24 March 2023

The Emptiness of Etymology

It is de rigueur to begin an essay about a Buddhist technical term with an etymology: a discourse on the "true" or "original" (etymos) meaning of the word by referring to historical usage. I have often done this myself because I learned to write academic essays by reading hundreds of them and it's a very common practice. Constructing etymologies is fascinating and fun, it's just that this scholastic approach is alien to most language users. One could be completely ignorant of historical linguistics and still be fluent in a language. As most children are, for example.

It makes little sense to offer an etymology up front as part of defining a term since no language-user ever defined words that way. Nor does etymology necessarily tell us anything at all about how words are used at any given time (even "originally"). Also note that the first recorded use is necessarily a reference to written language which is likely a poor reflection of how a language was spoken. Moreover, in Buddhism Studies our texts are religious texts composed by religieux for religieux and they are full of words used as religious jargon. Buddhist jargon was often created in defiance of etymology: e.g. vedanā or dhāraṇī.

This begs the question: How do we define words? And, in the light of the answer to this, how should we present the meaning of words when writing about them?

As Wittgenstein noted:

“For a large class of cases of the employment of the word ‘meaning’—though not for all—this word can be explained in this way: the meaning of a word is its use in the language” (Philosophical Investigations 43).

This is sometimes boiled down to the rule of thumb: meaning is use. The meaning a word conveys doesn't come from the word itself, it comes from how we use the word in language (see also the Humpty Dumpty theory of language). The general term for this approach to speech is pragmatics. Apart from Wittgenstein, this approach is also particularly associated with two philosophers of language: John L. Austin (1911-1960) and (his student and collaborator) John Searle (1932 -).* In a pragmatic framework we often don't ask what a word means, we ask what an utterance (or locution) was intended to do (illocution) and what it actually did (perlocution). The focus is less on words, and more on sentences.

* NB. Searle (now a nonagenarian) was a fine philosopher and hugely influential in thinking about language, mind, society, artificial intelligence, and much more. He is one of the few philosophers I've enjoyed reading and has been a big influence on me. However, in 2017 a series of allegations of sexual harassment and sexual assault emerged. Searle's emeritus professor status was revoked by UC Berkeley although, far as I can tell, he has not been charged with or convicted of a crime. Still, his legacy is severely tarnished by these allegations.

Philologers are concerned primarily with words in written language. For all that writing has a way of formalising language, in fact as we start to deal directly with ancient manuscripts we soon discover that no two of them are identical. And this leaves us with the awkward task of reconciling multiple competing versions of the text. The philologist has to assess which of the plurality of possible readings is the most plausible one. But the assessment relies heavily on the knowledge and experience of the editor and thus often involves subjectivity.

As my readers will know from my work on the Heart Sutra, editors sometimes make mistakes or deliberately mislead us about texts. Edward Conze is an obvious example of what can go wrong: his editions are full of mistakes (in Sanskrit grammar and syntax) and misreadings that made his idiosyncratic theory of Prajñāpāramitā more plausible. I still can't decide if he knew what he was doing or not. I suspect he did.

Authors may also allow different versions of their works to circulate, leaving philologists to argue over which is better. There are two editions of Kant's Critique of Pure Reason and they are substantially different. Clearly the author preferred the second reading but scholars sometimes think he said it better the first time around. We are not bound to take author's view of what the text says and remakes are notoriously inferior. Moreover, scribes make mistakes and these can be incorporated into the text, as has happened to the Tibetan Heart Sutra text.

In any case, even when we have a settled and reliable text, we may still encounter unfamiliar words quite often. And these present a problem, especially with an oral text, no reference books, and the spoken language has drifted for a millennia, which was the situation faced by Ṛgveda scholars in the ancient world.

Unfamiliar Words in the Ancient World

Many people will know that modern linguistics was kickstarted when European intellectuals encountered the Aṣṭādyāyī, a complete descriptive grammar of Vedic produced by Pāṇini ca 400 BCE. Pāṇini was an advocate of the analytical (vyākaraṇa) approach to what words mean, which involves analysis of the construction of words and identifying verbal roots as the principal bearer of meaning.

Alongside this was a less well known tradition of linguistic analysis known as nirvācaṇa or "semantic analysis". This approach is associated with the figure of Yāska and is contained in his work the Nirukta, which is a commentary on a list of rare and unusual words in a text called the Nighaṇṭu. My account of the Nirukta is largely based on Eivind Kahrs (1998) and some articles by Paolo Visigalli (2017) and Johannes Bronkhorst (1981, 2001).

For Yāska, as for Pāṇini, the name of a things derive from its activities. Kahrs points out that proposition is conceptually more fundamental that deriving nouns from verbs. This is why explaining a noun by referencing a verb is seen as meaningful. Yāska asks "Why is dawn (uṣā) so called?" And answers: uṣā ucchati "Dawn dawns"; That is, "dawn is called 'dawn' (uṣā) because she carries out the action of dawning (ucchati)." Here the verb ucchati explains the noun uṣā by linking it to an action. (See also my essay on what explanation is and what makes for a good explanation)

The Nirukta proposes three levels of analysis. Durga's commentary on the Nirukta refers to these as the cases of pratyakṣavṛtti "evident formation", parokṣavṛtti "obscured formation", and atiparokṣavṛtti "opaque formation" (c.f. Kahrs 1998: 35-39; Visigalli 2017: 1148).

Evident Formations

Firstly, there are obvious examples where grammatical analysis (vyākaraṇa) gives us all the information we need. For example, the common noun dharma is regularly derived from the root √dhṛ "hold" by adding the primary suffix -ma to the root in its guṇa grade (i.e. dhar). If you know the morphology and you can identify the root, then you can parse the unfamiliar word.

Another good example for Buddhists might be the word buddhi. This is not one of our technical terms so might not be familiar. But it looks a lot like buddha, the past participle of √budh "aware, awake" with the suffix -ta and a series of entirely regular sound changes. Similarly, I can parse buddhi as √budh with the action noun suffix -ti and guess that the resulting word means something like "awareness, waking". Actual usage provides the context that allows us to refine this guess to something appropriate to the specific occurrence.

Obscured Formations

Secondly, some words are tricky because they are irregular formations. Consider the Pāli word vedagū (see also Some Notes on -in Vedagū). The veda part is easy and tractable with a standard vyākaraṇa approach. Veda is from the causative stem of the root √vid "to know".

On the other hand, is not a word in Pāḷi and it's not a word in Sanskrit. Nor is it obviously derived from a known word (verb or otherwise) in the Indo-European language family. On the other hand it does not appear to be a loan word either. And yet we have not only vedagū but a bunch of other words ending in -gū like lokantagū (an epithet of the Buddha). When we look at how the word is used it becomes apparent that - must be related to words meaning "go" and thus can be explained as some kind of alteration of √gam or √ which is closely related. Moreover we can see that a few other words that ended in -ā got changed to -ū, like kataññū or viññū. With some effort, then, we can see that vedagū means much the same thing as *vedagā or vedagata, i.e. "in a state of knowing", "knowledgeable". Still, the form -cannot be arrived at by an allowed phonetic changes applied to a known form or root. Therefore - cannot be explained using vyākaraṇa. I have argued that is an example of an analogical change in Pāli (though the JPTS was underwhelmed by my argument).

Opaque Formations

Thirdly, and most importantly for Yāska, there are very obscure examples which defy logical grammatical analysis entirely. It seems that there were a large number of unique words (hapax legomena) in the Ṛgveda whose meaning was already obscure by the time Yāksa wrote his treatise. Yāska was focussed on this class of words that didn't yield to standard linguistic analysis of the Pāṇinian vyākaraṇa tradition. Presumably the hundreds of loanwords from Dravidian and Muṇḍa languages in Ṛgveda (c.f. Witzel 1999) complicated this process since they are never traceable to a Vedic verbal root.

Take the example of śraddhā "faith". Vyākaraṇā suggests a root √śrad but this is unknown in Sanskrit. We can see from context that the word is understood as referring to the heart but the standard word for this is hṛd. The only other related word is śrāddha "death rite". There are no Sanskrit verbal forms related to śrad or hṛd. So we can't easily define śraddhā in terms of an action. What tends to happen however, is that scholars note that dhā appears to relate to the root √dha "to place" and the definition often includes a note about faith being "placing the heart".

Thus, Yāska was not idly speculating about the philosophy of language, his practical task was to try to elucidate the meaning of unfamiliar words in the archaic and partially obscure language of the Ṛgveda. Something like ten centuries separate the composition of the verses of the Ṛgveda and Yāska. Spoken language in daily use changes a lot in that time, and in this case enough so that intelligibility began to break down. And full intelligibility has never been restored since even modern scholars are still puzzled by some of the unique words in the Ṛigveda.

Yāska's approach to dealing with these intractable problems was a form of phonetic analysis that attributed meaning to sounds.

The Sound Alike Principle, or Sound Symbolism

There is a curious fact about language that is not accepted by mainstream linguists but which was the subject of a PhD (2001) and a popular book (1999) by Margaret Magnus. If you take all the simple* words that begin with the same sound and cluster them by (broad) meaning, one will find different patterns for different sounds.

* here I use "simple" for Magnus's term monomorphemic: words that do not rely on prefixes or suffixes for their meaning. So for example "gnosis" is monomorphemic but "cognition" is not, since it involves the Latin suffix com-.

Describing the range of semantic fields covered by English words beginning with the /b/ sound, Magnus (1999: 52) says in her book:

The essence of /b/ concerns two things of which one is the lesser. /b/ identifies with this lesser element, the baddie, the back side. /b/ is not found in the 'ifs' or 'ands'. /b/ is found in the 'buts', those things which are secondary and adjoined, 'beside the point', 'by the way'.
/b/ is also high pressure within and low pressure without, like a bubble on the surface of the sea waiting to burst. English perceives this in various ways: as a birth, a death, a barrier, a transgression. Whatever the interpretation, /b/ conveys an explosive, large, and uproarious experience to the world. It blocks up openings or processes until the pressure is to great, and then it blows up. BAM! It disperses in all directions and cares not a whit where it lands.

Another, independent, manifestation of this phenomenon can be found in the little book Euphonics by John Michell (1998: 2)

An image evoked by the B sound is of balloons blown up near to bursting. They are broad, bluff, burly, obese, bulging, bulbous, burgeoning, billowing, blooming, blubbery blimps. These bouncing orbs attract adjectives of bounty: blessed, benevolent, benign, abundant, bland, buttery, and beautiful

So in the last resort, when all other methods have failed, one may try to guess using the soundalike principle. An unusual English word beginning with /b/ may well fall within the semantic fields outlined in this way. It's not guaranteed or reliable. This is a last resort and must still be combined with understanding the context.

