11 October 2019

Heart Sutra: Work, Text, Document

Some time ago I uploaded a draft of an eclectic edition of the Chinese Heart Sutra (Xīnjīng) for comment on academia.edu. Richard K. Payne responded with a terse, but ultimately very interesting question: "what do you see as the value of critical editions in general?"  I had to confess that I was not sure. Some years later and  I'm struggling with the plethora of versions and variations on the text. People keep asking me for my translation of "the Heart Sutra" but there isn't just one Heart Sutra to translate. There are many. The history of the text is complicated.

The problem is non-existent for most people since there is a canonical version. But the thing is that almost no one uses the canonical version unchanged. The version in the popular Taishō Shinshū Daizōkyō edition of the Chinese Tripiṭaka has a mistake in the dhāraṇī and the popularly chanted version is better. This is just one character out of about 250, but still it means that one cannot simply take the canonical version on trust.

Worse, Huifeng (now Matthew Orsborn) has convinced me that there were two other errors in the existing versions of the Xīnjīng. One goes back to Kumārajīva's 5th Century translation of the Large Sutra, and the other probably dates from the mid 7th Century when the Xīnjīng was composed. The former is moderately serious since it confuses the meaning of the text at that point, but it also contributes to obscuring the latter mistake which is much more serious and completely changes how we even approach the text. We definitely want to repair this damage but it dates from the 7th Century and has (more or less) always been this way. What would be the warrant for changing something that has been accepted as authentic for 1300 years?

Subsequently, I looked into the old inscriptions of the Xīnjīng. I located a full image of and transcribed the Beilin Stele, believing Kazuaki Tanahashi's claim that this was the earliest dated Xīnjīng (678 CE) and noted several places where the scribe used different characters with more or less the same meaning. Then I discovered the Fangshan Stele is considerably older (661 CE). I noted that it too had different characters to the canonical version (my article on this is due out any day now).

So which version should I translate? The canonical? The popular? The oldest? Or should I fix the errors, creating a new variant that had never existed before and translate that. I had a hunch that no Buddhists in the world would accept a change to the ancient text. Probably no academics would either. So was I stuck trying to translate a defective text, knowing it to be defective and gritting my teeth while transmitting a falsehood that was accepted because it had become canonical?

In parallel to this I was still working on the Sanskrit text, following up work done by Jan Nattier. She had concluded, quite rightly, from her investigation, that the Hṛdaya was a translation of the Xīnjīng. My own work repeated Nattier's method on another copied passage (the "epithets") and drew the same conclusions. Then, I believe put the conclusion beyond any doubt by finding a Chinese idiom encoded in Sanskrit in a portion of the text that was not copied, but composed. This showed that the language of composition was Chinese.

As part of this I had identified and corrected two errors by Conze and shown that the phrases rūpaṃ śūnyatā śūnyataiva rūpaṃ  rūpān na pṛthak śūnyatā śūnyatāyā na pṛthag rūpaṃ had history linking them back to very old Buddhist similes exemplified by rūpaṃ māyopamaṃ "appearance is like an illusion". I had also begun to assemble the Sanskrit parallels to the copies passages in the Xīnjīng as they appeared in the Giligit manuscript of the Pañcaviṃśatisāhasrikā Prajñāpāramitā Sūtra (Pañc). And it showed that the person who created the Hṛdaya was unaware of the idioms of Pañc. A Hṛdaya reconstructed in idiomatic Buddhist Sanskrit looked very different, indeed. I could put all this together into a text, but such a text had never existed before.

I had been somewhat resistant to committing to making a translation while the source text remained unclear. This seemed to frustrate some of my colleagues because they all assumed that the source text was a settled matter. My last paper on the source text is out being reviewed right now. I never know what reviewers and editors will find. There are usually many problems to overcome to get to the point of publication. Until the last article is published, I regard the matter as unsettled. Nonetheless, I am confident about the conclusions of that forthcoming paper, partly because it rests on the excellent work by Jan Nattier and Matthew Orsborn. And partly because my old Sanskrit teacher seemed very impressed by the argument. So I began to think about translating from three texts:
  1. A repaired version of the received Sanskrit text (which is expected but, in fact, a bit pointless given what we now know about the history of the text). 
  2. A repaired version of the received Chinese text.
  3. An idiomatic Sanskrit translation of No. 2.
And then I thought about Richard Payne's question again and I re-read an article by Jonathan Silk (2015). Payne also pointed out that Lewis Lancaster has written on this subject (and I still need to read those articles). This raised some serious questions about my project. This essay is part of my attempt to see beyond a conceptual impasse.


As I understand it, the methods of modern philology were born out of European imperialism. As Europeans conquered and occupied the so-called Middle East (including eastern North Africa, the Levant, and Arabia) they began to discover old manuscript copies of the Bible in Greek and other languages like Syriac. These were different to the versions of the Bible in Europe. And this might not sound so surprising in our modern world, but back then the Bible was the literal word of God and if there were different versions of the it, then this was a disaster. So they developed methods for reconstructing what they called the "ur-text", i.e. the original text as composed by God and obscured by man. This turned into classical textual criticism, which consists of four stages:
  1. Heuristics
  2. Recensio
  3. Emmendatio
  4. Higher Criticism
 The goal here was to recreate the Word of God, to rediscover the intention of God. There are some serious questions about whether this approach applies to a literature that is not considered a divine revelation. For example, there is a real question about my wish to revise a text that is unquestionably accepted as the Heart Sutra amongst Buddhists in China and Japan. Such people have every right to go on using the familiar text and my attitude might well be seen as imperialist and colonialist. Just another stupid fucking white man who thinks he knows better. I don't think this is my attitude but, nevertheless, I'm taking the time to think about what my project is and who it is for. 

In his recent article Silk (2015:205-6) draws out a threefold distinction first made by Chaim Milikowsky. First we have the Work, which is the author's or editor's product. This may only exist conceptually and never have been committed to words. Or the author may have attempted to put it in words and be more or less satisfied that the result, but still consider this as inferior to their conception of the Work. A presentation of the Work in words is a Text. A single Work may generate multiple Texts; for example, one story that is told many times, but with minor differences each time. No single Text is the "original" in this case, because the Text is not the Work. Lastly a Document is some physical instantiation of a Text. Typically, in studying Buddhist manuscript cultures, we are faced with multiple Documents representing multiple Texts. This is certainly in the case of the Heart Sutra.

One can see this hierarchy in my outline of the problems with the Heart Sutra. The Work was conceived, probably by Xuanzang and a Text was created. Early evidence is that the Text was quite fluid at first. Characters were freely substituted by different scribes (who thus become secondary authors). Various Documents were created on the basis of the different Texts - the oldest we still have is carved in stone and dated to 13 March 661. The oldest plausible literary reference is from 688 but refers to an event in 656 CE. The dhāraṇī was probably copies from another Text (Dhāraṇīsamuccaya) translated into Chinese in 654 (T 901).

After this various elements of a standard myth appeared over the next few decades, especially in the Biography by Huili and Yancong (688 CE), and in the Kaiyuan Catalogue (730 CE). There appears to be a concerted effort to promote the idea that the Work is Indian, that the Chinese Text is the result of the translation by Xuanzang. Later, other stories about translations are added.

In the midst of this a Sanskrit Text was created by a Chinese monk who translated the Xinjing into Sanskrit without much of a grasp of Buddhist Sanskrit idiom. In all likelihood a Document was produced to make the story plausible. This Text purports to represent the Work; or to at least be the ur-Text from which the Chinese texts derive. This attempt at deception succeeds to the point that the Xinjing (of which it is a translation) is traditionally misinterpreted due to a mistranslation by the Translator.

Also around 730 CE the Damingzhoujing is constructed which remains more faithful to the Kumārajīva Large Sutra, and thus purports to predate the standard Text and better represent the Indian Work. Is the Damingzhoujing a new Work, or merely a new Text? The earliest surviving Document of the Damingzhoujing is from the 11th Century. Consider this, the Damingzhoujing stays faithful to the Large Sutra translation by Kumārajīva (T 223); it copies from it word for word. And this means that whoever created it must have known that the Xīnjīng was not a translation, but rather a selection of copied passages from the Large Sutra. The author of the Damingzhoujing restored a missing line (character for character) from the middle of the longest copied passage (often called the "core section") which could only have come from the Large Sutra. We have been deliberately deceived about the provenance of the Hṛdaya and also of the Damingzhoujing

The story is still incomplete, because another editor saw fit to provide the text with the requisites that it lacked: a nidāna starting "thus have I heard", announcing the location and occasion for the teaching; the addition of the the Buddha who endorses the teaching of Guanyin; and a colophon including praise from the audience. This extended version of the text never took off in China. No commentaries seem to have been composed on it, though it was translated from Sanskrit into Chinese four times, suggesting perhaps that the additions were made in Sanskrit (we could check the idiom of the extension to see if they better fit the Indian Buddhist context). The extended became the standard in India (eight commentaries survive in Tibetan translation), Nepal, and Tibetan (where it seems to have been translated twice).

For a text of about 250 words (about 1/3 of an A4 page in Times 12 point) this seems an absurdly complex and convoluted history. The Work is reflected in various Texts, and the Texts each in various Documents. In addition, there is a trail of seemingly deliberate misinformation about the provenance of the various Texts and how they relate to each other.  

I'll pause here, but there are several more layers of the onion to peel yet. The next step will be to the look at the composite nature of Buddhist texts. 



Jonathan A. Silk (2015) 'Establishing­/­Interpreting­/­Translating: ­Is­ It­ Just­ That­ Easy?' Journal of the International Association of Buddhist Studies. Vol. 36/37: 205-225.

Watanabe, Shōgo. (1991) 「般若心経成立論序説」 『仏教学』 “An introduction to the Theory on the Formation of the Prajñā-hridaya-sūtra,” Journal of Buddhist Studies 31, (July): 41-86. [Japanese]

13 September 2019

Another Failed Attempt to Refute the Chinese Origins Thesis

In 2002, Japanese scholar Harada Waso published an annotated translation of the Heart Sutra which included many notes on why he did not believe the Chinese Origins thesis. The article was written in Japanese and remains untranslated, but amidst the background disputes over the Heart Sutra article on Wikipedia (now edited by religious fanatics), user Pat457 (25 July 2017) gave a fairly long summary of the arguments in Harada (2002). I have no way of knowing how accurate or reliable this English resumé is but I here I will take it at face value and explain why the argument is wrong.

