood and used by everyone for the same purpose.

With respect to the basic colours in Pāli, note that nīla does not mean "blue" until considerably later, here it means "dark" and this takes in black, blue, green, brown, grey, etc. Thus, while a blue object would be labelled nīla, not all nīla-coloured objects would be blue. The Sanskrit of this period also has just four basic colour terms, as does Homeric Greek. See also my essay Seeing Blue (6 Mar 2015). Colour terms evolve in predictable ways, so that if a language has four basic colour terms one of the expected lines of development is to first split off red from light-coloured and then to distinguish red and yellow. And we confidently predict that the next colour to be added will be blue or green. And so on. So the colours named in the sutta are not arbitrary or random. We have eleven basic colour terms in English: black, white, red, yellow, green, blue, brown, purple, grey, pink, orange. I suspect that, like some other languages, English begins to distinguish blue from cyan at a basic level.

Vetter assumes that he has a reliable definition and then leaps to the conclusion that the definition derived from the Khajjanīya Sutta "seems to presuppose some knowledge of a description of a kasina." (24). This is based on the appearance of the four basic colour terms (sans any insights into colour terminology) and of the verb sañjānāti,which occurs over 100 times in the suttas, often in the sense of "he recognises X as X" (cf. Mūlapariyāya Sutta MN I.1). Vetter leaps ahead, yet again, to suggest that this means "he imagines X as X". And I frankly do not follow the reasoning for this second leap. Nor do I follow his reasoning when he leaves behind the context of the khandhas and discusses saññā in a completely different context without making any distinction. We already know that the meanings of these words are sensitive to context. Vetter's citations of the Suttanipāta (25) are for example in the context of a person abandoning views. E.g. 793: "He becomes dissociated from all dhammas that are seen, heard, or thought (mutaṃ)." (which also seems like a description of the absence of sense experience, doesn't it?). If we are discussing the khandhas then we ought to stick to texts that deal with the khandhas, unless the case is being made that the use with respect to the khandhas is just the ordinary usage. Vetter does not make this case.

Saññā is yet another term that changes its meaning depending on context and over time. Like words such as manas, citta, and viññāna, saññā can just be a general word for mind or mental activity. Vetter suggests that Johansson has it right when he summarises the relevant mental activities as "ideation" (25) and Gethin supports Alex Wayman's suggestion that it means "idea" (1986: 36). Again, this seems to be based on the word in a different context. Moreover, even Gethin is too reliant on the Khajjanīya Sutta and on the modern psychological interpretation of the significance of the colour terms when it comes to saññā. When one says of an object that it is blue, one is not having an idea, one is putting a name to a quality. Perhaps this does involve conceptualisation, but that conceptualisation is transparent to the person and would not have been obvious to early Buddhists. I see no evidence that the authors of the Pāli text generally had any insight into this abstract way of thinking about perception. We have to remember the the terms developed in Iron Age India long before sophisticated psychology and theories of consciousness emerged. All the authors I'm considering in this essay seem to be projecting modern ideas backwards and making an anachronism out of saññā (another example of the anachronistic fallacy).

A deeper sense of incoherence emerges when we see an alternative set of three khandhas in the Mahāvedalla Sutta (MN 43) where they are also matched to verbs from the same root: viññāṇa/vijānāti, vedanā/vedeti, and saññā/sanjānāti. The sutta tells us that viññāna cognises (vijānāti) pleasant, painful, and neutral; Vedanā experiences (vedeti) pleasant, painful, and neutral, while saññā recognises (sañjānāti) dark, yellow, red, and white (Note the first three colours are given in the form nīlaka, pītaka, lohitaka, where -ka can be adjectival with a possessive sense or diminutive). Here, the three factors are in a different order than we are used to meeting them. The text tells us, however, that they are conjoined and inseparable (ime dhammā saṃsaṭṭhā, no visaṃsaṭṭhā. MN I 293) and this is because "What one feels, that one recognises; and what one recognises, that one cognises." (yaṃ vedeti taṃ sañjānāti, yaṃ sañjānāti taṃ vijānāti). And this is the order we meet in the khandha passages.

But this is strange, because what one feels is sukha/dukkha/asukhamadukkha, what one recognises is colour, and one what cognises is again sukha/dukkha/asukhamadukkha. Two of these form a related set and one is completely unrelated, unless the colours are somehow a code for combinations of sukha/dukkha. If saññā really is the recognition of colours then the khandhas are incoherent at this point. This is reinforced when we think of saṅkhāra as karmic actions (obviously, I'm jumping the gun here a little), because karmic actions involve grasping the pleasant (sukha) and shunning the unpleasant (dukkha). The khandhas appear to narrowly miss out on having sukha/dukkha as a unifying idea: the appearance, feeling of, recognition, reaction to, and discrimination based on, the pleasantness and unpleasantness of sensory experience.