As with etymology, the Nirukta soundalike principle escaped the grammarian's milieu and found a popular expression. My favourite example of this in Buddhaghosa's explanation of bhagavant:

bhagī bhajī bhāgī vibhattavā iti
Akāsi bhaggan ti garu bhāgyavā
Bahūhi ñayehi subhāvitattano
Bhagavantago so bhagavā ti vuccati
(Vism VII.56)
The weighty one (garu) has blessings (bhagī), is a frequenter (bhajī), a partaker (bhāgī) a possessor of what has been analysed (vibhattavā). He has caused abolishing (bhagga), he is fortunate (bhāgyavā). He has fully developed himself (subhāvitattano) in many ways. He has gone to the end of becoming (Bhagavantago) thus he is called “Blessed” (bhagavā)

At face value, the Pāḷi is pure nonsense, but note how this fits with the approach by Magnus and Michell. This doesn't always work. In the absence of an obvious verbal root for rūpa for example, some Buddhists related it to ruppati "to harm" in Pāli (e.g. the Khajjanīya Sutta which has been seen as an important source for understanding the skandhas). In answer to Yāska's question standard question: rūpaṃ kasmāt "Why is [it called] rūpa?", the Khajjanīya-kāra says that because rūpa does the action of ruppati "harming". The inferred relation is from √rup "harm" to the noun rūpa. But this is not accurate because there is no grammatical relation between rūpa and √rup. A version of this pericope in Aṣṭasāhasrikā uses the denominative verb rūpayati instead and this makes much more sense, especially in the light of my aphorism: "rūpa is to the eye as sound is to the ear." While a kind of sense can be made of rūpaṃ ruppati (since Buddhists consider sensory experience to be harmful), the sense of rūpaṃ rūpayati "an appearance appears" is too obvious to be wrong. This kind of statement is extremely common in Pāḷi.

Finally we should note that Plato also accepted a soundalike principle and essays it unsystematically in the Cratylus Dialogue.

"Now the letter rho, as I was saying, appeared to the imposer of names an excellent instrument for the expression of motion; and he frequently uses the letter for this purpose: for example, in the actual words rein and roe he represents motion by rho; also in the words tromos (trembling), trachus (rugged); and again, in words such as krouein (strike), thrauein (crush), ereikein (bruise), thruptein (break), kermatixein (crumble), rumbein (whirl): of all these sorts of movements he generally finds an expression in the letter R, because, as I imagine, he had observed that the tongue was most agitated and least at rest in the pronunciation of this letter, which he therefore used in order to express motion".

An article by Johannes Bronkhorst (2001) outlines and compares the two approaches found in the Nirukta and the Cratylus.

I am not saying that this sound alike principle was or is a first line of resort in explaining unfamiliar words, I'm saying that in the face of the failure of more rational methods, we may resort to this rather dubious method. This is definitely the last resort and all it does is potentially enrich our guesses, but there is some limited evidence that it can enrich guesses.

Unknown Words in the Present

When we encounter an unfamiliar word in a text, human beings have a limited set of possible actions to try to figure out what it means. Like Yāska, I see three levels these actions as occupying three levels. The first reaction for most of us is to ask the nearest (familiar) person if they know and then broaden the circle out. People seek information from other people before we seek it in other ways. There is also trial and error: a kind of Bayesian approach in which we infer the meaning from the context and then look to see how this works in other contexts and then adjust as required. Finally, we can guess based on clues like similarity to other words and hope for the best.

These intuitive responses have been refined by European* philologers and what follows is an attempt to generalise about how we approach the unfamiliar word in the present.

* The people we are talking about are either grew up in Europe or in one of the colonies of Europe.

As scholars encountering an unfamiliar word in a text we also have three levels of response though they have been refined somewhat: consult an expert, explore the context, and etymology.

Consult an Expert

Encountering an unfamiliar or unknown word is relatively common. I still come across new English words from time to time. Of course we live in an age in which there are comprehensive modern dictionaries and most of them are now online. It is usually a matter of a few seconds of searching to get a definition. These days as major references works are almost all online now, we can get pretty reliable answers to most questions if we are selective about which sources we use.

Recently, I've taken to watching American Football for entertainment and this has involved learning the jargon that the commentators and players use. Some terms that are used frequently like "fourth down and ten" or "out of bounds" or "snap" can easily be decoded from watching the action and equating words with what I'm seeing. However, it's not entirely obvious what terms like "go route" or "ineligible man downfield" mean in this way. In order to understand these terms I had to look them up.

When I google "ineligible man downfield", I'm asking an expert, Google's role is to put me in touch with the most relevant experts. When I look something up in a dictionary or grammar text I am consulting an expert. Experts are not simply knowledgeable about what words mean but (in the words of the late Roy Norman), they can say why words mean what they do, largely from looking at how the have been used in the past. The lexicographer Margaret Cone used to literally look at every single occurrence of a word in Pāḷi before writing her Dictionary of Pali entries. It is this that makes her an expert and her definitions authoritative.

One of the results of printing was to democratise knowledge, though it was some time before literacy became a general condition. Once a person could write down their knowledge, reproduce it relatively cheaply, and distribute it far and wide, knowledge began to escape from the milieu of experts. It also gave birth to meta-experts like librarians: we don't know everything; our expertise is in knowing how and where to find out anything.

This means that even though we seldom have direct access to experts, we have access to their knowledge in a systematic, albeit generalised, form in their written works.

Humans are social animals. We first look to each other for support; this includes knowledge support. But what to do when the experts fail, as they do from time to time.


In one of my first publications (Attwood 2010) I noticed that a translation from Pāḷi included the word "confess" but I could not find a word in the Pāḷi that meant "confess". The passage in question was cited from T. Rhys Davids' translation of the Dīghanikāya in which Ajātasattu goes to visit the Buddha and tells him about killing his father, Bimbasāra, in order to take his throne.

Verily O King it was sin that overcame you while acting thus. But in as much as you look upon it as sin, and confess it according to what is right we accept your confession as to that. For that, O King, is the custom of the Noble Ones, that whosoever looks upon his faults as a fault and rightly confesses it, shall attain to self-restraint in the future. (94-95)

I had many questions after reading this. I also discovered that in some cases of this passage in other contexts the phrase "we forgive you" occured in the English translation. Let's look at the Pāḷi and my (recent) translation.

Taggha tvaṃ, mahārāja, accayo accagamā yathābālaṃ yathāmūḷhaṃ yathā akusalaṃ, yaṃ tvaṃ pitaraṃ dhammikaṃ dhammarājānaṃ jīvitā voropesi. Yato ca kho tvaṃ, mahārāja, accayaṃ accayato disvā yathādhammaṃ paṭikarosi, taṃ te mayaṃ paṭiggaṇhāma. Vuddhihesā, mahārāja, ariyassa vinaye, yo accayaṃ accayato disvā yathādhammaṃ paṭikaroti, āyatiṃ saṃvaraṃ āpajjatī ti.
Certainly, Mahārāja, you transgressed (accaya) by going too far (accagamā) when—like a fool, like an idiot, like an incompetent—you deprived your legitimate father, the rightful king, of his life. Since, however, Mahārāja having seen the transgression as transgression, you have returned to righteousness, we accept that from you. It is an ancient custom in this noble discipline that one who sees his transgression as transgression returns to righteousness, and in future will exhibit restraint.

While we can see how a nineteenth century translator might have rendered accaya as "sin" there is no word that means "confess" nor less "forgive". And where we might expect to find that word, we have instead the verb paṭikaroti.

The entire entry from PED reads:

Paṭikaroti [paṭi+karoti) 1. to redress, repair, make amends for a sin, expiate (āpattiŋ) Vin i.98, 164; ii.259; iv.19; S ii.128=205; A v.324; DhA i.54. — 2. to act against, provide for, beware, be cautious J iv.166. — 3. to imitate J ii.406. — ger. paṭikacca (q. v.). — pp. paṭikata (q. v.).

And yet this is not how the verb is being used in the passage in question. Notably here etymology is completely useless because the verb karoti is so vague. It can mean "make" or "do" and the paṭi- prefix adds the sense of "against, back to, return". So paṭikaroti might have originally meant something like "counteract", but has clearly been used in a variety of senses.

To get the correct sense of the term for the context we have to look at the phrase and how it is used, and this is what I did in Attwood (2010). My conclusion was that the word had to be seen as part of the expression: yathādhammaṃ paṭikaroti. I argued that this must mean something like "return to righteousness" in order to make sense in context.

The passage does not use a word that means "confess", however Ajatasattu does confess in the sense that he tells the Buddha "I killed my father". And the Buddha accepts the confession (taṃ... paṭiggaṇhāma "we accept that [confession]" not "we forgive your [sin]"). And he acknowledges the resolution to return to lawfulness, but he does not "forgive" Ajātasattu and the sutta brings this out because after king leaves, the Buddha says to his monks: "The king is wounded, monks, the king is done for" (khatāyaṃ, bhikkhave, rājā. Upahatāyaṃ, bhikkhave, rājā. DN I 85) . The commentary records the story that, on death, Ajātasattu went straight to a hell realm for his transgression to be purged through extreme suffering (a hangover of the non-Buddhist idea that suffering purifies the soul). Later versions of this story see the king being purified of his unforgivable karma merely by confessing to the Buddha. This reaction is part of an ongoing process of deification of the Buddha: Buddhists could not conceive of his presence having no effect on Ajātasttu so they changed the rules of karma to allow the Buddha to save sinners.

In this case consulting the experts failed and I had to become the expert. I did this by looking at the local and broader context of how the word was used; by reading every single occurance of this term in the Pāli suttas and thinking about what it might mean in each context. Still there are times when this method fails also.

And this brings us to the last resort: etymology.


The word etymology is made up of Greek etymos "true, real, actual" combined with logos "study" and thus originally meant "the study of true [meaning]." This is a wonderfully optimistic idea. With the reconstruction of Proto-Indo-European more or less complete (notwithstanding arguments over details) and widely accepted to be accurate, we can usually trace a word in English or Sanskrit back a few thousands years to PIE; where the word is not a loan from some other language family.

Etymology is a useful tool in comparative historical linguistics because it helps identify cognate words, but as much as anything we are concerned at one level with the changes in phonetics over time on one hand and with changes in grammar (the structure of language) on the other. I think lay people often place too much emphasis on the conservation of words over time. For example, kin terms like father (pitṛ, πατέρας) or numbers like two (dvi, δύο) are strongly conserved. But what's interesting historically is that despite conservation we see sound changes. Indo-European initial /p/ becomes /f/ in the Germanic languages (including English). This is known in Europe as Grimm's Law.

The limitations of etymology for defining words are beautifully brought out by the example of nice. Our English word derives from the Latin nescius "ignorant, unaware,". It was used in the thirteenth century to mean "foolish, ignorant, frivolous, senseless." By the late fourteenth century, it was being used in the sense of "fussy, fastidious" and "dainty, delicate". By the sixteenth century, the meaning had changed to "precise, careful" (a "nice distinction"). In the eighteenth century, it came to mean "agreeable, delightful"; in the nineteenth century "kind, thoughtful"; and, finally, in the twenty-first century it now means "twee, bland, uninteresting".

At more or less any time in history, the etymology of the word nice tells us almost nothing about how it is used by English speakers. And keep in mind that we are basing this on the occurrence of the word in written texts. In fact we have little idea of how people spoke at any given time prior to audio recordings.

Examples of poor use of etymology abound in Buddhist studies. For example, Edward Conze insisted that avalokita means "looked down" and vyavalokayati "to look down". And he was flatly wrong in both cases. Both words derive from √lok "to look" and mean "examine, observe". That is to say they both suggest forms of looking closely at something: more like "getting down and looking" rather than "looking down". The prefix ava- can simply mean "downwards" but Sanskrit prefixes have a range of senses and influence on the final meaning. Prefixes don't always result in logical changes to words. For example, avagacchati can literally mean "go down" as in "descend", but it's also used to mean "arrive, visit, approach"; moreover it has a cognitive sense of "understanding, knowing".