There are two main theories for the origins of the Heart Sutra: Indian and Chinese. It has never been disputed that the Heart Sutra reuses text from the Large Perfection of Insight Sutra. What is in dispute amongst some (mainly Japanese) scholars is in which language the copying occurred. Jan Nattier (1992) was the first to propose that the copying occurred in Chinese. I've since published a number of analytical articles showing why this has to be the case. Very few, if any, other scholars have really got to grips with the materials and methods used by Nattier so far as I can tell from their published work. So the dispute has been carried out in a superficial manner. Sadly, Nattier received almost no support from Western scholars when Buddhist establishment figures published apologetics for the traditional accounts.

The reuse of texts was prevalent in early medieval China. Chinese Buddhists created a unique genre of texts that bibliographers at the time called chāo jīng 抄經 or "digest texts". Hundreds of digest texts were in circulation by the early 6th Century. The digest text is a distinctively Chinese genre; there is nothing like it in Indian Buddhist literature. The Heart Sutra is self-evidently not a sutra (and the earliest Chinese commentaries acknowledge this). It reuses passages from the Large Sutra and perfectly fits the description a digest text. It appears to be the only one of its kind ever to be translated into Sanskrit, i.e. there are no other Sanskrit digest texts.

The fact that large parts of the Heart Sutra were copied gives us leverage on the problem of where the text was composed. Copying conserves the original whereas translation does not. When an original and a copied text are translated independently into another language changes are likely to be introduced. And where a copy in a second language is translated back into the first language (i.e. translated twice), this will inevitably result in some idiosyncrasies.

The source, in this case is the Large Sutra. We have various witnesses of this text: two recensions in Sanskrit; i.e. Gilgit 6th C and Nepalese 19th Cl along with three complete translations in Chinese T 221 (291 CE), T 223 (404 CE), and T 220-ii (663 CE) and one incomplete translation, T 222 (286 CE). Another complete text is embedded in a commentary (T 1509). The copy is the Heart Sutra which exists in many forms, including manuscripts, inscriptions, and edited versions in both Sanskrit and Chinese.

The Sanskrit Heart Sutra is packed with odd vocabulary, expressions, and idioms. Some of these are neologisms but others appear to be calques and idioms from Chinese. A "calque" is a word or phrase borrowed from another language by literal, word-for-word or root-for-root translation. For quick reference I made a list of these: The Chinese Origins of the Heart Sutra Revisited (19 January 2018). This is what we predicted for a back-translation.

Nattier identified a clear pattern with respect to the "core passage" and it only supports the Chinese origins thesis. Huifeng (2014) made a vital contribution to our understanding of the text which supports Nattier's conclusion. I've published five articles to date; no.6 is due out any day, no.7 is being reviewed, no.8 is awaiting an editor's initial response, while no.s 9 and 10 are being beta tested before submission. My work on the Heart Sutra has two main threads: 1. correcting the sloppy mistakes in Conze's Sanskrit edition (overlooked by other scholars); and 2. identifying other copied passages and showing that the pattern identified by Nattier in the core section also characterises the rest of the text. Indeed, last year I published what I consider a decisive argument, i.e. that tryadhvanvyavasthītāḥ sarvabuddhāḥ is a calque of the Chinese phrase 三世諸佛 sān shì zhū fó “all the Buddhas of the three times” (2018b).

In other words, not only is there overwhelming support for the Chinese Origins Thesis, but the Indian origins thesis has been refuted. Harada was writing in 2002 and was responding only to Nattier, though I can honestly say that none of my observations is rocket science and they ought to have been obvious to anyone. 

Problems with the Text

Harada is using a Sanskrit text of the Heart Sutra, but which text? I can't read Japanese and Pat457 doesn't say. But I can read Sanskrit and I'm familiar with the main Sanskrit versions. To start with, Harada gives the maṅgala as namaḥ sarvajñāya. This means he is not using Conze's critical edition, although the first iteration of the critical edition (Conze 1948) is cited in his bibliography (not the revised version 1967, or the popular version 1958/1975). Another clue is that Harada's text has na vidyā nāvidyā na vidyākṣayo nāvidyāksāyo which Conze, in a rare moment of clarity, realised was a late interpolation. The two terms nāvidyā and nāvidyāksāyo don't belong here in what is an abbreviated list of the 12 nidānas. Non-ignorance (na avidyā) plays no part here. It's a wild contradiction in terms since non-ignorance would be tantamount to enlightenment.

These two features suggest that Harada is using the Hōryūji manuscript (H) as his text. However, note that the sandhi in the maṅgala is wrong, it should be (and is in H) namas sarvajñāya. Another oddity is that H has no punctuation, but Harada's text has punctuation that is similar to Conze's and to Max Müller's (1881) diplomatic edition of H. It's possible that Harada is using Müller (1881) though this does not appear in his bibliography. The misplaced colon after vyavalokayati sma is a distinctive feature of Conze's edition. Harada's text is not H, nor any of Conze's variants, and does not appear to be Müller. So is he using an edition published only in Japan? Or has he created a hybrid for his own purposes? 

Wherever the text comes from, it has mistakes in it. As well as those noted, Harada's text has the same two mistakes found in all the versions of Conze's edition. In the first sentence, pañcaskandha is in the wrong case (Attwood 2015). It should be accusative plural, i.e. (with sandhi) pañcaskandhāṃs making it the object of vyavalokayati sma (and this is why the colon is extraneous). Secondly, following Müller (1881) and Conze (1967) Harada's text incorrectly has a full stop (period) after cittāvaraṇāḥ (Attwood 2018a). This leaves the next four words with no verb and no subject and thus they do not make a well-formed sentence. The short term solution is simply to remove the full stop, but there are much deeper problems that were partially addressed by Huifeng (2014) and which I forensically examine in my forthcoming article (2019b).

Now, these mistakes in the Sanskrit text ought to be obvious to anyone with basic Sanskrit. A transitive verb with no object adjacent to a noun in the nominative (meaning it has no relationship to any other word in the sentence) ought to ring alarm bells. Similarly, a sentence with no verb and no subject is not a properly formed sentence and any competent Sanskritist ought to have noticed this and corrected it. So this raises questions about Harada's competence as a Sanskritist. That said, many apparently competent Sanskritists did not notice these mistakes over the 70 years since Conze's edition was first published. Normally we'd notice a grammatical mistake precisely because a sentence doesn't make sense. I've argued (Attwood 2015) that the expectation of nonsense established by Suzuki and promoted by Conze made them and subsequent scholars insensitive to mistakes in the text. I have to say that this aspect of studying the Heart Sutra infuriates me. All the more so when incompetent commentators are cited as authorities.

Harada's Argument

We can now consider Harada's argument as presented by Pat457. I cite the whole of Pat457's text verbatim in blue and indented. My comments are interspersed in black.
Nattier doesn't give an answer as to why the shorter Sanskrit version does not contain the phrase 度一切苦厄 ("crossed over all suffering and affliction"). Harada cites Fumimasa Fukui's thesis (Hannya shingyō no kakushin (般若心経の核心 'The Core of the Heart Sūtra') in Toyo no Shisō to Shūkyō (東洋の思想と宗教 'Thought and Religion in the East') 4, Waseda Univ., 1987) that the core - the 'heart', if you will - of the Heart Sūtra is not so much the first half that speaks about emptiness, but the latter half that extols the merits of the Gate gate paragate...mantra. Fukui argues, and Harada apparently concurs, that the phrase 能除一切苦 ("able to remove all affliction") is actually the most important part of the sūtra - in fact, the very reason why the sūtra came to be so popular in China. The phrase 度一切苦厄 in the opening section - found in Kumarajiva (T. 0250) and Xuanzang (T. 0251), with equivalents in the longer versions of Prajñā and Li Yan (T. 0253: 離諸苦厄) and Prajñācakra (T. 0254: 離諸苦厄), but absent from other versions - is proposed to have been inserted by Kumarajiva in his version of the sūtra to prefigure 能除一切苦, which Xuanzang preserved in his own version. If, as Nattier said, the Sanskrit Heart Sūtra was a back-translation from Xuanzang's Chinese 'translation' (which in turn was based on Kumarajiva's Large Prajñāpāramita text) made in China, Harada argued that the omission of 度一切苦厄 would be an unthinkable move. (pp. 108-107/33-34)

It is quite true that Nattier doesn't attempt to explain 度一切苦厄. What can we say about this? Firstly, Watanabe Shōgo showed that the Damingzhoujing (T 250) was not by Kumārajīva. He called it a 偽経 (gikyō) "fake text". This echoes the term 偽經 wěi jīng (same characters and meaning) used by medieval Chinese bibliographers for texts which they did not accept as Buddhist texts. In an undated interview in Japanese, Watanabe says: 鳩摩羅什訳の『般若心経』は偽経であるという説が提示され、現在、学界で定説となっています。 i.e. “The theory that Kumārajīva’s Heart Sutra is a spurious scripture [偽経] was suggested and it has become an established theory in the academic world at present” (Translation by Dr. Jeffrey Kotyk; personal communication). I concur with Watanabe: the Damingzhgoujing is a fake, probably produced in the 8th Century to boost the myth of the Heart Sutra. Although Fukui missed this point himself, Harada includes the Watanabe article in his bibliography. We don't know why he ignored the finding. 

I have identified the probable source of this phrase 度一切苦厄, i.e. T 410 13.708.a26-7. The  《大方廣十輪經》 Dàfāng guǎng shílún jīng (T 410) is a translation of the *Daśacakra-kṣitigarbha-sūtra, made by an unknown translator during the Northern Liang Dynasty, ca. 397 – 439 CE. Note that the text used in Xīnjīng is not from Xuanzang's translation (T 411) because he translated this phrase as, 脫一切憂苦 (tuō yīqiè yōu kǔ). Unfortunately there are no Sanskrit versions left anymore, but the phrase probably translated something like sarvaduḥkhitaṃ samatikramati sma (note: the phrase sarvaduḥkh* is found only a once in the Nepalese recension of the Sanskrit Large Sutra and not in this phrase).

As Pat457 says, "Harada argued that the omission of 度一切苦厄 would be an unthinkable move". But no counterpart of this phrase is found in any Sanskrit manuscript. It is no less unthinkable that a random phrase would be added by a translator when it did not occur in their source. My opinion is that it makes less sense as an addition than it does as an omission. However, there is not enough evidence to resolve this issue and it should, therefore, be pegged as a mystery rather than disingenuously asserting one's opinions as facts.