Hamilton, as noted, translates saññā as "apperception" and associates with the action of "identifying" (76). "What one is doing in this process, according to the texts, is making manifold and naming what one is experiencing." (76). Unfortunately, with the terminology of "making manifold" Hamilton is at her least clear and her usual practice of referring to which texts she is referring breaks down. Hamilton appears to insist on a metaphysical interpretation of dependent arising against her own stated view that dependent arising describes experience (which is an epistemic or phenomenological interpretation). She appears to believe that dependent arising entails an undifferentiated world in which nothing is distinguished, and that identifying individual objects in this undifferentiated mass is "making manifold", Hamilton's translation of papañca. Papañca is in fact another tricky and under-defined concept that I have studied in some depth in two essays from 2012: Translating Papañca and Understanding Prapañca.

What Hamilton does not say is that the texts that use the word papañca are confused. There are four distinct ways of talking about papañca:

  • papañca as the sum total of the perceptual process: e.g. in M 18, S 35.94 A 3.294, Sn 4.11, Sn 3.6
  • papañca as metaphysical speculation: e.g. in A 4.173
  • papañca in relation to 'I am': e.g. in S 35.248, Sn 4.14
  • papañca = kilesas e.g. S 35.248, SA 2.381 (commenting on S 35.94). UdA (commenting on Ud 7.7)

Moreover, when it comes to the process, the term refers to three distinct sequences of terms that appear to describe a process and they all disagree on the order in which the process happens. That is:

  • Sn 4.11: papañca = saññā → nāma & rūpa → phassa → sāta & asāta → canda → piya → macchara etc.
  • M 18: rūpa + cakkhu + cakkhu-viññāna → phassa → vedanā/vedeti → sañjānāti → vitakka → papañca.
  • D 21: papañcasaññāsaṅkhā → vitakka → chanda → piyāppiya → issā-macchariya → verā etc.

No one can say definitively that papañca is one thing or the other. There is no overarching definition that fits every mention of it and Hamilton has simply chosen the definition which best fits her theory. As much as I find her theory useful, I don't find this aspect of it at all convincing. As I said in 2012: "What my study seems to say is that the ambiguity of papañca allows it to be co-opted to suit the agenda of the commentator.". Back then I was still puzzled by the lack of coherence in a Buddhist doctrine. Since then I have realised that incoherence is the norm.

Given Hamilton's emphasis on experience and the lurch into metaphysics at this point (undifferentiated reality vs objectified experience) her work here becomes practically useless. I retain enough naivete to find this quite disappointing.


I think colours associated with saññā in the context of the khandhas are a red herring. The colour words trigger particular associations in a modern reader that seem unrelated to how people in Iron Age India conceived of their world. We can be sure that they did not think in terms of psychology. We know that they didn't think of the mind at all like we do, as I show in my essay The 'Mind as Container' Metaphor.

There is a distinction between inventing a way of reading texts (a hermeneutic) that makes sense on our terms and seeking to understand how the authors understood the texts. The former ignores the meaning of the text qua text and opts for something that fits modern preconceptions. This is the approach of all the modern Buddhists that I know. Stepping outside this modernist insider approach to really read the texts and to unearth the mode of thought of the authors is difficult and perhaps ultimately impossible. However, the textual scholar is bound to try to do this and to highlight how they go about it.

Sanskrit dictionaries tell us that saṃjñā means "agreement, mutual understanding; consciousness, understanding; sign, token, signal; name, appellation"; while the verb means "to agree, to be of the same opinion; to appoint, assign, to acknowledge, to recognise; to claim, take possession". So the etymology offers a rich semantic field from which to choose, but "ideation" and "idea" don't really come up. At the outside edge of this we could imagine "perception" but really only if we employ psychology anachronistically. Of course, we have to avoid the etymological fallacy and keep in mind that the use may be unrelated to the etymology. Still, the usage still has to make some kind of sense.

We seem to come back to recognition as the principle idea associated with the word and the most likely application in this context also. The problem is that this is the conclusion of none of my informants. This feels a little awkward. The influence of the Khajjanīya Sutta on modern discussions of the khandhas continues to be problematic. Why was it not read critically by scholars? It seems that in the absence of a precise and coherent explanation this text was adopted as the next best thing. But it cannot carry this weight.



Biggam, C. P. 2012. The Semantics of Colour: A Historical Approach. Cambridge University Press.

Gethin, Rupert. 1986. “The Five Khandhas: Their Treatment In The Nikāyas And Early Abhidhamma.” Journal Of Indian Philosophy 14(1): 35-53.

Hamilton, Sue. 1996. Identity and experience: the constitution of the human being according to early Buddhism. London: Luzac Oriental.

Hamilton, Sue. 2000. Early Buddhism: A New Approach. London: Routledge.

Vetter, Tilmann. 2000. The Khandha Passages in the Vinayapiṭaka and the Four Main Nikāyas. Wien Verlag der Österreichischen Akademie der Wissenschaften.
Related Posts with Thumbnails