Anastomus oscitans

Another example from my own research (Attwood 2013) is the case of sithilahanu. This word is used only once in Pāḷi and Chinese translations of the text offer no insights. The context tells us it is a kind of bird, since the word occurs in a list of birds whose feathers are used to fletch arrows. But the etymology of the name is unhelpful: sithila "loose, lax, bending, yielding" and hanu "chin, jaw". One author guessed that it might be the open-billed stork (Anastomus oscitans), which has a gap when it's beak is closed. This doesn't really make sense: a "gap" is not "loose" and the bill is not particularly bent or yielding, it has a gap which is helpful for cracking the shells of water snails without crushing them. Moreover the beak of a bird is usually tuṇḍa.

Or consider vedanā (Attwood 2018) which we often see translated as "feeling" or "sensation", but which Buddhists use to mean "the positive, negative, and neutral hedonic quality of experience" (sukha-dukkha-asukhamadukkha). We actually have an English word that is used in exactly this sense by neuroscientists, i.e. valence (a word that is used in a variety of scientific jargons). But vedanā comes from the root √vid "to know": the closely related term vedana means "announcing, proclaiming" which suggests that it is from the causative stem of the root, i.e. ved- "to cause knowing, to make known, to inform". Again, the etymology of the word has no obvious connection to how it is used in Buddhist texts.

It turns out that when we encounter an unfamiliar word, the etymology is the least useful approach to understanding what it means. Indeed, etymology is often wrong about usage at any given time or simply irrelevant to that usage.


Citing an etymology in lectures and essays probably began as a way for philologists to show that they were bona fide members of the profession who knew their stuff: a kind of academic shibboleth. I doubt they ever imagined that the practice of talking about etymology would ever escape their ivory tower. It seems that, other scholars soon realised that they could simply consult an etymology dictionary and they would sound erudite in the matter of language. I've definitely done this.

Unlike Plato who seemed to be quite playful with the soundalike principle, Yāska was serious and his business, interpreting the Ṛgveda was serious. Even so, Yāska knows that a guess based on the sound-alike principle was a last resort when dealing with an unfamiliar word. Of course, phonosemantics has never taken off in mainstream linguistics because it's too opposed to the paradigms of academic linguistics which take as axiomatic the idea that verbal sounds are not symbolic (or gestural) in any way. The modern view, following Saussure, is that spoken sounds are entirely arbitrary. However, I still think Magnus had a point and good data. And I wonder if the discovery of the sophisticated gestural language of chimpanzees, which is largely intelligible to humans, might make someone think (with Magnus) that verbalisations are (or at least involve) gestures.

For scholars of language or texts encountering an unfamiliar word, the last resort is etymology. One uses etymology to guess the contemporary (synchronic) meaning of unfamiliar words based on the historical (diachronic) meaning. These days we have the marvel of the reconstructed language of all our Indo-European forebears, i.e. Proto-Indo-European. It is now a trivial matter to look up what Indo-European root a given word can be traced to. I've spent many happy hours doing just this (for around 500 terms in my book Nāmapada).

The first resort of explaining a word to someone ought to be offering them a sentence in which it is used in a typical manner. As is done in American spelling bees. If the word has multiple senses then offer a representative sample of sentences in which the word is used. This is what the best dictionaries do, with Franklin Edgerton's Dictionary of Buddhist Hybrid Sanskrit being a shining example.

An excellent example of how to approach the meaning of a word can be found in the article on saññā by Krishna Del Toso (2015). Rather than pfaff about with etymologies, Del Toso goes straight to looking at how the word is used in sentences in Pāḷi. He begins with a catalogue of existing translations and why they are unsatisfactory, offers provisional translations from a locus classicus, but then examines how it fits in other contexts leading to refinements in his definitions. In this view saññā, can be usefully translated as "recognition", but involves but "recognizing" and "naming" (693). Del Toso suggests that saññā, "Indicates an ordering activity that is carried out by grasping the distinctive marks of things of which one has a sensation. (709)". This is similar to the conclusion I came to when studying the khandhas via summaries by Vetter and Hamilton (see Modern Interpretations of the Khandhas: Saññā).

It's all too easy to fall for the hype of etymology, but it really isn't much help day to day unless the problems we are thinking about involve historical phonetic changes. The one thing the etymology can never tell us is the true meaning of a word, independent of use. Because in the final analysis, Wittgenstein was onto something: meaning is to be found in how words are used; and use is specific to a language using community which exists in a place and a time.



Bronkhorst, J. (1981). "Nirukta and Aṣṭādhyāyī: Their Shared Presuppositions." Indo-Iranian Journal, 23(1) :1-14

——. (2001). "Etymology and Magic: Yāska's Nirukta, Plato's Cratylus, and the Riddle of Semantic Etymologies." Numen 48(2), 147-203. doi: https://doi.org/10.1163/156852701750152645

Del Toso, Krishna. 2015. “The function of saññā in the perceptual process according to the Suttapiṭaka: an appraisal.” Philosophy East and West 65(3), 690-716.

Kahrs, Eivind. (1998). Indian semantic analysis : the nirvacana tradition. Cambridge: Cambridge University Press.

Magnus, M. (1999) Gods of the Word : archetypes in the consonants. Kirksville, Missouri : Thomas Jefferson University Press.

——. 2001. What's in a Word? Studies in Phonosemantics. PhD Dissertation, Norwegian University of Science and Technology, Trondheim.

Michell, John (1998). Euphonics: A Poet's Dictionary of Sounds. Powys, Wales: Wooden Books.

Skilling, Peter. (2007). Mṛgara’s Mother’s Mansion: Emptiness and the Śūnyatā Sūtras. Journal of Indian and Tibetan Studies 11, 225-247.

Visigalli, Paolo. (2017). "Words In And Out Of History: Indian Semantic Derivation (Nirvacana) And Modern Etymology In Dialogue". Philosophy East and West, 67(4): 1143-1190.

Witzel, Michael. (1999). "Substrate Languages in Old Indo-Aryan (Ṛgvedic, Middle and Late Vedic)." Electronic Journal of Vedic Studies 5(1): 1-67.

17 February 2023

What's the Difference Between a Meditator and Corpse?

At first glance, my title this week might seem like an odd question or the opening to a joke. In fact, the question is asked and answered in the Pāḷi Mahāvedalla Sutta (MN 43). This is one of those suttas that seems to be an attempt to comprehensively summarise Buddhism as it was understood at the time, but not in a standard Theravāda way. 

The Mahāvedella is a teaching by Elder Sāriputta for Elder Mahā-Koṭṭhita. The pair are also portrayed as speaking together in the Koṭṭhita Sutta (AN 9.13) and another Koṭṭhita Sutta (SN 35.232).

In this case, the sutta includes some ideas that are rare elsewhere. What the Pāḷi texts repeatedly show is that different ancient Buddhists thought about the same terms in different ways. Not everything that we find in a Pāḷi sutta was incorporated into Theravāda Buddhism, even in theory. 

The Mahāvedalla Sutta

The Mahāvedalla Sutta is a series of questions and answers. For example, the first question asks for a definition of "faulty pañño" (duppañño; Skt duḥprajñā) and compares this with someone endowed with pañño (paññavā; Skt. prajñāvat). Note how these are not quite opposites. The natural opposite of duppañño would be supañño; while the opposite of paññavā would be apaññavā. No doubt there was a story here, but it's lost to time. It's not clear how the Mahāvedalla-kāra understood pañño, the adjectival form of paññā, but in Prajñāpāramitā it seems to connote the knowledge gained by undergoing cessation (nirodha). The series of questions continues. Define "discrimination" (viññāṇaṃ; vijñāna)? What is the difference between viññāṇaṃ and paññā? The answer here is that paññā is to be cultivated; discrimination is to be comprehended (paññā bhāvetabbā, viññāṇaṃ pariññeyyaṃ). 

This explanation leaves me in the dark about the distinction, I think, because I lack the context in which to understand it. There is one other reference to cultivating paññā in Pāḷi. The Rāga Sutta (AN 6.107) describes a group of three things to be abandoned (raga, doha, moha) and three to be cultivated (asubha, mettā, and paññā) in order to eliminate them, i.e. cultivating understanding (paññā) dispels confusion (moha). This one is comprehensible on its own, but doesn't help us to distinguish paññā from viññāṇa. It seems that the Mahāvedalla-kāra did not see viññāṇa as something that could be cultivated or abandoned. But this doctrine was not developed by Buddhists and all we have is this incomplete snapshot. This happens a lot in the Pāḷi suttas. 

Then the sutta asks, what is valence (vedanā) and recognition (saññā)? And are these three—saññā, paññā, vedanā—inseparable? The sutta-kāra says they are not separable because "what one experiences, that one recognises; what one recognises one discriminates" (yaṃ hāvuso, vedeti taṃ sañjānāti, yaṃ sañjānāti taṃ vijānāti MN I 293). Note that the traditional skandha meditation practice is predicated on being able to distinguish these three, while here the three are said to be impossible to distinguish individually (na ca labbhā imesaṃ dhammānaṃ vinibbhujitvā vinibbhujitvā* nānākaraṇaṃ paññāpetuṃ).

* The repetition of vinibbhujitvā here is odd, but seems to be in the original texts. 

Then a change of pace. "Comrade, what can be inferred by purified mental discrimination that dismisses the five [physical] senses?" (Nissaṭṭhena hāvuso, pañcahi indriyehi parisuddhena manoviññāṇena kiṃ neyyan ti?)

* Ñāṇamoḷi & Bodhi "Friend, what can be known by the purified mind-consciousness released from the five faculties?

Interestingly, what can be inferred or understood (neyyan) from this are precisely the āyatana states. From the statement (or thought) "space has no limits" we can infer the stage of limitless space (ananto ākāso’ti ākāsānañcāyatanaṃ neyyaṃ); from "there is no limit to discrimination" we infer the stage of limitless discrimination can be inferred (anantaṃ viññāṇan ti viññāṇañcāyatanaṃ neyyaṃ); and from "there is nothing" we infer the stage of nothingness can be inferred (natthi kiñcī ti ākiñcaññāyatanaṃ neyyaṃ). And we know this phenomenon through the eye of paññā (paññācakkhunā). And what is the purpose of paññā? It is higher knowledge (abhiññatthā), exact knowledge (pariññatthā), and abandonment (pahānatthā). The latter refers to eliminating sensory experience (cf. Pahāna Sutta SN 35.24).

More questions follow on right view (sammādiṭṭhi), being (bhava), first jhāna, the five faculties, and then the section that really interests me.

Life and Heat

The pertinent question is, "On what condition do the five faculties depend?" (pañcindriyāni kiṃ paṭicca tiṭṭhantī ti); where the five faculties are eye, ear, nose, tongue, body. The Mahāvedalla Sutta says that they depend on āyu "life" (Skt āyuḥ; as in āyurveda). Life itself depends on the condition of "heat" (āyu usmaṃ paṭicca tiṭṭhati) but, at the same time, heat depends on the condition of life (usmā āyuṃ paṭicca tiṭṭhati). The relation between the two is explained by an analogy: it's just like how seeing the light of a lamp is dependent on seeing the flame, and seeing the flame is dependent on seeing the light. This mirrors the analogy between mutually conditioning viññāṇa and nāmarūpa in the Mahānidāna Sutta (DN 15) there conceptualised as two sheaves of harvested grain that lean against each other (called a "stook" in English).