Harada cites Fukui in arguing that "能除一切苦 ("able to remove all affliction") is actually the most important part of the sūtra". This seems very unlikely. It is based on a misreading of the text. As I showed in Attwood (2017), the latter half of the text does not "extol the merits of the Gate gate paragate...mantra" at all. In fact, the word mantra does not occur in the Chinese Heart Sutra or in the Sanskrit texts from which the epithets pericope comes. The epithets extol the virtue of Prajñāpāramitā (as is all too obvious in the Large Sutra, especially in Sanskrit) and the word 明咒 mistakenly translated as mantra in the Sanskrit Heart Sutra was intended to read vidyā (as it does in all the extant Sanskrit Prajñāpāramitā manuscripts) but in practice was read bright dhāraṇī (i.e. as two words) when the Heart Sutra was compiled in the 7th Century. Nor is the following spell a mantra; it is a dhāraṇī. Translating 咒 as mantra was simply a mistake. Most of this was stated explicitly in Nattier (1992: 211-3 n.54a - with credit to Yamabe Nobuyoshi who alerted Nattier to this issue but never published it himself). My article was more or less a systematic expansion of the point made by Yamabe and Nattier in fn 54a enabled by the power of electronic searching to systematically identify all the occurrences of the pericope.

 I recently discovered that the epithets did circulate separately in a version taken from the Small Sutra (aka Aṣṭasāhasrikā). However, Buddhists have always focused on the section that says 色不異空,空不異色;色即是空,空即是色。 "Appearance is emptiness, emptiness is only appearance, etc".

In proposing this theory about the "core", Fukui and Harada have made a easily avoidable error.

Harada answers Nattier's observation that the Heart Sūtra uses kṣaya for the Large Sūtra's nirodha by pointing out that the Sanskrit version of the Perfection of Wisdom in 8,000 Lines (Aṣṭasāhasrikā-Prajñāpāramitā) already uses the word akṣayatva (... avidyākṣayatvena subhūte bodhisattvena mahāsattvena prajñāpāramitā abhinirhartavyā / evaṃ saṃskārākṣayatvena vijñānākṣayatvena nāmarūpākṣayatvena ṣaḍāyatanākṣayatvena sparśākṣayatvena vedanākṣayatvena tṛṣṇākṣayatvena upādānākṣayatvena bhavākṣayatvena jātyakṣayatvena jarāmaraṇākṣayatvena śokaparidevaduḥkhadaurmanasyopāyāsākṣayatvena subhūte bodhisattvena mahāsattvena prajñāpāramitā abhinirhartavyā). Ergo, the Heart Sūtra's use of kṣaya is not unusual/without precedent. (pp. 96-95/45-46).

Except that kṣaya means "destruction" and akṣayatva means "indestructible" or Conze translated it as "non-extinction" (1973: 271). (i.e. the complete opposite): avidyākṣayatvena subhūte bodhisattvena mahāsattvena prajñāpāramitā abhinirhartavyā "Perfection of insight should be [internally] realised by the bodhisatva, mahasatva, through the indestructibility of ignorance." This comes towards the end of the text (and was probably added quite late). It's not entirely clear that it makes sense.

Note that Kumārajīva's text does not include an exact parallel. He does, however, have the immediately preceding passage with the same structure: 「須菩提!色無盡故,是生般若波羅蜜;受、想、行、識無盡故 (T 227; 8.578c22-3) "Subhūti [the bodhisatva] gives birth to the Prajñāpāramitā through the indestructibility of appearance". Xuanzang's translation does have a (abbreviated) parallel: 應觀無明乃至老死皆無盡故,引發般若波羅蜜多;(T 220; 7.315b1-3). Here akṣayatva = 無盡故; abhinirhartavyā = 引發; avidyā yāvan jarāmaraṇa = 無明乃至老死 .

Kumārajīva and Xuanzang confusingly use both 無明滅 and  無明盡 for avidyānirodha, though both have a preference for the former: T 223 4 x 無明滅 and 2 x 無明盡; T 220 43 and 2. Still, we know that the Sanskrit tradition is unequivocal in using nirodha. The confusion only exists in Chinese. There is a very long-standing convention of referring to the cessation of the nidānas (which is where Hṛd employs kṣaya) using the word nirodha - the same word in Pāli and Sanskrit and Gāndhārī (though the spelling is irregular in the latter). I know of no exception to this in Pāḷi or Sanskrit. The Pāli equivalent khaya is never found in this context (which is easily confirmed by electronic searching). In short using kṣaya here is simply a mistake; and a mistake that no Indian Buddhist would make. It's part of a pervasive pattern of this kind of mistake in the Sanskrit Heart Sutra that tells us the text was composed in Chinese.

Regarding Nattier's observation that while the general meaning of the Large Sūtra and the Heart Sūtra are the same but their vocabulary is not (the Large Sūtra employs singular verbal forms while the Heart Sūtra uses plural adjectival forms), Harada points out that such grammatical differences are natural, since the subject in the Large Sutra is śūnyatā (singular feminine), while in the Heart Sutra, the subject is sarva-dharmāḥ (plural masculine). Note that such grammatical differences do not exist in Chinese, so where the Sanskrit differs (na ... utpadyate na nirudhyate / anutpannā aniruddhā), the Chinese text of both the Large Sūtra and the Heart Sūtra simply just say 不生、不滅.
Harada also makes the same observation as Nattier (p. 203) that while the Chinese versions say 不增不減 ("(they) do not increase, (they) do not decrease"), the Sanskrit formulates it in reverse: anūnā aparipūrṇāḥ "they do not decrease, they do not increase." While Nattier says that "it is difficult to explain this reversal no matter what direction of textual transmission is postulated," Harada points out how the Chinese title of the Anūnatva-Apūrṇatva-Nirdeśa shows the same quirk as the phrase in the Chinese Heart Sūtra: 不增不減經 - i.e. putting 'non-increase' (apūrṇatva 不增) before 'non-decrease' (anūnatva 不減). All in all, he proposes that the Heart Sūtra was indeed compiled in India and sees it very likely that its vocabulary was taken from a source/s different from that of the Large Sūtra (he proposes the Tathāgatagarbha sūtras (如来蔵経典)), thereby explaining the different word choices. (pp. 100-99/41-42)
This is a reference to the passage:



Iha Śāriputra sarva-dharmāḥ śūnyatā-lakṣaṇā, anutpannā aniruddhā, amalā avimalā, anūnā aparipūrṇāḥ.
是諸 法空相 不生不滅不垢不淨不增不減。
是諸 法空相 不生不滅不垢不淨不增不減。
ya Śāradvatīputra śūnyatā na sā utpadyate, no nirudyate, na saṃkliśyate, na vyavadāyate, na hīyate, no vardhate.

As Nattier points out, the Chinese text inexplicably shifts the subject from śūnyatā (singular) to sarvadharmāḥ (plural). With this shift the grammatical forms have to change. And, since we have this text in all four versions, we can do our basic check of asking where the similarities and differences are. As we can see, Xīn and Dajing are identical, while Hṛd and Pañc are different in vocabulary and sentence structure and, crucially, the subject in Hṛd is sarvadharmāḥ (following the Chinese rather than the Sanskrit Large Sutra). This gives a clear conclusion. But we can say more. The passage in Pañc is a pericope. It occurs, with minor variations, throughout the Prajñāpāramitā literature and even though the conjugations vary, it always uses words from the same roots, i.e. ut√pad and ni√rudh; saṃ√kliś and vi-ava√dā; √ and √vṛd. Sometimes the pairs appear individually.

Now this is easily refuted. All we have to do is show a Sanskrit text with the same wording as the Heart Sutra. Even one. I've looked for seven years and I cannot find one. Harada has evidently not found one either. But we would not expect to. This is simply not how Indian Buddhists expressed these ideas, whereas the forms found in the extant Pañc and Aṣṭa texts are how they expressed them. The Japanese scholars all seem to adopt the approach of seeking Chinese texts that might explain the situation. But this cannot help us here. We are not trying to explain the similarity in Chinese, we are trying to explain the difference in Sanskrit and only another Sanskrit text will do. Chinese translations are notoriously poor guides to the Sanskrit source texts. As Jan Nattier says in her other famous work:
"In short, when reading any given line of a Chinese Buddhist sūtra—excepting perhaps those produced by someone like Hsüan-tsang, who is justifiably famous for his accuracy—we have a roughly equal chance of encountering an accurate reflection of the underlying Indian original or a catastrophic misunderstanding" (2003: 71).
And as we've just seen, Xuanzang is not 100% reliable. What does Xuanzang have in his translation? He has: 是諸法空相,不生不滅、不染不淨、不增不減 (T 220; 5.22b67), i.e. exactly the same as Kumārajīva. If Xuanzang is famed for his accuracy and all the extant Sanskrit manuscripts have "śūnyatā na sā utpadyate, no nirudyate, na saṃkliśyate, na vyavadāyate, na hīyate, no vardhate" then this is what those characters refer to in T 220 but also, since the two translators use identical expressions, this is what T 223 refers to as well. Xuanzang has approved of Kumārajīva's translation and reused it so it must have matched his Sanskrit source text. And thus the Sanskrit terms in the Hṛd are the odd ones out. The Chinese characters in Xīnjīng cannot be a translation of the Sanskrit terms  in Hṛd because we know that these Chinese characters fit the expected pattern. This is all quite elementary. We have no need to go looking for alternative sources in Chinese.

Harada needlessly proposes the Tathāgatagarbha Sūtras as alternative sources but of course most of the main Tathāgatagarbha Sūtras do not exist in Sanskrit any more. Of those that do, the Ratnagotravibhāga does not contain these terms except for one mention of anutpannā aniruddhā (Rgv 112); the Mahāyāna Mahāparinirvāṇa Sūtra does not contain the terms at all; the Laṅkāvatāra Sūtra has the words anutpanna and niruddha paired (e.g. Laṅk 77) but never the others. So, as far as Sanskrit sources go, Harada's theory falls flat. 

The Anūnatva-Apūrṇatva-Nirdeśa does indeed show the same quirk as the phrase in the Chinese Heart Sūtra: 不增不減經. The problem with this text as a potential source is that it does not include the phrase 不垢不淨. The Nirdeśa does include 不生不滅 but it is clearly being used in the singular. All three pairs would have to occur to consider the Anūnatva-Apūrṇatva-Nirdeśa as a potential source for the Xīnjīng. Ideally, we'd see the three pairs used as a set, but the Nirdeśa doesn't even combine the two pairs it does include. Harada seems to have done a half job here. With a little more effort he would have seen these serious objections to this proposition. 

Note also that the Dajing has the pair 不增不減 in the opposite order to the Sanskrit text. Now, Anūnatva-Apūrṇatva-Nirdeśa was translated by Bodhiruci, who was active around a century after Kumārajīva made his translation. So why is Harada proposing that we need another source later than Kumārajīva? We know that Xīnjīng borrows other phrases from the Large Sutra, so why not this one? The fact that Harada's alternative theory doesn't make sense should not detract from the fact that it is not needed in the first place.

Regarding the differences in expression used between the Large Sūtra (na anya X anya Y "X is not other than Y") and the Heart Sūtra (X na pṛthak Y) "Y is not distinct from X") despite their word-for-word similarity in Chinese (X不異Y), Harada pretty much argues that the similarity in the Chinese version could have been caused by Xuanzang being a conservative translator, citing how in his own translation of the Large Sūtra (T. 0220). Xuanzang retained Kumarajiva's 色不異空、空不異色. Harada adds, if the Sanskrit Heart Sutra was really back-translated from Xuanzang's text, why did the translator not render 空 as śūnyaṃ/śūnyān, but as śūnyatā/śūnyatā(ḥ), which would have been 空性 in Chinese? (pp. 106-105/35-36)

The Indian Prajñāpāramitā literature never uses the form na pṛthak. Never. The Indian Prajñāpāramitā authors always make this kind of comparison using the anya pronoun. And here again the two Sanskrit texts are very different, while the Chinese texts are the same or very similar. Which is exactly what the Chinese origins thesis predicts.

Harada's point about 空 is spurious because 空 is routinely used for sūnyatā in Chinese texts of many genres and across different time periods. What's more śūnyatā is the form the word mostly takes in Sanskrit; the adjectival form śūnyan appears occasionally and it also translated as 空. The alternate translation 空性 is sometimes used, but much less often. For example, in his Large Sutra translation Kumārajīva uses 空  2280 times and 空性 just 34 times. Most of the time 空 means śūnyatā and Harada must be aware of this because it is elementary to the field of Chinese Buddhist studies. 

A more serious objection was raised by Huifeng (2008), i.e that Nattier had ignored the notes in the Taishō which showed that the older editions of the Tripiṭaka have an alternative form of the phrase, i.e. 非色異空. This is what we find in Damingzhoujing (T 250), although we know that this text is a fake. The early Chinese commentarial literature has the phrase as 色不異空: for example, the Zhào lùn 肇論by Kumārajīva’s student Sēngzhào 僧肇 (T 1858; 45.156c5-6); and the Móhēzhǐguān 摩訶止觀 (T 1911), a collection of lectures by Zhìyǐ 智顗 published by his student Guàndǐng 灌頂 in 594 CE (T 1911; 46.5b19-20). This suggests that 色不異空 is the original expression and that the alternative was introduced in the editing of the Canon.

Regarding the following excerpt: (clause Ia/b) rūpān na pṛthak śūnyatā śūnyatāyā na pṛthag rūpam; (clause IIa/b)evam eva vedanā saṃjñā saṃskāro vijñānaṃ; (clause IIIa/b)yad rūpaṃ sā śūnyatā yā śūnyatā tad rūpam, Harada makes the following points:
(1) He refutes Nattier's claim that (Ia/b) rūpaṃ śūnyatā śūnyataiva rūpaṃ "is absent from all the Chinese versions of the text" (p. 203) by citing the translations of Amoghavajra (From a manuscript from Dunhuang: (Ia/b) 色空、色性是空。 (IIa) 色不異空。(IIb) 空亦不異色。(IIIa) 是色彼空、(IIIb) 是空彼色。), Dharmacandra (法月 Fayue, 738 - T. 0252: (Ia/b) 色性是空、空性是色 (IIa) 色不意空。(IIb) 空不意色。(IIIa) 色即是空、(IIIb) 空即是色。) and Prajñācakra (智慧輪, after 855 CE - T. 0254: (Ia/b) 色空、空性見(?)色。(IIa) 色不意空。(IIb) 空不意色。(IIIa) 即色是空、(IIIb) 即空是色。)

Nattier has confused the issue here and phrased her observation poorly. Harada is technically correct here. What Nattier meant to comment on, I'm sure, is that the pairs of phrases rūpam śūnyatā śūnyataiva rūpam and rūpān na pṛthak śūnyatā śūnyatāyā na pṛthag rūpam are inverted in Chinese, i.e. 色 不異空,空不異色;色即是空,空即是色。 And in this case Hṛd follows Dajing rather than Pañc. More evidence against the Indian origins thesis.

Harada manages to show that the phrase occurs in the Heart Sutra versions, because of course it does, but so what? The other versions of the Heart Sutra all dated later than the Xīnjīng. And we know from Watanabe that Damingzhoujing is a fake.
(2) Harada agrees that many Nepalese manuscripts lack (IIIa/b) yad rūpaṃ sā śūnyatā yā śūnyatā tad rūpam. Among the Chinese translations, those of Facheng (法成, 8th c. - T. 0255: (Ia) 色即是空、(Ib) 空即是色。(IIa) 色不異空、(IIb) 空不異色。) and Dānapāla (施護, after 982 CE - T. 0257: (Ia) 即色是空、(Ib) 即空是色。(IIa) 色無異於空。(IIb) 空無異於色。) reflect this omission. Out of the Indian commentators on the Heart Sūtra, only Praśāstrasena has this third sentence; all others omit it.

(3) Versions that include this third sentence (aside from Praśāstrasena's text) are the Hōryūji manuscript and the transliterations by Amoghavajra and Maitrībhadra (慈賢: 10th c.), the Chinese translations mentioned in (1), and a Tibetan manuscript of the short version found in Dunhuang.
(4) The versions of Xuanzang and Prajñā+Li Yan, meanwhile, apparently omit the third sentence and switch the first two sentences around: (IIa) 色不異空、(IIb) 空不異色。(Ia) 色即是空、(Ib) 空即是色。
(5) In both Kumarajiva's version of the Heart Sūtra and the Large Sūtra, however, the sentence structure is completely in reverse:
(-IIIb') 色空故無惱壞相。受空故無受相。想空故無知相。行空故無作相。識空故無覺相。何以故。(IIa) 舍利弗非色異空。(IIb) 非空異色。(Ia) 色即是空。(Ib) 空即是色。

There's no real argument against the Chinese origins thesis here. Details of a textual variation cited without analysis don't help much. And Harada seems to ignore chronology here. If later versions show variation, then so what? This is precisely what we expect for a Mahāyāna sūtra. All that we need to point out is that the line yad rūpaṃ sā śūnyatā yā śūnyatā tad rūpam is absent from both recensions of Pañc and from the Chinese translations of Pañc. It was clearly added in some manuscripts of the Heart Sutra after the initial composition creating a second recension. And again, this is just what we expect for a Mahāyāna sūtra text over time.

One begins to suspect that Harada has not looked at the Sanskrit Pañc at all. Admittedly, the edition by Dutt (1934) was quite flawed and the facsimile edition of the Gilgit manuscript he had access to was difficult to read. We are fortunate now to have much better access to these texts in the editions by Kimura (2010) and Karashima et al. (2016).  The whole point that Nattier makes is that the pervasive differences are in Sanskrit. We are not really trying to explain similarities in Chinese, we are trying to explain differences in Sanskrit. No amount of citing Chinese texts is going to achieve this (unless, of course, the Chinese texts are the source text for the Hṛdaya).

Harada then proposes the following scenario: the original sentence order as found in the 25,000-verse Prajñāpāramitā and in the 'primitive Heart Sutra' (原初的な 『心経』) in which this section was incorporated was (-IIIb')-(IIa/b)-(Ia/b). (cf. Kumarajiva) At some point, (-IIIb') was excised, leaving only (IIa/b)-(Ia/b). (cf. Xuanzang, Prajna and Li Yan) Afterwards, a new sentence (IIIa/b) was inserted and the first two sentences were switched around, giving the structure (Ia/b)-(IIa/b)-(IIIa/b). (cf. Hōryūji and Dunhuang MSS, Amoghavajra, Maitrībhadra, Dharmacandra, Prajñācakra, Praśāstrasena) However, (IIIa/b) was finally dropped, leaving only (Ia/b)-(IIa/b). (cf. Facheng, Dānapāla, Nepalese and Indian MSS.)
Harada thus argues that the shorter Sanskrit version cannot then have been a back-translation from Xuanzang's text as Nattier proposes: if that was the case, one should expect the text to reflect the (IIa/b)-(Ia/b) structure found in Xuanzang, whereas texts such as the Hōryūji MS or Amoghavajra's transliteration have the (Ia/b)-(IIa/b)-(IIIa/b) structure. (pp. 105-103/36-38)

This is just nonsense. There is no need for such a convoluted approach to this textual variation. This argument ignores the obvious conclusions and tries to twist them so that they support an article of faith. yad rūpaṃ sā śūnyatā yā śūnyatā tad rūpam is simply missing from both the recensions of Pañc. It was added directly to the Heart Sutra some time after the composition of the Xīnjīng.

What Harada has done here is compare all the texts except the actual source of the reused passage. What's more, I have shown (Attwood 2017b) that this phrase has a long history and undergoes an important transition when it is copied from the Aṣṭasāhasrikā-prajñāpāramitā-sūtra (Aṣṭa) into the Large Sutra. Here is how this phrase evolved.

rūpam māyopamaṃ
"appearance is like an illusion"
(simile in Pāḷi suttas)
rūpam māyā
"appearance is illusion"
(metaphor in Sanskrit āgamas)
rūpameva māyā mayaiva rūpam
nānyad rupām anyā māyā nānyā māyā anyad rūpaṃ.
"Appearance is just illusion; illusion is just appearance.
Appearance is not different from illusion; illusion is not different from appearance"
rūpameva śūnyatā śūnyataiva rūpam
nānyad rupām anyā śūnyatā nānyā māyā anyad rūpaṃ.
色不異空 空不異色
色即是空 空即是色
(Dajing) [inverted pairs]
非色異空 非空異色
色即是空 空即是色
(Dajing alternate text)
色不異空 空不異色
色即是空 空即是色
rūpam śūnyatā śūnyataiva rūpam.
rūpān na pṛthak śūṇyatā śūnyatāyā na pṛthag rūpam
rūpam śūnyatā śūnyataiva rūpam.
rūpān na pṛthak śūṇyatā śūnyatāyā na pṛthag rūpam
yad rūpaṃ sā śūnyatā yā śūnyatā tad rūpam

Inverting the pairs was a quirk introduced into the text by Kumārajīva in his Large Sutra translation. The key point that Harada misses is that although na pṛthak is valid Sanskrit, it is never used by the Indian authors of Prajñāpāramitā texts. No amount of shuffling the Chinese texts around can erase this vital fact. It was ignorance of the Sanskrit Prajñāpāramitā literature that tripped up the original translator and made him chose the wrong word on many occasions. And it has tripped up Harada as well.