Life and heat are not a common topic in Pāḷi; they occur together in just three texts including the Mahāvedalla Sutta, and I will digress briefly to consider the other two. We find life and heat together in a verse at the end of the Pheṇa­piṇḍ­ūpama Sutta (SN 22.95) where death is equated with the absence of āyu, usmā, and viññāṇa (SN III 143). In the Kāmbhū Sutta (SN 41.6), which features a discussion between the patriarch* Citta and the bhikkhu Kāmbhū, we find a similar discussion of the difference between a corpse and a meditator experiencing cessation (Starting at SN IV 294). Here the bodily, verbal, and mental formations (kāya-, vācī-, and citta-saṅkhāra) cease in a meditator undergoing cessation. However, they still have life and heat, and their "faculties are serene" (indriyāni vippasannāni).

*Gahapati refers to the patriarch of an extended household or possibly an extended family within a clan structure. Standard translations like "householder" seem to miss the point.

Note the inconsistency here: a living person in both texts has life and heat, but the third factor is viññāṇa in one account and indriyāni in another. Here we might conjecture that viññāṇa is intended as the function of the indriyāni, i.e. objectification is the function of the sense faculties. We could, at a pinch, see the two terms in this context as synonyms. Though this is a neat solution, we have to consider other possibilities as well. The two texts may be trying to say something different and incompatible that we no longer understand (this is not uncommon between two Pāli texts).

I don't understand how we came to translate viññāṇa as "consciousness" but it seems plain wrong to me. Notably, viññāṇa is an action noun rather than an abstract noun, so viññāṇa and consciousness are not even on the same level of abstraction. It is my view that no Pāḷi word can be translated into English as an abstract noun "consciousness" and that our whole philosophical concept of "consciousness" is absent from ancient Buddhist dialogues (see also The 'Mind as Container' Metaphor). The use of "consciousness" in discussing ancient Buddhist discourses is a Whiggish anachronism (in which we imagine ancient Indians to be primitive precursors of ourselves).

In any case, the gist here is clear. It can be very difficult to distinguish a meditator from a corpse by the usual signs of life that we look for in a conscious and aware person, because we cannot interact with them. We could say that following cessation a person becomes completely unresponsive to the world around them. People undergoing cessation of sensory experience necessarily lack all sense of time, since all of the clues to the passing of time have, by definition, ceased. Hence, perhaps, the Buddhist insistence that the Buddhadharma is akāliko "timeless", though in a culture where death is often referred to as kālaṅkato "having done one's time", akāliko could also be a synonym for amata "deathless" (Skt. amṛta). The phenomenon of people sitting lost in samādhi for days on end is likely related to their undergoing cessation and having no sense of time passing. It is likely that thirst, i.e. a need for water, is what rouses them. Being dragged out of samādhi by thirst may explain why "thirst" (Skt. tṛṣṇa; P. taṇha) became such a key word in the Buddhist lexicon.

Life Force

Coming back to the Mahāvedalla Sutta and moving to the next section the subject is now "life" (āyu) and the "constituents of life" (āyu-saṇkhārā). The sutta explicitly states that these "constituents of life" are not phenomena that one can experience (na kho, āvuso, teva āyusaṅkhārā te vedaniyā dhammā). And then it says that, if the āyu-saṅkhārā were phenomena to be experienced, the one who experienced the cessation of awareness and experience would not emerge from their meditation, that is to say they would die. The logic here is that if āyu and āyu-saṅkhāra were part of the experienced world, then when the experienced world ceased, so too would life. Rather, the text makes the apposite observation that life continues even when all sensory experience ceases. 

What did the sutta-kāra mean by āyu and āyu-saṇkhārā? It is difficult to say, because the terms are not defined. Sujato has blogged about how the words āyusaṇkhāra and jīvitasaṇkhāra are used. There is not a great deal more to be said. In the Mahāparinibbāna Sutta (DN 16) the Buddha mentions jīvitasaṅkhāra in a sense that Sujato interprets as a "will to live". He is, I think, here relying on the traditional idea that saṅkhāra means "volition" because it is explained as the six kinds of cetanā associated with the six sense spheres.

This meaning of saṅkhāra derives from the earlier Brahmanical use of the Sanskrit equivalent. In Vedic ritual, a saṃskāra is a rite of passage. When performing these rites, the Brahmin priests carry out a series of actions (karman). Hence, in Buddhist usage, saṇkhāra/saṃskāra is "an opportunity for doing karma". Keeping in mind that all intentional acts carry a karmic debt. At the same time, the unique but influential passage in AN 6.63 famously says "intention is how I talk about karma, monks" (cetanāhaṃ bhikkhave kammaṃ vadāmi). Thus an opportunity for doing karma becomes an intention to act. 

Whether this meaning can be applied to āyusaṅkhāra is moot and, since Sujato doesn't make this case, we are none the wiser. He finds a way to make sense of jīvitasaṇkhāra as "the will to life" and then retrospectively relates āyusaṇkhāra to this as a kind of "vital force". In the end, however, Sujato concludes that distinction between āyusaṇkhāra and jīvitasaṇkhāra probably emerged later and that the two words are synonyms for "vitality" and "vital energies" and are best translated as "life force". This is a self-consistent explanation and it might be right. But there is presently no way to confirm such conjectures: we are trying to make sense of how a word was used in the absence of any contemporary explanation and from just a few instances that are vague and/or ambiguous. This is a common problem when dealing with older Buddhist texts (in any language). 

Across the ancient world we repeatedly encounter the idea of a "life force", but it is almost always conceptualised as breath. Words indicating breath as life force include: psyche, anima, spirit, qi 氣, and prāṇa. For more on this theme see my 2014 essay: Spiritual I: The Life's Breath. In the Indian context the vital force is āṇa "breath" which itself is caused by the action of the element of wind (vāyu). Vāyu conceptualises all forms of movement. The word āyu, however, does not refer to "breath". Rather, it is related to the words aeon and age, and often refers to lifespan or longevity. Breath (āṇa) is what animates the body (kāya); the resulting animation seems to be called āyu (and is accompanied by usmā). Similarly, jīva is not related to breath but is cognate with Greek bio, Latin vivarus, and Germanic quick; all meaning "life; living".

These are not ideas that were integrated into later Buddhism. Nor does the concept of a life-force as distinct from mind and body ever become mainstream. The reason is obvious, and has also bothered European philosophers. If there were a "life-force", then it would surely have a roll to play in facilitating life after death. And if it is present in all living things, as appears to be implied, then we are in the realms of eternalism: that is to say āyu starts to sound suspiciously like ātman. Not surprisingly most Buddhist schools of thought set the idea of a "life force" outside of their orthodoxy and āyusaṅkhāra never became a mainstream Buddhists' technical term. Moreover, Buddhist knowledge of physiology never really developed beyond this Iron Age conception.


To answer the question in the title, a meditator and a corpse are similar in that signs of life in the form of actions of body, speech, and mind are absent. Even though the meditator is insensate, or even catatonic, they are still alive; still warm. The corpse is cold and lifeless (and decay sets in almost immediately). 

Presumably, this was enough of an issue for the early Buddhists thought that it required some doctrinal explanation. That said, the terms used to explain the difference—like āyu and āyusaṅkhāra—did not seem to need an explanation in the minds of the author(s). Leaving us scratching our heads. 

This sutta is not consistent with Theravāda Buddhism, if only because it unequivocally states that vedanā, saññā, and viññāṇa cannot be distinguished from each other. Nor is this statement consistent with any form of Buddhism I am familiar with. The Mahāvedella Sutta appears to be from an unknown sect of Buddhists, missing from the historical record. Their text was preserved, but the teaching lineage associated with it was not. I suspect this is true in a large number of Pāḷi suttas.

However, that āyu and usmā occur together in three texts suggests that at least some Buddhists believed in some kind of "life force" as distinct from a soul (ātman). A life force (jīva) was also important in Jain theology, where it provided the necessary continuity for rebirth. At least some Buddhists further conceptualised life as composite and posited life-constituents (āyu-saṅkhāra). However, in the end we don't know precisely what words like āyu or āyusaṅkhāra meant to those people then, because they didn't say and there is not enough context to guess.

In this case it is very tempting to smooth over the difficulty by conjecturing an answer that solves all the problems, is plausible, and self-consistent. However, this is not sufficient to establish how the author(s) thought. Any number of plausible, self-consistent answers are possible. But we have no objective facts available to help us choose between them.


03 February 2023

Does Buddhism Provide Good Explanations?

This is my 600th essay on this blog. Thanks to all my readers over the last 17 years. Although I've slowed down in order to focus on publishing in academic journals, I still enjoy writing these less formal essays.


What is an explanation? And more to the point, what is a good explanation? To answer these questions we have to formulate a "good" explanation of "explanation" along with an  explanation of what constitutes "good" in this context. There is a whole branch of the philosophy of science concerned with characterising scientific explanation. We can draw on this to outline such an explanation. In a classic article, Jan Faye (2007) outlines three modes of scientific explanation. These will form the basis for an exploration of what kinds of explanations that Buddhism offers.

Formal-Logical Mode of Explanation

In this approach, A and B are both (logical) propositions. We may say that A explains B if B can be inferred from A using deduction. This approach involves some idealisation because both A and B are propositions rather than brute facts. Moreover, if A explains B according to this definition, then A is ipso facto a good explanation. Here a "good explanation" is one that follows the rules of logic. Formal-logical explanations are thus prescriptive rather than descriptive.

The problem with this mode of explanation is that in applying deduction we inevitably seem to reproduce our starting assumptions. All exercises in logic involve axioms or propositions that we hold to be true but cannot substantiate. For example, many Buddhists assume that the Buddha was an historical character. We can't prove this, since there is no direct evidence of the Buddha, but the proposition has very broad appeal and most people uncritically accept it as true. The problem, then, is how this unevidenced commitment affects the outcome of deductions from circumstantial evidence. 

If we were simply to begin with known facts then we might deduce new facts. But in all cases of applying logic to the real world we cannot avoid axioms, or at least propositions that we take to be axiomatically valid. If we set out to prove that the Buddha was an historical character when we have a prior intellectual commitment to that proposition, i.e. if we treat the Buddha's historicity as axiomatic, then the details of the logical argument are irrelevant. There will always come a point when we deduce that the Buddha was an historical character and we will judge this to be a valid deduction precisely because it concurs with our axiomatic presupposition. In other words, if our axiom is "the Buddha was an historical character" then any sequence of deductions that reproduces the axiom will automatically be accepted as valid. This boils down to Aristotle's law of identity: A is A. 

Recently, David Drewes (2022) challenged the axiom of the historical Buddha by arguing that the basic definition of an historical character is that they can be connected to an historical event or fact. If this is our criterion then the Buddha is not an historical person since he cannot be connected to any historical event or fact. Drewes concludes that it does not make sense (at least for academic historians) to speak of an "historical Buddha". Indeed, the same can be said of all the characters in early Buddhist texts, including the kings. Usually, if anyone can be associated with history, it is kings. Lāja Piyadasi (aka  Emperor Asoka whose name means "remorseless") is the first historical character in Indian history. He left artefacts that we can definitely link to him. 

This mode of explanation has long been out of fashion in science, but it still holds sway in theology, including Buddhist theology.

Ontological Mode of Explanation

Ontological explanations seek to identify and understand causes. The idea here is that facts, events, and states explain observations by revealing law-like casual relations between them. This mode of explanation is familiar to anyone who has studied science. In this mode, A explains B if A is the cause of B: where A and B are facts, events, or states. 

In a classical scientific explanation we might say, for example, that a force applied to a mass causes it to accelerate (i.e. undergo a change in speed and/or direction). The size of this effect is given by Newton's second law of motion: F = ma; where F is force, m is mass (kg) and a is acceleration (m/s2). From this we can say that the magnitude of the change in acceleration is equal to the force applied divided by the mass. It also allows us to define a unit of force: one Newton of force (N) is 1 kg⋅m/s(i.e. mass multiplied by acceleration). However, note that there is no term in the equation that corresponds to "causation". We assume that the applied force is the cause of the acceleration, but we don't need to encode that in the descriptive law. 

As Faye says, in this mode, "An explanation is both true and relevant if, and only if, it discloses the causal structure behind the given phenomena." (I'm referring to an unpaginated preprint). As another commentator, Sam Wilkinson puts it, "events explain other events, and they do so by standing in predictable law-like relations" (2014: 2)

Causation is a very tricky subject since, as David Hume observed some 300 years ago, we never observe causation, we only ever observe sequences of events. Immanuel Kant went further with this and asserted that causation and other metaphysical notions, like space and time, are actually structures imposed on experience by our minds in order to make sense of them. This basic insight seems to have stood the test of time. The explanations of causation I have seen tend to be rather abstract. In my view, our understanding of causation is related to learning how to use our own limbs and especially our hands. The archetype of causation, then, is action initiated by desire. This is why the concept of causation seems so intuitive but also why it is so confusing. Desire is an emotion and is not present outside of animal life. It's not present at all in most sequences of events in the universe, and as far as we know the only agents in the entire universe all live on earth. Thus our internal model for causation is not typical of causation generally. The conscious initiation of an action is an anomaly, not a model. 

Moreover, most events have multiple causes and we tend to highlight one or a small number from amongst them. The real world is far too complex for single causes to give rise to single effects, so most causal explanations introduce simplifications: for example, we often ignore gravity when it is perpendicular to the plane of interest. Our perception of initiating movements is biased because we are simply not aware of the complexity involved in making our limbs move: our kinesthetic sense, for example, is very coarse grained compared to our physiology. We don't sense, in any way, the nerve impulses involved, or the muscle fibres contracting and relaxing. We just have a broad sense of the limb moving. 

Thus, causal explanations are also fraught with uncertainty because the basic concept of causation is ill-defined and our understanding of it is based on a local anomaly. 

This leaves us with the third mode of scientific explanation. 

The Pragmatic Mode of Explanation

"Pragmatic" here is a technical or jargon term that refers to a particular approach to philosophy that is mainly focussed on speech and "speech acts". A speech act is an utterance that is intended to do something. This approach is contrasted with semantics, which is focussed on what words mean. It is not that pragmatists deny the meaning of words, but they do see meaning as secondary to doing. A classic example is irony, when we say one thing, but mean another, and do so "for effect". Semantics struggles to explain irony, though semantic explanations of irony have been proposed. Irony is simple for pragmatics since the effect of irony is to highlight the gap between expectation and reality. This is what the ironic utterance does

A question, in this view, is a speech act with the intention of seeking information. According to Faye, an explanation begins with a question. A pragmatic explanation, then, is a response to a question. As such, then, the purpose of an explanation is to satisfy the questioner. And an explanation is "good" to the extent that it does this. 

This may seem like a slippery slope to "relativism". Pragmatist philosophers address this by requiring that the questioner not be delusional and thus the answer only has to satisfy the rational requirements of a questioner and not their irrational requirements. Different philosophers have addressed this in different ways. For some, it requires a metaphysical turn, that is, linking "good" to being "true". 

Less problematic is the idea of accuracy: the closeness of an observed value to the true value. A good explanation has to attain a somewhat arbitrary level of accuracy. In physics, for example, scientists look for a confidence level of 5-sigma (5 σ). This means that there is a one in two million chance that your measurement is a statistical fluke that can be explained by some alternative route such as measurement error. So when, for example, someone announced that they had measured neutrinos travelling faster than the speed of light, the first reaction was to look for some problem in the measurement rather than an immediate overturning of general relativity. And sure enough, a loose wire in the measurement device was fixed and the neutrinos "slowed down" to subluminal speeds.

Note that all measurements have an inherent level of error due to the limitations of our methods. Real science will always state what level of error the scientists have identified in their measurements. If you see a measurement that is not followed by the ± sign, then you should be on alert because someone is, at best, oversimplifying things and, at worst, attempting to deceive. 

An explanation that matches observations with a 5 σ level of accuracy is taken to be a good explanation in science. General relativity is a good explanation to certain questions asked by scientists. What is gravity? "Gravity is the geometry of spacetime". Why does the orbit of Mercury precess at the rate it does? Because at perihelion the curvature of spacetime caused by the sun is much greater than at aphelion, causing the planet to get ahead of where its orbit is expected to go under classical descriptions. And so on. 

Other philosophers try to apply criteria like "usefulness" a good explanation is one that is "useful", though usefulness is itself a nebulous concept: Useful for what? Useful to whom? 

In the pragmatic view, a good explanation entails the person asking the question understanding the answer that addresses their rational concerns. And if it does this, then it is a good explanation for that situation. For example, it remains to be seen whether my explanation of this is good from the reader's point of view. It gets interesting when authors raise questions that readers have never thought to ask, a situation I meet everyday as an author. 

Incidentally, this is consistent with Mercier & Sperbers contention that reasons are post hoc explanations that we produce in response to questions about motivations. Humans don't act for reasons, we act and then produce ad hoc reasons as required. Our deliberations on whether to act or how to act are generally handled by inferential processes below the threshold of awareness. Moreover, we tend to adopt the first plausible explanation for our actions that comes to mind. A striking demonstration of this can be seen in people with neurological damage who cannot form new memories. Oliver Sacks reports one such case in his book The Man Who Mistook His Wife for a Hat (1985). The patient is in hospital but cannot remember why. Asked to account for why he is there, the man confabulates a story that seems plausible to himself. He's there for a check up, he's waiting for someone, and so on. Each time it's different because he can't remember his previous answers. One of the functions of memory is to put limits on our speculations about how we got here. 

By this criterion, if a five-year old asks me why the sky is blue, and I respond by explaining how the scattering of incident sun light by atoms in the atmosphere is dependent on the type of atom and the wavelength of the light; blue light is scattered by more nitrogen gas than other frequencies, so we see more blue light than other colours in any direction in the sky away from the sun itself. The answer is objectively true, to the best of my knowledge, and accurate to some arbitrary level. However, I have still failed to give a good answer because my five-year old interlocutor is unlikely to understand it. This is not relativism, since we can still say that some answers are better than others. The sky is not blue because that is God's favourite colour, for example. Rather there is a limit on the "goodness" of an explanation that is dependent on the knowledge and capabilities of the questioner. Similarly, if a scientist asks why the sky is blue and I give a mythological answer that would satisfy the people of, say, Iron Age India, it would not be a good explanation in that circumstance.

So a good explanation, in this mode, is an utterance that addresses a particular question, asked by a particular person whose rational needs (especially for understanding) must be satisfied by the answer. This means that explanations in this mode are not universal, they have to fit the requirements of the question/questioner. In this mode, contrary to the others, the explanation is not inherent in nature, not a fixed thing, but changes dependent on who is asking and why. An explanation has to do more than simply reveal a pre-existing truth, it has to communicate something to a specific audience. And as such, an unmitigated scientific explanation is seldom the best explanation except to other scientists. 

Other Modes?

These modes are how philosophers of science explain explanation to each other. I was alerted to this by reading Sam Wilkinson's (2014) application of Faye's schema to a problem in neuroscience. One mode that is not included, though it was the primary mode of explanation for ancient Buddhists is, explanation by analogy. In this mode, one can explain an unknown event or process by analogy to some known process. One of the best examples of this from ancient Buddhism is the niyāma doctrine that we find in Pāli commentaries attributed to Buddhaghosa. 

In this view, for example, we can explain karma by reference to the way a seed grows into a plant (bīja-niyāma). Indeed, we refer to the results of karma as phala "fruit" and vipāka "ripening, maturing" (from √pac "to cook"). This also explains the like-for-like specificity of karma: good actions (kusala, puṇya) lead to a good rebirth (sugati); bad actions (pāpa, akusala, apuṇya) lead to a bad rebirth (duggati). The analogy here is that a rice seed can only grow into a rice plant. On the other hand, karma also has to be timely, i.e. to ripen at the appropriate time. The analogy for this is the way that plants bloom and fruit in season, or the way the monsoon rains (mostly) come at the same time each year (utu-niyāma).

However, sometimes ancient Buddhists could find no analogy. And in these cases they often resorted to theological arguments. Taking another example from the doctrine of pañcavidha niyāmaKamma-niyāma itself refers to the inevitability of consequences. Karma must ripen. There is no good analogy for the inevitability of this, since it is not true that real fruit must ripen. In nature, crops may fail to ripen for many reasons. Some unripe fruit fall from the tree, some are consumed by birds, insects, or fungus, some are affected by draught, etc. Where they could find no analogy from nature, Buddhists would resort to an appeal to authority. In the case of kamma-niyāma Buddhaghosa opted to cite a verse from the Dhammapada (127) which describes inevitability. 

In my articles on karma I have noted that the early belief in the inevitability of karma is overridden in  later Mahāyāna and Vajrayāna texts that allow for religious practices to mitigate karmic consequences. The epitome of this is the Vajrasatva mantra, the chanting of which is said to eliminate even the most negative forms of karma, such as that from killing a buddha. This alone suggests that argument from authority is less successful than argument from analogy.

Doctrine as Explanation

Buddhist explanations are typically codified as "doctrine", a word derived from Latin that means "teaching", but is also used to refer to the official body of teachings that are core to a particular religious community. Some religieux, notably Christians and Muslims, define membership of their group by acceptance of certain doctrines. Modern Buddhists will often ape this practice, setting out a Buddhist catechism, but ancient Buddhists seem to have defined membership in terms of behaviour rather than belief. And after all, a person can profess belief, but if their behaviour is not consistent with that belief such profession is meaningless. 

Doctrines are not necessarily arbitrary, but can reflect a considered position on a particular question or problem. That is to say, a doctrine can (and often does) function as an explanation. For example, in response to the question, "How should we live as Buddhists?", Buddhists codified sets of training rules or śikṣapāda, the best known of which are the pratimokṣa and the five precepts for lay Buddhists.  These constitute different levels of explanation for different people. Lay people were not expected to live according to the pratimokṣa, though they often knew the rules and reported monks who broke them. 