At the part "Because there is no attainment, the bodhisattva(s), relying on perfection of wisdom..." the Horyuji MS, Amoghavajra and Maitrībhadra all have bodhisatvānāṃ (genitive plural). The plural is also reflected in the Tibetan version (byaṅ-chub-sems-dpaḥ rnams), even if the word there is rendered in the nominative as the subject of thob-pa med-paḥi phyir ("because there is no attainment"). Harada asks, since Xuanzang's version reads 菩提薩埵 at this point, shouldn't have one expected the putative creator of the Sanskrit text to render it in the nominative singular (bodhisattvaḥ) instead of the plural bodhisattvānāṃ attested in the different versions? (p. 94/47)
(Harada (pp. 49-50/92-91) answers the question about how to square the plural 'bodhisattvas' with āśritya viharati, which requires a singular nominative by saying that the subject of this portion is actually Avalokiteśvaramentioned in the prologue: in other words, Avalokiteśvara examined the five skandhas, the twelve ayatanas, the eighteen dhatus, the twelve nidanas, and the four noble truths, saw their 'emptiness', and as a result, he dwells (in saṃsāra to save sentient beings?) in a state of nirvāṇa, with an unobstructed mind. Harada declares that Avalokiteśvara must be the one referred to in this portion; otherwise it will appear that he just had a one-time cameo in the beginning and then inexplicably disappeared. "I cannot imagine the author of the Heart Sutra seriously thinking of such an incomprehensible and unsightly scenario." (『心経 』制作者がそんな不可解で無様なドラマのシナリオを本気で考えたとは想像できない。)
His rendering of this portion (p. 119/22) therefore pretty much runs like (note: not a word for word translation of his translation): "Because there is no attainment (of the arhathood held to be the ideal in Śrāvakayāna) for bodhisattvas, he (Avalokiteśvara), relying on the perfection of wisdom, (continues to) dwell (in saṃsāra) with an unobstructed mind. Because his mind is unobstructed, he is unafraid [of saṃsāra], having overcome perverse thoughts/views; (while still being in the saṃsāra world) he is in nirvāṇa. It is (after all) due to relying on the perfection of wisdom that all the buddhas of the three times have attained supreme, perfect awakening."

Here Harada goes completely off the rails as is shown by Huifeng 2014 and my forthcoming article on this section (still in the review process). Harada has failed to notice a couple of things that Huifeng's sharp eyes have noticed.

The first is that where Xinjing has 無智亦無得 which we rightly translate as na jñānam na prāptir, the Sanskrit recensions of Pañc both have instead na prāptir nābhisamayo. I added the observation that both the Mokṣala (T 8.6a11-12) and Xuanzang (T 7.14a23) translations of this passage are consistent with na prāptir nābhisamayo. The text should read na prāptir nābhisamayo. What's more, I observe in my forthcoming article that the list of pairs that follow in all the Pañc texts, including Kumārajīva, are precisely the attainments and realisations. All the lists are based on the classic list of eight āryas. In the Gilgit Pañc ms. we find this list
na prāptir nābhisamayaḥ na srota āpanno na srota āpattiphalaṃ [na sakṛdāgāmī] {21r} [na sakṛdāgāmi]phalaṃ nānāgāmī nānāgāmiphalaṃ nārhan nārhatvaṁ na pratyekabodhir na pratyekabuddhaḥ na tatra mārgākārajñatā na bodhisatvaḥ na tatra bodhir na buddhaḥ. [my transcription of the ms in Karashima et al 2016].
No attainment, no realisation: no stream entry and no fruit of stream entry, no once-returning and no fruit of once-returning; no non-returning and no fruit of non-returning, no arhat and no arhatship, no individual awakening and individually-awakened, no knowledge of the path-maker and no bodhisatva, no awakening and no awakened.

Here prāpti and abhisamaya stand for the more usual marga and phala. Thus Kumārajīva, alone of the three Chinese translators, has made an error here. This very specific error is copied into the Heart Sutra and then translated into Sanskrit. This alone makes the Chinese origins thesis a certainty.


It is now 27 years since Nattier's article was published and it has yet to get the credit it deserves. I find it frustrating that the responses to it fall so far short of the standard that Nattier set, and yet still seem to outshine her exemplary work.

There is always the possibility that Pat457 has misrepresented Harada, in which case this critique can be considered a rebuttal of the Pat457's text. And my apologies to Harada-san. I'd be more than happy to read a full translation of Harada's article by a competent translator at some point, even though the summary by Pat457 makes it seem quite unpromising. But, taking it at face value, we have to question what Harada was doing when he published these comments and what kind of editorial process his work went through. The Sanskrit text he uses is faulty to start with. Does no one actually read the Heart Sutra in Sanskrit? The flaws in his argument seem so obvious that the whole enterprise strains credulity. How do these substandard articles on the Heart Sutra get published?

The Heart Sutra, the most popular of all Buddhist texts, seems to attract only cranks and zealots. Where is the careful, detailed, impartial scholarship? Why are the same old cliches endlessly repeated?

At present the Wikipedia article on the Heart Sutra says that Fukui, Harada, Ishii  et al have refuted the Chinese origins thesis, which is simply and emphatically not true. The traditional interpretation has some antiquity and should be properly documented. But so should the truth. The Heart Sutra appears to be popular only in a very narrow sense of the word: it is chanted and sung as a magical incantation, but it is not read or studied. If you want to tell me that you have read or studied the text then first you must tell me where the mistakes are. If you can't say, then you have not read it. And if you have only studied a translation, then you have been misled, probably deliberately. You've been had. And the sad thing is that the true history and story of the Heart Sutra is better than the cheesy traditional story.

"Ever get the feeling you've been cheated?"
John Lydon. 



Attwood, J.
— 2015. ‘Heart Murmurs: Some Problems with Conze’s Prajñāpāramitāhṛdaya.’ Journal of the Oxford Centre for Buddhist Studies, 8, 28-48.
— 2017a. ‘“Epithets of the Mantra’ in the Prajñāpāramitāhṛdaya.” Journal of the Oxford Centre for Buddhist Studies, 12: 28-59.
— 2017b. ‘Form is (Not) Emptiness: The Enigma at the Heart of the Heart Sutra.’ Journal of the Oxford Centre for Buddhist Studies, 13, 52–80.
— 2018a. A note on Niṣṭhānirvāṇaḥ in the Heart Sutra. Journal of the Oxford Centre for Buddhist Studies, 14, 10-17
— 2018b. ‘The Buddhas of the Three Times and the Chinese Origins of the Heart Sutra.’ Journal of the Oxford Centre for Buddhist Studies. 15, 9-27.
— 2019a (forthcoming). Xuanzang's Relationship to the Heart Sutra in light of the Fangshan Stele. Journal of Chinese Buddhism.
— 2019b (forthcoming). ‘Ungarbling Section VI of the Sanskrit Heart Sutra.’ Journal of the Oxford Centre for Buddhist Studies.

Conze, Edward. (1973). The Perfection of Wisdom in Eight Thousand Lines and its Verse Summary. San Francisco: City Lights, 2006 (reprint).

Dutt, N. (1934). Pañcaviṃśatisāhasrikā Prajñāpāramitā Sūtra: edited with critical notes and introduction. London, Luzac & Co.

Harada, W. (2002) "An Annotated Translation of The Prajñaparamitahrdaya." Association of Esoteric Buddhist Studies, Vol.2002, No.209, pp. L17-L62

— (2008). “A Survey Of Prajñāpāramitā Sūtra Translations In Chinese.” Unpublished Essay. http://prajnacara.blogspot.com/2008/10/survey-of-prajnaparamita-sutra.html.
— (2014). “Apocryphal Treatment for Conze’s Heart Problems: Non-attainment, Apprehension, and Mental Hanging in the Prajñāpāramitā Hṛdaya.” Journal of the Oxford Centre for Buddhist Studies. 6: 72-105.

Karashima, S., et al. (2016) Mahāyāna Texts: Prajñāpāramitā Texts (1). Gilgit Manuscripts in the National Archives of India Facsimile Edition Volume II.1. The National Archives of India and The International Research Institute for Advanced Buddhology, Soka University, Tokyo.

Kimura, Takayasu. (2010). Pañcaviṃśatisāhasrikā Prajñāpāramitā. 5 Vols. Tokyo: Sankibo Busshorin 2007-2010.

Nattier, Jan (1992). “The Heart Sūtra: a Chinese apocryphal text?” Journal of the International Association of Buddhist Studies. 15 (2) 153-223.

06 September 2019

Notes on the History of the Dàmíngzhòujīng

My article on the Fangshan Stele as the oldest dated Heart Sutra text is about to appear in the Journal of Chinese Buddhism. And it got me thinking about one of the other Heart Sutra texts. I've already written about the Sanskrit text transcribed in Chinese characters and accompanied by a modified version of the standard text (T 256). It's now attributed to Amoghavajra, but as with the others we don't really know who composed it. I have yet to write much about the Dàmíngzhòujīng, i.e. the 摩訶般若波羅蜜大明呪經 Móhēbōrěbōluómì-dàmíngzhòu-jīng (T 250). This is the text that is traditionally attributed as a translation by Kumārajīva. So what do we know about this text?

Dàmíngzhòujīng: Provenance

Like the Xīnjīng, the Dàmíngzhòujīng reuses several passages from Kumārajīva's Large Sutra translation but, unlike the Xīnjīng, it has not been modified with terms drawn from Xuánzàng's lexicon. Also in the Xīnjīng version of the so-called "core passage", the copied passage starts a line later and a line has been removed from the middle. Dàmíngzhòujīng restores the line in the middle and starts one line earlier. The resulting text is much closer to Large Sutra translation completed by Kumārajīva in 404 CE than the Xīnjīng is. This led to the idea that the Dàmíngzhòujīng came first and the Xīnjīng is a modified version of it (something that is still part of the popular mythology of the Heart Sutra).

It is now widely believed by scholars that the Dàmíngzhòujīng was created relatively late. Watanabe Shōgo (1991) first drew the conclusion and later said in an (undated) interview:
"The theory that Kumarajiva's Heart Sutra is a spurious scripture [偽経] was suggested and it has become an established theory in the academic world at present." (translation by Jeffrey Kotyk)
The phrase 偽経 (gikyō) is the same one that is used by medieval Chinese bibliographers (i.e. 偽經 wěi jīng) for texts that they did not consider to be authentically Buddhist. I would probably adopt a more brusque tone and translate it as "fake text". Other modern authors have tended to adopt the more euphemistic term "apocryphal" (in various conjugations).

I can't read Japanese, but the arguments that the Dàmíngzhòujīng was not produced by Kumārajīva are not too difficult to tease out. It's not recorded amongst the works Kumārajīva is known to have translated. That said, a number of texts were falsely attributed to him for a time and have since been deprecated. Kumārajīva's translation process was very public. He would translate and comment on the texts during lectures, with the audience numbering in the hundreds. A lost Kumārajīva translation is very unlikely, whereas a spurious attribution is very likely. In addition, he is not known to have produced any digest texts (抄經), i.e. collections of reused passages that are supposed to convey the meaning of a larger text. And the Heart Sutra is certainly in that genre. Nor is the Dàmíngzhòujīng recorded in any of the dozen or so surviving bibliographies of Buddhist texts until the Kaiyuan Catalogue of 730 CE, i.e. the Dà táng kāiyuán shìjiào lù 《大唐開元釋教錄》 (Catalogue of Śākyamuṇi’s Teachings of the Kaiyuan Era of the Great Tang), compiled by Zhìshēng智昇  (T 2154). In order to fit these facts into the traditional myth of the Heart Sutra, some scholars conjectured that the Dàmíngzhòujīng was produced by one of Kumārajīva's students after his death. Perhaps his student Sēngzhào 僧肇 who was one of his principal collaborators (Liebenthal 1968). Sēngzhào's role was to listen to Kumārajīva's explanation of the text and write it down in elegant Chinese. He is one of the people responsible for the enduring appeal of Kumārajīva's translations. 

Everyone who ever wrote about the Xīnjīng (the standard Chinese Heart Sutra) in English tells us at the outset that this is probably the most beloved text in Mahāyāna Buddhism and that it is chanted daily in temples around the world. This popularity has never extended to the Dàmíngzhòujīng or, if it comes to it, to the Sanskrit fake produced in Tang Dynasty China.

I have conjectured that Dàmíngzhòujīng was only created to shore up support for the emerging myth of the Heart Sutra as a classic of Indian Prajñāpāramitā literature. It helps to fill out the back story. But the focus was always on the Xīnjīng (T 251)

I know the English language literature on the Heart Sutra quite well and the only serious discussion of the history of the Dàmíngzhòujīng that I know of occurs in Jan Nattier's 1992 article. And even that is quite sketchy. We know that the Dàmíngzhòujīng first appears in literary record in the Kaiyuan Catalogue in 730 but where are the physical examples of the Dàmíngzhòujīng and what is the oldest one? Why are there no images of the Dàmíngzhòujīng on the internet for example? (Search for "Dàmíngzhòujīng" or "Dàmíngzhòu jīng" and it's mainly my own work, such as it is). After attempting to search for information specific to this text and not finding anything, I began to try randomly asking the question online and sending emails to likely informants. And this led to some insights that I might never have got to on my own.

Ji Yun

Responding to my email, Ji Yun pointed out that the Dàmíngzhòujīng does not occur amongst the 100,000 or so Dunhuang manuscripts. This in itself is quite telling. Most of the texts there are from the 8th Century onwards. There were many Heart Sutra manuscripts at Dunhuang covering a variety of versions of the text in Chinese and Tibetan but the Dàmíngzhòujīng was not one of them. It did not travel beyond China.

Ji also consulted Huìlín’s (慧琳, 736-820)  一切經音義 Yīqièjīng yīnyì "Pronunciation and Meaning of All the Sūtras". This early dictionary of Chinese was begun in 649 by Xuanying 玄應 but completed by Huìlín in 807. Xuanying completed 25 chapters but the final version has 100. Xuanying was a contemporary of Xuanzang and worked with him.

The Yīqièjīng yīnyì lists two versions of the Heart Sutra one of which is attributed to Kumārajīva. However, Huìlín appears to be confused. The two texts he mentions are labelled:
  1. 《大明呪經》(前譯般若心經 慧琳音). Dàmíngzhòujīng (A previous translation of the Prajñāhṛdaya Sūtra. Entry by Huìlín)
  2. 《般若波羅多心經》(羅什譯 慧琳音). Bōrěbōluó[mi]duō xīnjīng (sic). (Translated by Kumārajīva. Entry by Huìlín). 
The tripiṭaka attributes Dàmíngzhòujīng to Kumārajīva and convention treats it as a "previous translation"; whereas the Xīnjīng is attributed to Xuanzang. Clearly, Huìlín is somewhat confused in his attributions. Along with each title are some associated words with guides to pronunciation and definitions. Just three terms are discussed and they don't add much to the picture. Thanks to my old friend Maitiu O'Ceileachair for advice on how to understand the dictionary entries.

Jason Protass

Jason Protass (Brown University), responding via Twitter, was also most helpful. He responded by looking through the older editions of the Chinese Tripiṭaka:
"an extant printed Damingzhoujing appears in the N. Song canon, and the sponsors colophon is dated 1085."
"The Damingzhoujing is in the catalog for the Kaibao canon 開寶藏, the first printed canon completed in 983, but is not among the surviving fascicles." (tweet)
Ji Yun also consulted a Chinese reference work that shows the Damingzhoujing occurs in all of the editions of the Tripiṭaka following the Kaibo Canon.

From the collection of Kunaichō shoryōbu, the Japanese Imperial Household Agency, we get some images from the Chongning Canon 崇寧藏, sometimes also referred to by the place of production, i.e. the Dongchan Temple (東禪寺) edition. Jason says it is dated between 1080 and 1112 CE (tweet). However, the date on this specific text is 元豐八年, i.e. the 8th year of the Yuanfeng era in the Song Dynasty or 1085 CE. The Dàmíngzhòujīng covers two pages, but in the accompanying image I have stitched them together using Photoshop.

Dàmíngzhòujīng as it occurs in the Chongning Canon 崇寧藏 (p. 17-18)
There is also a Stele from Fangshan (another one) from the Liao Dynasty (916–1125) which contains a copy of the Dàmíngzhòujīng along with the Xīnjīng. Several related pieces in the catalog are dated 1081 so it's probably from a similar period. I haven't yet had an opportunity to study the inscription, but it looks like this:

Buddhist Association of China (2000 VII: 399)


These may well be the earliest examples of the Dàmíngzhòujīng in existence. Both from about 300 years after the first literary reference and 400 years after the first physical evidence of the Xīnjīng. Please email me if you know of any earlier evidence (full credit will be given in any future publications on this subject).

Even given how periferal the Dàmíngzhòujīng was and is in Chinese Buddhism, the resources for this text are quite thin on the ground. No one has studied this text. I think maybe there is more information available in Chinese or Japanese sources, but my medieval Buddhist Chinese is not much help in reading modern Chinese and I don't know any Japanese. And here lies one of the problems in this field: information gets trapped in silos and those who could span the divide seem not to have much interest in doing so.

It would be very interesting to try to dig out all the Heart Sutras from all the extant canons and look at the variations!


As a footnote to my essay on svāhā, note that the Yīqièjīng yīnyì (807 CE) lists the spelling in T 250 as 僧婆訶 whereas Taishō has 僧莎呵 with no notes. However Chongning Canon 崇寧藏  (1085 CE) gives svāhā as 莎訶.


Buddhist Association of China and Chinese Buddhist Library. 中国佛教协会 / 中国佛教图书文物馆 (2000). Fangshan shi jing 房山石經 (30 Vols). Beijing: Huaxia chubanshe 華夏出版社.

Liebenthal, Walter (1968). Chao lun; the treatises of Sengzhao. A translation with introduction, notes, and appendices (2nd Rev. Ed). Hong Kong: Hong Kong University Press.

Watanabe, Shōgo. (1991) 「般若心経成立論序説」 『仏教学』 “An introduction to the Theory on the Formation of the Prajñā-hridaya-sūtra,” Journal of Buddhist Studies 31, (July): 41-86. [Japanese].

23 August 2019

Bad Free Will Philosophy

Philosophy is an important activity. Ideally, philosophy helps us to make sense of the world, ourselves, and our place in the world. Unfortunately, philosophy, at least on the level that I engage with it, is plagued with unhelpful legacy concepts from the Victorian period. Victorian accounts of subjects like reason, consciousness, and free will are all anachronistic and contradicted by a weight of evidence. It's not clear that these terms offer any advantage as starting places for discussions about the world, ourselves, or our place in the world. Also, virtually all philosophy seems to be solipsistic. whereas we human beings are social animals and we make sense of our world in a social setting. 

Free will is one of the most aggravating subjects to be interested in because the whole discussion is poorly framed: bad definitions, bad methods, and bad theoretical frameworks. Of course, the three coincide in some cases to make for spectacularly bad philosophy, but it only takes one to spoil the whole enterprise. In this essay, I'll walk through what seem to me to be the most egregious aspects of bad free will philosophy.

Bad Definitions

Almost no one starts off a discussion of free will by defining what they mean by free will. And, don't laugh, I'm not going to, either (well, maybe a bit, later on). It is seldom clear what any commentator means by free will, what kind of evidence they think is relevant to the discussion, or what they would consider a valid source of knowledge on the subject. And it gets worse, because not only is free will not defined, but neither are "free" or "will". Much of the time it's not even clear why we need to talk about free will.

Of course, one may sometimes infer from what someone says what assumptions they are making after a while, but this is a very inefficient way to communicate. Worst of all, after one has suffered through enough nonsense to collect sufficient information to triangulate what they actually mean, they usually mean some form of contra-causal free will.

Contra-causal free will is the view (almost universal amongst physicists) that our decisions are not caused by anything. And by "anything" we include everything physical, visceral, and social. So for example, if our own emotions are involved in decision making, then we have no free will. Also, it doesn't count if the decision we make is unconscious. Our state of knowledge tells us that our emotions are involved in every decision, every choice, and every evaluative thought we have because we encode the value or salience of information as feelings. And it seems very likely that all decisions rely on unconscious inferential processes. Ergo, physicists argue that we don't have free will, meaning, we don't have contra-causal free will. But so what? Contra causal free will is a nonsense idea to start with.