However, Buddhists seldom codified the questions or problems that gave rise to the doctrines. And thus we have to be careful. We do not know, for example, why anyone believed in rebirth in India. Vedic speakers were originally from a milieu that did not believe in rebirth. They seem to have embraced it when they encountered it in India (traces of the process are found in the Ṛgveda stories about Yama as the discoverer of the way to the pitṛloka). We know a good deal about the content of what various communities believed about life after death, and we know how they responded to the belief, but we don't know why they found rebirth a more compelling version of the afterlife than the one they arrived in Indian with (every culture has a view about this). Nor do we fully understand how rebirth happened, since even within Buddhism there was no general agreement on this. 

There are times when we can attempt to reverse engineer the problem that gave rise to a doctrine, but since we have no way to go back in time and confirm this directly, such reconstructions are always speculative. 

According to many Buddhist texts, the most fundamental thing that Buddhism seeks to explain is suffering. In Christianity, for example, suffering must be explained with reference to the creator and is often referred to as the problem of theodicy or "God's justice"; or simply the problem of evil. If God created the world with evil in it, was he compelled, complicit, or indifferent? Christians have addressed this problem in different ways. Evil is not God's fault because he gave humans free will (evil is our fault; or worse, evil is women's fault). Evil is part of God's plan and not being omnipotent we simply cannot understand it. Many of us now say that the explanations of evil offered by Christians are not good explanations because all of the answers implicate God as the creator of evil. 

Buddhists, free of the scourge of a personified creator, had a number of approaches to explaining suffering. One of the most important reasons offered is the belief that we are repeatedly reborn in one or other of the realms of saṃsāra. To be reborn is to suffer. In response, all ancient Buddhists sought an end to rebirth as their primary goal. Arguably all modern Buddhists do also, but they seldom couch it in these terms. Modern Buddhists often try to explain evil from the point of view of individual psychology (often with a psychoanalytic perspective): evil is a result of an individual's desire, since it is desire that fuels rebirth. 

However, in this example, rebirth is a given. The belief in rebirth seems to have been ubiquitous in pre-modern India. Rebirth is not a conclusion, it is an axiom. That we are reborn is never in question for Buddhists (and when it is questioned, many Buddhists throw up their hands and declare that without rebirth Buddhism is meaningless). There are early Buddhist references to non-believers, but we don't know if they actually existed, because they did not leave their own records and where we have external confirmation (as through archaeology) we find Buddhist texts are quite unreliable witnesses to history. As we know, none of the characters in the Pāli stories can be linked to historical events. Indian history begins with writing and writing in India begins with Asoka's inscriptions. Before that, we have archaeological evidence, such as potsherds or stupas, but this tends to shed light only on material culture, not on individuals or mores.

As Sue Hamilton (2000) says repeatedly, early Buddhists were interested in how certain things worked; they did not have much to say about whether or not anything existed. This is also true for Prajñāpāramitā.

Is Dependent Origination a Causal Explanation?

Recall that the causal mode of explanation aims to explain something by showing a causal relation with something else. X explains Y, if X is the cause of Y. Superficially, the formula of dependent arising resembles a causal explanation, so much so that many modern Buddhists speak of dependent arising as a "theory of causality". This is inaccurate. Dependent arising tells us when causation happens, i.e. when the condition is present; it does not explain how causation happens. Moreover, the relations between the twelve nidānas are extremely varied: ignorance is the condition for volitions, for example, but birth is the condition for death. In the latter we get a dramatic example of why this is not a theory of causation. If X causes Y, then we are left stating the proposition that "birth causes death". This is obviously false. 

Of course one cannot die without first being born, but it would be ridiculous to say that birth is the cause of death. Such a causal explanation fails to satisfy any rational question about death. Take the case of a person who dies at age 100. Their birth was 100 years prior. Just as Buddhists insisted on the presence of the condition, we too think of causation as happening locally, both physically and temporally. A cause with a 100 year delay in fruition doesn't work as a causal explanation because we don't perceived widely separated moments in time as being present to each other (note that in English we almost always discuss time using spatial metaphors). We are constantly interacting with the world and its living creatures; constantly experiencing causation at various levels. A 100 year delay opens the door to uncountable causal relations that would smother any causal connection between two events. 

Leading up to bhava the nidānas can, at a stretch, be understood as relating to experience. But bhava means "existence". It may be that this was the end of the sequence at some point, because bhava here implies a series of rebirths driven by karma. This raises larger questions about whether the nidānas are an explanation at all. Many would say that the nidānas are not an ontology (not an explanation of what exists), but merely a framework for reflecting on the nature of experience in the pursuit of ending rebirth. The whole thing is predicated on axiomatic rebirth. 


I started with an overview of different modes of explanation as understood by philosophers of science:  the formal logical mode, the ontological mode, and the pragmatic mode. Philosophers of religion have to add a few more modes of explanation that scientists don't accept, such as explanation through analogy and appeals to doctrine. This looks like a fruitful approach to exploring Buddhism. For example, it would be useful to identify what kinds of questions Buddhists were asking. We get some idea of this from the work of people like Richard Gombrich who have identified where Buddhists were in dialogue with Brahmins or people of other faiths. For example, it seems that everyone in ancient India who believed in rebirth looked for an explanation of rebirth. Just saying we are reborn again and again is not satisfying, especially in a culture that sees birth as a misfortune. Ancient Buddhists wanted to know why we are repeatedly reborn into misery because the idea of being reborn into misery was taken to be axiomatic true. And the doctrine of karma seeks to explain rebirth rather than the belief in rebirth. Buddhists never sought to explain their belief in rebirth because everyone took it to be axiomatically true. Background beliefs are almost never explained, which is why we still need philosophers. 

What other kinds of questions were Buddhists trying to answer in their myriad doctrines? And why do those doctrines compete? We tend not to even think in these terms so the answers are not forthcoming.

Then there is the question of why doctrines change. I have been interested, for example, by changes in the doctrine of karma over time, especially where we get oddities like the sarvāsti doctrine or the pudgala doctrine. What caused such changes? Buddhists themselves are uninterested in such questions. They are more concerned with the question of authenticity. For example, there is now a regular stream of articles saying that secular mindfulness is not authentic because the suite of practices under that  rubric have been removed from their "spiritual" context (see for example White 2023). 

What does this word "spiritual" mean any more? What question is "spiritual" the answer to and for whom? Moreover, there is no word in the Buddhist technical lexicon that can legitimately be translated as "spiritual"? The word has no meaning in modern English anymore, but it never had any meaning in a Buddhist context. 

In the end, what do the Iron Age doctrines of Buddhism explain and for whom? For example, the intended audience for many texts is monastics, living a life that has almost no parallel in modern Buddhism, i.e. mendicant monasticism: living rough, depending on begging, mostly engaged in meditation. Do answers for this community constitute a good explanation of anything for someone living in the 21st century? Many modern Buddhists require us to believe that not only are the Iron Age doctrines still apt, but that they are the acme of all explanations: Buddhists purport to explain reality itself. 

But these explanations strike few of us as salient or good. And by "us" here, I mean modern Europeans and people living in European colonies. Traditional explanations are still popular where Buddhist traditions thrive. But they clearly do not relate to the majority. If we look at the statistics, Buddhists are typically less that 1% of the population of European states. 200 years after the first Buddhist monks arrived in the UK we are still less than half of 1% of the population. The fact is that, while many Europeans are sympathetic to and tolerant of Buddhism, they have no interest in becoming Buddhist. The largest and fastest growing category of religion in the 2021 UK census was "no religion". Religious explanations are not winning people over, secular explanations are. 

Moreover, I don't find the Buddhist account of reality convincing anymore, either. I'm not sure I was ever wholly convinced, but I did take on a lot of doctrine in order to facilitate a sense of belonging. Now I find the metaphysics of Buddhism quite dull and uninteresting, especially in the light of Sue Hamilton's heuristic of treating the Buddha of the Pāli canon as addressing issues of phenomenology and knowledge, and not as addressing metaphysical problems. Indeed, the Buddha famously avoids answering  questions framed as "Does X exist or not exist?" And this attitude persists into the Prajñāpāramitā literature that I have spent the last ten years studying. 

Buddhist metaphysics leads to paradoxes and contradictions; to philosophical kludges like the two truth doctrine (dvisatyavāda) or the momentary doctrine (kṣanavāda). One paradox, for example, is that dependent arising forbids action at a temporal distance while karma demands it: conditionality requires the presence of the condition, while some karmic actions only mature (vipāka) into consequences many lifetimes later. We cannot have both and ancient Buddhists produced all kinds of fudges to try to overcome this, with simple denial being at the forefront. For example, one academic journal editor, who is a Theravādin Buddhist, refused to publish an article of mine because he refused to admit that dependent arising requires the presence of the condition. We need to be clear about this: causation in the absence of a cause is totally incoherent, even in Buddhism. Causation in which the cause and effect appear at arbitrary times doesn't even meet the requirement of being a repeating sequence of events. It is not a sequence at all since any number of other events intervene and are implicated in conditionality generally. 

The internal contradictions of Buddhist doctrine are a sign that something has gone wrong. Moreover, ancient Buddhists must have been aware of this, since all Buddhist sects introduced doctrinal innovations and this still goes on today. Some of the explanations, such as sarvāstivāda and pudgalavāda, make clear that they introduced innovations to deal with  perceived problems in doctrine, but most never acknowledge the motivations for innovations. Changes in the doctrine of karma are never justified. Later Buddhists simply adopted a different definition and did not acknowledge the older one. 

The modern Buddhists' use of ancient Buddhist texts as justification for a belief is not a good explanation for me, as it doesn't answer any question that I have posed about Buddhist doctrines. Moreover, not even ancient Buddhists could cite textual justification for changing the doctrine of karma, because that would mean admitting that the Buddhavācana had mistakes in it. 

What Buddhism does explain is this: there is a real state of contentless awareness that a human being can reach by withdrawing attention from sensory experience and enduring the resultant sensory deprivation (often accompanied by hallucinations). 

In terms of conditionality we have to leave dependent arising behind in explaining contentless awareness. This is the state that occurs when all conditions for sensory experience have been eliminated through the practice of nonapprehension of sensory experiences. Philosophers would refer to it as a non-intentional conscious state. That is to say, a conscious mental state that is not "aimed at" anything. 

However, the vast majority of Buddhists reify this state and talk about it as "reality". The word may be explicit (especially in English language publications) or it may be implicit. And this is ironic for several reasons. There is no word in Indic languages that quite covers the same semantic field as the abstract noun "reality". Reality is never explicit in Pāḷi, for example, because there is no such term in Pāḷi. The same applies for Sanskrit and Chinese. Buddhist accounts of reality are often merely silly, like the common claim that "mind creates matter". It doesn't. Moreover, Buddhist texts don't claim that it does, because they almost never discuss "matter" (dravya) in relation to mind; they discuss phenomena (dharma) which they clearly distinguish from objects. 

My conclusion is that traditional Buddhism does not provide me with good general explanations. Modern Buddhism does little better since it subordinates all knowledge to the claims of the traditional buddhist: where a Buddhist acknowledges the validity of a scientific explanation, for example, they almost always assert that the Buddha got there first. For Buddhists, science or any foreign body of knowledge can only ever be a handmaiden to Buddhism. In this all too common scenario Buddhists co-opt the explanatory power of science to assuage their anxieties over orthodoxy and to bolster metaphysical claims (that are otherwise unreasonable).  Buddhism does not provide me with good explanations of reality, truth, or any other metaphysical issue. Buddhists have no explanation whatever of morality, just a bunch of lists of moral rules. As Damien Keown long ago noted, there are no ethical treatises in traditional Buddhist literature. 

Buddhism can do a decent job of explaining the state of contentless awareness that follows the cessation of sensory experience. And yet I know that scientists are now exploring this phenomenon and are likely to do a better job of explaining it, just as they do for other events and states. 

We have an interlocking network of good explanations of reality in science (in the form of theories that satisfy many levels of questioning). Science gives us extremely accurate and precise values for some of the physical properties of changing systems. However, even science is not a complete description of the world as it stands because the vast majority of scientists are committed to a reductionist explanatory paradigm: any event or state is best explained in terms of events and states at a lower order of structure and/or complexity: the parts explain the whole. Only... bricks and mortar don't explain architecture. I don't accept reductionism as a complete metaphysics. Rather, I hold that structure is equally important to substance, and that in order to have knowledge of structures we cannot employ the destructive analysis of reductionism (which destroys structures). Any good explanation will explain both substance and structure and privilege neither. So, while I am critical of Buddhist explanations, I'm not advocating scientism as an alternative. As useful as reductionist science undoubtedly is, we have a long way to go understanding the place and role of structure. 



Faye, Jan.(2007). "The Pragmatic-Rhetorical Theory of Explanation." In Rethinking Explanation. Boston Studies in the Philosophy of Science. 43-68. Edited by J. Persson and P. Yikoski. Dordrecht: Springer.

White, Curtis. (2023). "How Corporations Attempt to Co-opt Buddhism." Yes Magazine. Jan 24, 2023. https://www.yesmagazine.org/opinion/2023/01/24/corporate-buddhism

Wilkinson, Sam. (2014). "Levels and Kinds of Explanation: Lessons from Neuropsychiatry." Frontiers in Psychology 5 (article 373): 1-9. 

27 January 2023

Reading rūpa (phenomeno)logically

Because I read everything that academics publish on the Heart Sutra, I see a lot of translations of the Sanskrit word rūpa (Chinese: 色). The most common translation of the word is "form", but one often sees it translated as "body" or, especially in relation to Tibetan Buddhism, even as "matter". Even when selecting "form" many translators and commentators appear to have substance in mind. Over the years I've become convinced that this must be incorrect.

I've written about rūpa before, especially in the context of the skandhas, based on extended essays on the khandhas in Pāli found in Vetter (2000) and Hamilton (2000). As I noted in my previous essay, the modern definition of rūpaskandha has been influenced by an ancient mistake in the Khajjanīya Sutta (SN 22.79), which related the word to a verb ruppati "to strike" (Skt rupyati), when in fact it was meant to be rūpayati "appearing", a denominative extrapolated from the noun. However, the denominative doesn't occur in Pāli, which may explain the incorrect etymology that follows from ruppati.

In this essay I want to try to show the utility of what I'm now calling a "phenomenological approach". I had been calling it "epistemic", but I realised that the focus is phenomena and the cessation of all phenomena in meditation. The result is certainly a kind of knowledge, prajñā, but the focus is on phenomena and prajñā appears to be context dependent. I believe, but cannot yet prove, that the cessation of experience in meditation, and the subsequent contentless awareness (aka emptiness), were common knowledge in ancient India. Contrary to the perennial philosophy, I do not believe that all religions point to a single truth. Rather, I believe that each sect interpreted contentless awareness in their own way, giving us a multitude of religions all based on one kind of experience, but quite diverse in how they understood the meaning and significance of contentless awareness.

We can begin with dictionary definitions, but in order to understand rūpa we need to see how it is used in context. I will argue that, in a Buddhist context, we should always at least try to understand it in terms of phenomenology. In this view, rūpa always means "appearance" whether visual or across sensory modes.

The noun rūpa refers to "any outward appearance or phenomenon or colour" (Monier-Williams). Mayerhofer (1976: III 70-71) tells us it means "appearance, colour, shape, beauty". According to William K. Mahoney (1998: 247, n.5), the word is based on the verbal root rūp "to exhibit" or "display"; however, there is no such root in W. D. Whitney's The Roots, Verb-forms and Primary Derivatives of the Sanskrit Language. Monier-Williams does include an entry for rūp but states the verb is probably a denominative, which is to say that the verb rūpayati means "to appear; appearing" since, in this case, it derives from the noun rūpa "appearance". As noted there is an unrelated root √rup "to harm" which forms a third person singular indicative rupyati "he/she/it harms" (Pāli ruppati). For example, in the Sutta Nipāta someone who is deprived of kāma is sallaviddhova ruppati "hurt as though pierced by an arrow" (Sn 767).

In Pāli, rūpayati and ruppati became confused. This may be because rūpayati doesn't occur in Pāli until the medieval commentaries began to be composed (likely by people knowledgeable about Sanskrit or, at least, grammar). Around one third of the occurrences of ruppati in the suttas occur in the Khanjjanīya Sutta. I've previously noted that I became aware of this because in the Pañcaviṃśatisāhasrikā Prajñāpāramitā we find:

yā śāradvatīputra rūpaśunyatā na sā rūpaṃ... tathā hi śāradvatīputra yā rūpaśunyatā na sā rūpayati | (Zacchetti 2005: 393)

Śāriputra, that absence of appearance, is not appearance... since the absence of appearance does not appear.

And the same is true of the other skandhas (vedanā does not vedayati and so on). As it happens this is the line immediately preceding the opening line of the core passage of the Heart Sutra. What comes next is the question "And why is that?" (tat kasya hetoḥ) and the answer is the now famous lines (though in their generally unfamiliar original form ): "For, Śāriputra, appearance is not different from emptiness; " (na hi śāradvatīputra anyad rūpam anyā śunyatā).

Given this as a starting point, we now need to look more closely at how the word is used in a Buddhist context.

Buddhist Usage

The Buddhist account of sensory perception is spelled out in some Pāḷi texts, e.g. The Loka Sutta (SN 12.44):

And what, monks, is the origin of the world. Conditioned by eye and appearances, visual discrimination occurs. The three together are contact. With contact as condition, valence occurs; with valence as condition, desire occurs; with desire as a condition, grasping occurs; with grasping as condition, becoming occurs; with becoming as condition, birth occurs; with birth as condition, aging and death occur: grief, lamenting, misery, depression, and despondency are born. This, monks, is the origin of the world.

Katamo ca, bhikkhave, lokassa samudayo? Cakkhuñca paṭicca rūpe ca uppajjati cakkhuviññāṇaṃ. Tiṇṇaṃ saṅgati phasso. Phassapaccayā vedanā; vedanāpaccayā taṇhā; taṇhāpaccayā upādānaṃ; upādānapaccayā bhavo; bhavapaccayā jāti; jātipaccayā jarāmaraṇaṃ sokaparidevadukkhadomanassupāyāsā sambhavanti. Ayaṃ kho, bhikkhave, lokassa samudayo.

This pericope is repeated for each of the sensory modes.

  • eye and appearance give rise to eye-discrimination
  • ear and sound give rise to ear-discrimination
  • nose and smells give rise to nose-discrimination
  • tongue and tastes give rise to tongue-discrimination
  • body and tangibles give rise to body-discrimination
  • mind and dharmas give rise to mind-discrimination

In the Sabba Sutta (SN 35.23) and elsewhere, these twelve items (eye and appearances, etc.) are called āyatana and they are everything (sabbaṃ). In the Pahāna Sutta (SN 35.24), the twelve āyatana are to be abandoned (pahātabba). In terms of understanding, and therefore translating, rūpa, we can see that rūpa stands in the same relation to the eye as sound does to hearing, or the body to the sense of touch. Certainly, in this context, rūpa is not "body".

Note that, for in each sensory mode, there is no consideration of the object being sensed; rather, we have the sensory organ (eye, ear, nose, tongue, body, mind) and something (appearance, sound, smell, taste, touch, dharma) that crosses the space between us and the object so that it contacts (phassa) our sensory organ and causes the arising of discrimination (viññāna). Here, then, rūpa is not "the object of perception", it is the medium through which visual objects are perceived.

In modern terms the something, the medium, related to the visual sense, is simply the light reflecting off the object and striking the eye, from which we extract visual information concerning the colour, shape, contrast, edges, etc of the object. While we can think of rūpa as reflected light, it would be anachronistic to use this concept in talking about early Buddhist religious doctrines. The ancient Buddhists do not seem to have thought in terms of reflected light facilitating seeing. Rather they thought in terms of a vague "visual appearance" emanating from the object and striking the eye to create a moment of visual awareness.

Evameva kho, bhikkhave, assutavā puthujjano cakkhusmiṁ haññati manāpāmanāpehi rūpehi (SN IV 201)

Thus also, monks, the uneducated hoi polloi are struck in the eye by pleasant and unpleasant appearances.

If rūpa were the object of perception, then being "struck in the eye by an appearance" (cakkhusmiṁ haññati rūpehi) is not vision, it's a trip to the hospital. There is no object that strikes the eye and gives rise to pleasure.

Corā gāmaghātakāti kho, bhikkhave, channetaṃ bāhirānaṃ āyatanānaṃ adhivacanaṃ. Cakkhu, bhikkhave, haññati manāpāmanāpesu rūpesu; sotaṃ, bhikkhave…pe… ghānaṃ, bhikkhave…pe… jivhā, bhikkhave, haññati manāpāmanāpesu rasesu; kāyo, bhikkhave…pe… mano, bhikkhave, haññati manāpāmanāpesu dhammesu. (SN IV.175)

Monks, "like thieves who attack a town" is a way of talking about the six external senses. Monks, the eye is attacked (haññati) by pleasant and unpleasant appearances; the ear..., nose.., the tongue is attacked by pleasant and unpleasant tastes; monks, the body... the mind is attacked by pleasant and unpleasant thoughts.

From all this we can deduce an important definition:

rūpa is to the eye as sound is to the ear

This is a general definition of rūpa in a Buddhist context. Making analogies was a very popular method of inferring knowledge in ancient India, so it is fitting that our general definition of rūpa should emerge from an analogy. Rūpa is not, and cannot be "the body", since in this context "the body" is part of the scheme as kāya (literally "a collection") and is in a different category, i.e. the category of sense organs, not the category of sense media. Moreover, rūpa cannot be the object of vision, since the object itself striking the eye does not result in vision, usually the opposite. Rather, we visually know an object (cakkhu-viññāṇaṃ) by seeing the light reflected from it. This reflected light is conceptualised in Pāli as "appearance" (rūpa).

However, this definition raises a problem in terms of how rūpa is used in other terms, notably rūpaskandha and nāmarūpa. In these terms rūpa is traditionally defined as meaning "the entire physical universe" and "the body". And neither of these definitions offers much in the way of coherent interpretation. How can one word mean both "visual appearance" (and definitely not "body" which is kāya), "body", and "the entire physical universe"? My answer is that it cannot. It does not. This interpretation is old, but it misses the point.

So how does the phenomenological reading help here?