Coming back to the problem of bad definitions, the people who are talking about contra-causal free will almost never use the words "contra-causal"; they may not ever have heard the words "contra-causal" (I hadn't, until recently). So, while they appear to be talking about the same thing as other people, they are not, and they probably don't really know what they are talking about, and don't know that they don't know.

A major problem with all these kinds of discussion is that people conclude what they believed at the outset. Deduction from axioms only reproduces the axioms in the end. Assume that we have free will and you can deduce that we must have it because at some point we will judge a proposition to be true on the basis of our belief in the axiom. Assume that we don't have free will and an equally valid line of reasoning will deduce that we cannot have it. This is a built-in flaw of deductive reasoning. We ought to know better by now, but one of the basic assumptions about free will debates is that we don't need to examine our starting assumptions before giving our opinion.

And the reason is obvious. By the time most people have given an accurate account of what they axiomatically believe about free will, it's apparent that they are not interested in having a discussion about it, they merely wish to assert a more or less elaborate belief system. Either that or, by spelling out their assumptions, they realise how stupid the subject is and give up before attempting to communicate it. Most of what makes it into the public domain is ipso facto stupid.

The most egregious examples of this are the ones that grant that I feel myself making decisions, but assert that because the equations that govern the movement of atoms are deterministic, that my decisions are an illusion. In other words, yes, decisions do get made, but we cannot think of them as decisions because that contradicts the model (in which the axiom is "we don't make decisions").

Moreover, the mythical "rational faculty" that is supposed to be the deciding faculty for free will really doesn't exist. This is explained in Mercier and Sperber's book The Enigma of Reason, which looks at the data on how people use reason and shows that we don't. At least, we don't use it for solving problems. 90% of people fail at simple tests of logic, though 80% of us state that we are 100% confident about our answer. All of us do better at solving problems in small groups. What we call "reason" is, in fact, used to propose reasons for things that have already happened. We make decisions using unconscious inference, then, when we need to know why, then reasoning kicks in and produces a reason.

It would be helpful is everyone could spend some time identifying what they believe and why they believe it before contributing to a discussion. 

Bad Methods

Almost everyone who still argues against free will relies at some point on the opinion of Benjamin Libet, which has been proven wrong by his peers. I comprehensively debunked Libet in a blog post called Free Will is Back on the Menu, so I don't really want to go over this ground again. Really, I suppose all I did was repeat the many ways in which other people, Libet's colleagues, debunked his opinion about his results. Libet wasn't exactly a fraud, he just misinterpreted the data based on a faulty model. The intellectual frauds are all the people (mainly physicists) over whom Libet exercised a powerful confirmation bias and who have been uncritically repeating his opinion ever since, without ever looking at the literature within which it is embedded.

Included in the data on human decision making we ought to include all the tests like the Wasson Selection Test that show we don't use reasoning to solve puzzles.

And again, if someone sets out to study decision making, but they take as axiomatic that there is no contra-causal free will, then they are much more likely to design experiments to show this. And again, so what? Contra-causal free will is not a useful way of thinking about human experience. 

Bad Theories

Almost everyone I've come across who denies free will does so either on the basis of a metaphysical commitment to reductionism or a metaphysical commitment to absolute being. So let's look at these.

Metaphysical Reductionism. 

Metaphysical reductionists believe that only the finest possible layers of the universe are real. The search for the nature of reality is the application of a conceptual microtome, slicing the universe so thinly that it cannot be sliced any thinner: atomic means "uncut, indivisible." Obviously, the atom is very cuttable, but we're stuck with calling atoms "atoms" even though our search for the truly atomic continues. This connection with the thinnest layer is why some people link quantum physics and reality.

What's more, they assert that the properties of the atomic entities that exist on that smallest scale are the defining properties of the whole universe. Thus, because they believe it is accurate to describe the universe on the smallest scale as deterministic, then everything, on every scale, is deterministic.

However, there are huge problems with this view. As Sean Carroll will explore in his new book and has talked about in several recent podcasts and various blog posts, we don't know what the world is like on that scale. Of course, we know how to manipulate the equations to predict what kinds of effects we can expect to manifest at a macro-level, but we have no idea what this connotes in terms of physical reality. How does the quantum Hilbert Space relate to reality? No one knows. We don't know what is real at this level and this is the level at which reductionists decide what is real. So... at present we know nothing about reality on those terms.

A majority of physicists have come out against the Copenhagen interpretation of the measurement problem, which in simple terms is the idea that the universe behaves one way when our back is turned and another way when we look at it, which is trickier than it sounds in a system where everything interacts with everything else. But they cannot agree on what does happen. Are there hidden variables that determine how the universe unfolds? Or does each quantum event cause the universe to split into different versions? Are their quantum pilot waves that push the particles around? No one knows. And at present, no one is sure whether we can know. There may be an epistemic horizon beyond which reality exists but we cannot know it or say anything about it. But right now, there is an epistemic horizon and we don't know what lies beyond it or if we ever will. 

Part of the epistemic problem is that we may be able to solve the quantum equations for a single hydrogen atom, but we cannot do so for a deuterium atom, not even in principle. Three particles in a  quantum system make it impossible to provide a precise mathematical description. We have to introduce some pretty gross simplifying assumptions. These assumptions give answers that are pleasingly accurate and precise. When we're already unclear about what the unsimplified equations tell us about reality, how does adding a series of increasingly gross assumptions help get us in touch with reality? Adding simplifications to make the math work takes us further away from reality (if we take the reductionist view). Why is anyone in quantum physics talking about reality

Here's the thing. Metaphysical reductionism is just a bad theory. It ignores the role that structure plays in the universe. It's all very well saying that water is really one atom of oxygen and two of hydrogen; but if you have a litre of water and atomise it, you now have no water. You cannot slake your thirst by drinking liquid oxygen (-219 °C) or liquid hydrogen (-259 °C) or any mixture of the two. If water is not real, over and above the existence of its component parts, then the whole category of "real" is nonsense.


The idea of absolute being manifests in many different ways and has a very long and varied history. It was very popular and became highly developed and differentiated in India. And it is still very popular in Advaita Vedanta circles. 

In this idea we are all just manifestations of a larger entity which is characterised by absolute being: it transcends notions of time and space and causation (and all that other metaphysical stuff). Our individuality is an illusion, our "being" as separate from universal being is an illusion, and especially our sense of having free will is an illusion. Absolute being demands strict determinism. So the irony is that even if human beings have free will, God is wholly deterministic. 

However, there is no need to take seriously any theory of absolute being. They are all figments of our imagination. It is not a theory that rests on evidence or makes any testable predictions. Indeed, the  very idea that a spatio-temporal being can experience the Absolute is nonsensical. This is why religieux have to keep making up ad hoc supernatural entities (like a soul or ātman) that are a little bit of the absolute in us; allowing us to bridge the unbridgeable gap between absolute and temporal. Nonsense compounded by more nonsense.

So much for the arguments against free will. However, rather than argue for some version of free will, I want to try to outline the kind of philosophical discussion I find useful. 

Is There A Way Out?

Back in 2016, I wrote a long three-part essay on reality called A Layered Approach to Reality. I was influenced mainly by Richard H. Jones and John Searle. But also by other philosophers and scientists. My small contribution has been a new way to think about the ancient philosophical problem of the Ship of Theseus. In the Layered Approach essay, I argued that reductionism is fine for discovering knowledge about substances, i.e., what the universe and things are made of. And I argued that a universe in which there is one kind of stuff is the only one that is consistent with all the observations and other theories of science. But this is less than half the story of the universe. 

The basic stuff is made into a lot of other stuff, i.e., structures that persist over time, that are insensitive to swapping out identical parts, and which act as causal agents in ways their component parts alone cannot (like water dissolving salt). In other words, structures are real by any useful definition of that term. In Feb 2019, Sean Carroll recently interviewed James Ladyman on the subject of Reality, Metaphysics, and Complexity. Ladyman's philosophy is similar to what I've proposed in that he argues we have to treat persistent structure as real, but there are some differences between us as well. Listening to him wrestle with the status of numbers I wanted to shout, "Read John Searle's The Construction of Social Reality!" Anyway, I just wanted to point out that I'm not the only one. Incidentally, note that John Worrall's (1989) argument for structural realism is a different kettle of fish.

Any biologist will tell you that dissection can only reveal so much about an organism. You could sequence the entire genome, all the epigenetic info, and map all the genes to proteins and you'd still know nothing about how an organism behaves. You have to observe the living organism interacting with its environment as a system in order to appreciate that organism. Analysis and dissection are the methods of reductionism. And again, these are great for studying substances. It's just that if the object we wish to study is a structure, then reductionism is useless because the moment we dismantle a structure to find what it is made of, we cease to have a structure. 

So, we combine reductionism for understanding substance and anti-reductionism for understanding structures. Anti-reductionism is also sometimes called emergentism. George Henry Lewes (1817-1878) first referred to "emergent properties" of structures in 1875. An emergent property is a property of a complex object that is not possessed by any of its component parts alone or in simple combinations (lacking structure). Generally speaking, emergent properties are not predictable from the properties of the components.

John Searle's analysis of kinds of facts can help us understand how this relates to our daily life.

Kinds of Facts

My Searle-y explanation of the ancient problem of the Ship of Theseus illustrates the principle. Timber has certain intrinsic properties that are ontologically objective: they are real and don't depend on an observer, or we could say that they are true for all observers. Intrinsic properties don't allow a pile of timber to transport a hero across a sea. The timber has to be assembled in a particular way to create a range of new properties. The hull of a ship encloses a volume that has a net density that is much less than the density of the timber and less than the density of water. So a structure floats even if the building material does not. Thus we can build ships from steel which is 8 times as dense as water. Low density is an (emergent) property of the structure that its component parts do not possess. Similarly for the shape that makes a ship move easily through the water, and which resists sideways movement, and so on.

Such a structure is then fit for transporting Theseus across the sea. Functions are observer-relative, and require prior knowledge. A naive dweller may look at a ship and conclude, rather, that it is a cistern for keeping water in. However, for a knowledgeable observer, the fact that a ship is a ship, is epistemically objective. It is true for everyone who knows what a ship it.