As noted in previous work on rūpakkhandha, the traditional definitions are based on incorrect information, especially the folk etymology in the Khajjanīya Sutta (SN 22.79). As Sue Hamilton (2000: 70) noted: "there is no text which gives a full and clear account of what is being referred to by the term khandha.". Still, there is another way in which rūpakkhandha is defined, in the Mahāhatthipadoma Sutta (MN 28) that carries a little more weight and is more amenable to metaphysics:

Katamo cāvuso, rūpupādānakkhandho? Cattāri ca mahābhūtāni, catunnañca mahābhūtānaṃ upādāya rūpaṃ. Katamā cāvuso, cattāro mahābhūtā? Pathavīdhātu, āpodhātu, tejodhātu, vāyodhātu.Katamā cāvuso, pathavīdhātu? Pathavīdhātu siyā ajjhattikā, siyā bāhirā. Katamā cāvuso, ajjhattikā pathavīdhātu?

Yaṃ ajjhattaṃ paccattaṃ kakkhaḷaṃ kharigataṃ upādinnaṃ, seyyathidaṃ – kesā lomā nakhā dantā taco maṃsaṃ nhāru aṭṭhi aṭṭhimiñjaṃ vakkaṃ hadayaṃ yakanaṃ kilomakaṃ pihakaṃ papphāsaṃ antaṃ antaguṇaṃ udariyaṃ karīsaṃ, yaṃ vā panaññampi kiñci ajjhattaṃ paccattaṃ kakkhaḷaṃ kharigataṃ upādinnaṃ. Ayaṃ vuccatāvuso, ajjhattikā pathavīdhātu. (MN I 185)

And what, comrade, is the branch whose fuel is appearance? It is the four great "beings" (bhūtā) and the appearances dependent on them. And what, comrade, are the four great "beings". The elements (dhātu) of earth, water, heat, and wind. And what, comrade, is the earth element. The earth element may be internal or external. And what, comrade is the internal earth-element?

That which is internal to oneself and is hard (kakkhaḷa) or solid (kharigata) when grasped, such as: head hairs, body hairs, fingernails, teeth, skin, flesh, sinews, bones, marrow, kidneys, heart, liver, diaphragm, spleen, lungs, intestines, bowel, stomach, excrement

And the proper attitude to all this solidity is

taṃ netaṃ mama, nesohamasmi, na meso attā (MN I 185)

This is not mine, I am not this, this is not the essence of me.

This is all usually construed in a metaphysical framework because our Theravāda informants tell us that they believe that it concerns metaphysics (based on their traditional commentaries on the Abhidhamma). This reinforces the sectarian idea that rūpakkhandha is substance generally. As I say, I no longer believe this to be true or even plausible.

A lot of this material comes either from the general Indian background at the time of the second urbanisation (from ca 600 BC onwards) or from Brahmanism. We learn a lot about the conception of the elements from Vedic texts. For example, the idea of the element of tejo "heat", is not simply "fire". Rather, the element of tejo is conceptually connected to actual flames, but also to the sun, anything hot, and digestion. Fire might be the prototype that defines the category, but clearly the category itself is not composed only of "fire". Moreover, the general word for fire is agni (Pāḷi aggi).

This language of prototypes defining categories and membership of the category being based on similarity to the prototype comes from George Lakoff's expansion on Wittgenstein's family resemblances, especially as set out in his book Women, Fire, and Dangerous Things (1987). I've often cited Lakoff's work on metaphors, but his approach to categorisation is useful here. The "beings" (bhūtā) in this view represent prototypes (or models) by which the categories are defined. Earth is the conceptual model for defining solidity, water for cohesion, heat for transformation, and wind for movement. We could still define these in ontological terms (leading to metaphysics), but we can also define them experientially (leading to phenomenology).

For example, if we grasp some hair we will have an experience of solidity or resistance. Hair is an example of "earthiness" because the solidity of hair is analogous to the prototype of solidity, i.e. the earth. Moreover, although Buddhaghosa goes into similar detail about where we experience "earth" in the body (Vism XI.27 ff), it's clear that this is not an ontology (i.e. not an account of real things) and not intended to be an ontology; this is a meditation practice. Making it into an ontology was a project of later Buddhist monks who seem to have completely missed the point (perhaps because they didn't meditate).

This means that we are not forced to read rūpakkhandha (metaphysically) as "the body" as Vetter and Hamilton suggested. We might even say, given comments on rūpa above, that it would incoherent to think of rūpakkhandha as "the body". There is no doubt that we can experience solidity in our body as well as outside it, but our body is not the prototype for solidity, earth is the prototype.

This opens up the possibility of reading rūpakkhandha phenomenologically, in which case the way the word is used ought to reflect the basic meaning, i.e. appearance. I take rūpakkhandha to refer to the "appearance" of sensory experience across the six sensory modes. Here rūpa is a metonym for appearance across the sensory spectrum: appearance, sound, smell, taste, tangibles, and thoughts (rūpa, saddo, gandho, raso, jivhā, phoṭṭhabbo, dhammo).


Like rūpakkhandha, there is a long tradition of treating nāmarūpa as an ontology. In this ontology we divide the world into "physical" and "mental". We have Pāli technical terms for these categories, viz. kāyika and cetasika. These terms are adjectives meaning "concerned with or pertaining to the body (kāyo)" and concerned with thought (ceto). Theravāda exegetes take kāyika to be another metonym for the entire physical world, along with with rūpa. This is a dualistic ontology that simply divides the world into material and non-material. We are quite familiar with this dualistic ontology in Europe, because it was central to our intellectual tradition. For Descartes, for example, this dualism allowed a place for God in an otherwise materialistic universe.

That said, we can make a valid epistemic distinction between what we know about what goes on in our body and what we know about the world. This is the distinction between internal (ajjhattika) and external (bāhirā). The distinction is epistemic because we get different kinds of information from our different senses. Interoception gains us knowledge of our internal physical state, and exteroception knowledge of the external world. We may well infer metaphysical conclusions from such knowledge, but we don't have to. We can think of this as a basic distinction between the kinds of knowledge we can have and how we get it. And, notably, different people infer different metaphysics apparently from the same kinds of experience (especially where meditation is concerned).

Again, we are not forced to read nāmarūpa as an ontology or as rooted in a particular kind of metaphysics. I think we can read rūpa here as appearance also. Note that as a nidāna, nāmarūpa is conditioned by viññāṇa, which I have defined as "discrimination of the object ". I have contrasted this with saññā which is "recognising (and thus naming) the experience". Two different kinds of knowledge, both sharing the etymological root jñā "to know".

In the Mahānidāna Sutta (DN 15) we find a curious variant of the nidānas. In this sutta, viññāna is the condition for nāmarūpa, but nāmarūpa is also the condition for viññāna. In the usual presentation of nidāna doctrine as an ontology (a way of dividing up reality) this doesn't make any sense. How can two things mutually condition each other? Conditionality requires the presence of the condition for the effect to arise. If the effect is also the condition for the condition, then conditionality ought to break down. Thus, at face value, in an ontological reading, this sutta is incoherent.

In a phenomenological reading viññāna involves identifying the object of experience from the experience of it. We can do this because the experience has a sui generis quality (or a sabhāva; Skt svabhāva). We know from the inside that greed and generosity are different, that anger and love are different. And such differences are essential to Buddhist soteriology because they influence how we act and, therefore, where we are reborn. This in turn either supports or undermines our attempts to end rebirth (the ultimate goal of all Buddhist traditions). As we know from considering the khandha doctrine, viññāna depends on the appearance (rūpa) of a sensory experience and putting a name (nāma) to it. In a sense ,then, viññāna and nāmarūpa are two different ways of talking about the same aspect of sensory experience.

Another aspect of the ancient account of sensory experience is that when an object is discriminated we name that discrimination according to which sensory mode it occured in:

yaññadeva, bhikkhave, paccayaṃ paṭicca uppajjati viññāṇāṃ tena teneva saṅkhaṃ gacchati. cakkuñca paṭicca rūpe ca uppajjati viññāṇaṃ, cakkhuviññāṇanteva saṅkhaṃ gacchati... (MN I 259)

Whatever condition gives rise to a discrimination, it goes by that name. A discrimination arising on the basis of the eye and appearances, goes by the name "eye-discrimination"...

I think we cannot overstate the importance of this. "Goes by that name" translates saṅkhaṃ gacchati. If, for example, there is a viññāna dependent on cakkhu and rūpa, then the name (nāma) of that is cakkhuviññāna. How do we know that it is a cakkhuviññāna? Because we perceived it via the eye and because it has the appearance of a visual percept. Vision is different from other modes of perception, i.e. vision has a sui generis character than enables us to distinguish visual sensory experience from other kinds of sensory experience.

We may notice that all these doctrines that Theravādins treat as ontologies are simplistic. The idea that just five categories cover all possible phenomena is overly simplistic. The fact is that these doctrines are mostly meditation practices rather than ontologies. They are subjects to contemplate in an attempt to gain liberation, not existential explanations. They all seek to undermine our conviction that sensory experience is the acme of being, because the authors were familiar with the idea of the cessation of sensory experience and possibly with the experience of cessation.

Early Buddhists (and Prajñāpāramitā Buddhists) did not, for example, make much of the contrast between pleasant and unpleasant sensory experiences. Clearly there is some discussion of these in relation to karma and rebirth. But the much clearer contrast is between any sensory experience (kāma) and no sensory experience (aka nibbāna, suññatāvhāra).

What the Mahānidāna Sutta appears to argue is that discrimination of the object and its name and appearance are much the same thing. That if they really are two different processes, then they coexist. And this is a phenomenological argument, based on epistemology; it is not a metaphysical argument based on inferences drawn from experience.


In this essay I've tried to make two co-existing arguments. On one hand the traditional interpretation of rūpa is incoherent in many ways, largely because it is misinterpreted as "body" or "substance". This seems to be due to a tendency towards metaphysical speculation in the post-canonical period (and up to the present). On the other hand, a phenomenological reading of rūpa as "appearance" stops short of inferring a particular metaphysics and avoids the kind of incoherence that we see in, for example, Theravāda orthodoxy.

We notably conclude that: rūpa is to the eye as sound is to the ear.

In a Buddhist context, then, rūpa means "appearance". In modern terms, this amounts to reflected light, but I have tried to avoid the anachronism of attributing knowledge of physics to the Iron Age authors of the Pāli suttas. The translation "appearance" is a much better reflection of the state of their knowledge of the physics of visual perception.

I have tried to show that we can retain this approach to better account for the idea of rūpakkhandha. Here, also, rūpa means "appearance" but in a more general way. And the same applies to nāmarūpa

I believe that this means we are not forced to take the metaphysical speculations of Theravādins, or any other Buddhists, seriously. We can, and I argue that we should, read these passages as concerned with phenomenology in the service of soteriology. Early Buddhists did not speculate about the nature of phenomena or the nature of reality. Because they can, and did, reduce sensory experience to the point where from their point of view it was completely absent (suñña), they were simply not interested (at first) in the nature of experience, they were interested in the implications of the cessation and absence of sense experience in meditation. 



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

Lakoff, George. 1987. Women, Fire, and Dangerous Things. Chicago University Press

Mahony. William K. (1998). The Artful Universe: An Introduction to the Vedic Religious Imagination. State University of New York Press.

Mayrhofer, Manfred. (1976) Kurzgefaßtes etymologisches wörterbuch des Altindischen. A Concise Etymological Sanskrit Dictionary. Heidelberg: Carl Winter. Universitätsverlag.

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