Most discussions of this ancient problem centre on "identity" which is, at best, an ontologically subjective fact. I would argue that since identity is only apparent with prior knowledge, that identity is likely to be epistemically subjective as well. The question "Is it the same ship?" has to be followed by the question "To which observer?". All the accounts I have read of the problem assume an unchanging observer with prior knowledge, which is nonsense, and why the problem presents as a paradox.

To come back to the relevant point, the timber has intrinsic properties that make it suitable for shipbuilding. But the ship qua structure also has unique intrinsic properties that are limited by, but not determined by, the properties of the components: the density of the building material does not determine the density of the ship's hull. Structures, in other words, are every bit as real as components.

Structures are real

In my essay about layered reality, I accepted the pragmatic premise that structures are real. But I also pointed out that emergent properties accumulate with complexity. Something as fiendishly complex as a biological cell has many layers of properties that cannot possibly be predicated of mixtures of its individual atoms. There are 1000s of relatively simple chemical compounds as well as 10s of thousands of complex polymers such as peptides, proteins, and nucleic acids. 

As I say, we don't really know what subatomic reality looks like. But the atomic theory of matter is a very successful theory in that it explains a great deal and makes nice and highly accurate predictions. Matter at the atomic scale (just beyond the quantum indeterminacy) is deterministic. The laws that govern matter give (relatively) simple answers: the way the universe evolves on that scale is described by relatively simple equations and if we know the state at any given time, we can use the equations to determine its state at any arbitrary time. 

But this very soon breaks down. As with quantum systems, macro systems quickly become too complex to calculate. If we consider the problem is one of calculability, that is, strictly speaking, an epistemic problem, and we call this view weak emergentism. In this view, the entire universe is still deterministic even if we cannot understand it well enough to predict it. Reductionists who dabble in emergentism (like Sean Carroll) tend to favour this kind of emergentism.

However, if emergent properties are real, if they result in more than just increasing complexity and actually produce wholly new properties, then we have a new ontology at each new level and this is strong emergentism. Reductionists argue for a single, fundamental, ontology combined with some necessary approximations to cope with complexity. Metaphysical antireductionists argue that only the universe considered as a whole, with everything affecting everything else all the time is real (this position is rare). I take a middle path: reductionism for substance, and antireductionism for structures. 

One complicating factor is that in non-linear systems (typically where a large number of components are interacting) predictability may fall to zero. And this happens quickly. A simple pendulum is entirely predictable. But add another degree of freedom halfway along, a pendulum hanging from the end of a pendulum, then the result is apparently chaotic and certainly unpredictable. But this does not make it non-deterministic. The system is still evolving according to patterns (which we call laws when we can codify them), it's just that the system is highly sensitive to changes in the initial conditions. The pattern of a double pendulum is too complex to be computable with any usefulness. The question is whether at some point the unpredictability becomes non-deterministic, i.e. not simply that we cannot determine the pattern from observation, but that the evolution of the system is not governed by simple laws at all. No one would argue that living cells do not change in ways that have patterns, but do such patterns as exist constitute determinism?

The difference between a mass of unstructured matter and, say, a living cell, is vast. So vast that it opens the door to strong emergentism. And if matter organised into biological cells is not deterministic, then how much less so an organism composed of trillions of such cells, themselves structured into organelles, organs, and systems, all in multiple feedback loops. And as we now learn, all in meaningful relationships with our symbiotic microorganisms on the skin and in the gut.

Cutting Loose the Legacy of God

One might ask why we debate free will at all. It is, after all, a theological concept designed to make God seem to be less of a monster for having invented evil and suffering. We're under no obligation to the legacy philosophy and theology of the past. Indeed, the question of whether we have free will is not really the best place to start a discussion about morality. It doesn't even come into my long essay on the evolution of morality for example. What kinds of questions might we really interested in? 
Mainly, as far as I can see, we're nowadays interested in the issue of culpability. It is through this issue that discussion of will has become naturalised in the secular world, i.e., in the absence of god what is the basis for our continuing with the idea that good people deserve to be rewarded and bad people deserve to be punished. In a sense, this is now the issue. Not God's big evil, but our petty human evil. But culpability admits to degrees. I discussed this to some extent in an essay in 2015 called Why Killing is Wrong and I'm actually working on a more nuanced version of this in an essay provisionally entitled Objective Morality (chosen to be provocative). I also touch on relevant issues in my recent essay We Need to Talk About Utilitarianism, which criticises the assumptions that utilitarians make and the way they address moral questions.

If I kill someone, the question is not "am I culpable", but "to what extent am I culpable?" My role in society may involve killing or allow it in certain circumstances (soldiers, police, doctors). As a citizen, I am allowed to defend myself, my loved ones, and my property and lethal force may sometimes be justified. And so on. There are many nuances. 

We know that decisions and choices are influenced by many factors, not least of which is our social environment. It's now many decades since social psychologists pointed out that assuming a person's behaviour is 100% because of their internal motivations is a fallacy (the fundamental attribution fallacy). We are social animals, and much of our behaviour is influenced by what our group expects from us, or at least how we perceive their expectations. We have mutual obligations, sometimes these take the form of rights and duties. We're also subject to "priming", by which I mean if we're having a bad day, for whatever reason, we make different decisions than if we're having a good day. It may even be that what we encounter in the moments before making a decision unconsciously influences the outcome.

Societies do best when there is political stability and citizens are prosperous. Too much stability and a society will stagnate, cease to innovate, and when the time comes they will fail to respond to changes in the environment.  Too little stability and the society will become chaotic and fall apart from the inside out. So we consider everyone to be under mutual obligations. And in large societies, we formalise rights and duties in law codes, the oldest examples of which are almost as old as civilization itself. No human being ever had absolute free will because we live, we exist, in a social network with mutual obligations. Any philosophy that ignores this aspect of humanity is worthless. 


Discussing free will in a reductionist framework is filled with traps. For example, reductionists conclude that anything which is dependent on something else is not real, because it can be reduced to its components. And we've seen how badly physicists go wrong already: If Amy has six apples and Sheldon reduces them to a quark-gluon plasma in a super-apple collider and captures the plasma in a specially designed container that prevents any loss of matter or energy, how many apples does Amy have? None. Reductionists literally cannot see the forest for the trees. Or they cannot see the universe for the quantum fields.

One of the most common reductionist tropes is that human experiences are "just an illusion". It doesn't matter that you have a persistent sense of self, a lasting personality, are able to remember your life, and experience love. In a reductionist framework, it makes sense to say that free will is an illusion, because making decisions is a mental activity, and because everything that is involved in the decision-making process is complex and dependent on component parts. 

If we take an anti-reductionist approach to structure, the fact that an object or entity is complex and made of parts is not important as long as the structure persists over time. Of course, some reductionists also say that time is an illusion. Certainly, the way we measure time is somewhat arbitrary - we simply count the number of iterative processes or events that occur over the period of observation. Time measurements are arbitrary in this sense, but this does not mean that time is an illusion, far from it. Time is a way of talking about the patterns of change that we perceive in the universe around us. Because we can retain information about previous states and compare them to the present, we can perceive change. Change is ubiquitous and unidirectional with respect to the second law of thermodynamics. This gives us the so-called "arrow of time", by which we mean that far in the past the universe was in a low entropy state and the total entropy has been steadily increasing ever since. So time is also real. It doesn't matter that time is not absolute, because nowhere in my definition of real is there any reference to absolutes. Indeed, I'm inclined to argue against absolutes on principle. For example, we know that relativity is wrong at the beginning of time (the big bang) because it predicts a universe of infinite density. That kind of absolute tells us we've made an error, no matter how good the equations are in less curved spacetime. Even if someone manages to prove beyond reasonable doubt that time is an emergent property of quantum fields (and it already seems likely that space is such an emergent property) it won't make time an illusion. 

The problem here is that illusions are not causal. An illusion doesn't make a difference in the world because it cannot interact with the world. Thus, to say that free will is an illusion is to say that humans make no difference in the universe. This is not merely dismal fatalism, it's self-defeating. If humans make no difference, then it makes no difference what we believe and there is no reason to believe that we don't have free will. It is equally valid (at least) to believe that we do have free will. As a philosophy, it ought to lead to passivity, but it doesn't. People who don't believe in free will go on being active and making decisions; they just tell themselves a story about the experience of deciding that makes sense in a legacy/reductive framework, but doesn't in a more sensible framework. 

The same arguments occur for having a sense of self. Of course, self is not an entity; of course, it is generated by the brain, but to argue that our sense of self is not causal, that it makes no difference, is clearly ridiculous. Else why would so many people want to persuade us to stop believing in it? 

The (ill)logic of the Free Will Illusion

The argument is that free will is an illusion, i.e., that there is no free will, and that our apparent free will is not causal, i.e., it makes no difference in the world. But if it is not causal, why is it a problem? The answer is usually that our belief in free will (or self or whatever the "illusion" is) is problematic in some way (usually it makes us unhappy). So free will is an illusion but, being a potential causal agent, a belief is not an illusion. Indeed, in this argument, a belief is real and has causal potential. Beliefs make a difference in the world or they would not be a problem. 

We often see that the same metaphysical reductionists who get so exercised about free will being an illusion seem to become apoplectic about people who hold religious beliefs or even those people who continue to believe in free will. But if free will is an illusion and the world is deterministic why does it matter what anyone believes? Indeed, if there is no free will then no one has a choice about what they believe and trying to persuade them to change their mind is a wild contradiction in terms. If there is no free will then no one ever changes their minds because that would require us to be free to do so.

The reductionist argument about free will being an illusion is not followed through to its logical conclusion by any of its proponents (that I know of). There is clearly a glaring contradiction in asserting, on the one hand, that "free will" (whatever we mean by it) is an illusion and, on the other, asserting that beliefs are persistent in time and causal (i.e., real). Because believing, willing, and selfing are all of the same kind; they are all forms of mental activity (and this epistemically and ontologically subjective). If a belief is causal, then so is our will. Or if will is not causal, then neither are beliefs. You can't have it both ways. 

It does matter what we believe and it matters what we do, if only to the people around us. Because of the latter, the reasons we discern behind our own actions also matter. Will, belief, and behaviour have to be seen in a social context. We need to be able to produce accounts of our behaviour (i.e., reasons) that make sense to those around us, more especially when our behaviour contravenes group norms. Morality evolved in, and only makes sense in, a social context. The broad parameters are limited by our biology, but our flexibility as a species allows for huge variety in mores and customs (and interpretative frameworks).


Related Posts with Thumbnails