Matthew Dasti, in a comment to a post, kindly informed me of his article on Testimony, Belief Transfer, and Causal Irrelevance: Reflections from India's Nyāya School (can be found here on Jstor and also read online without charge, even for people who are not part of a research institution). The article is quite interesting from several points of view, among which I would like to stress the following ones:
I wonder, however, whether a Sanskrit philosopher would not answer that one infers the hour from a clock because of the position of the hour hand and the minute one. Thus, this would be a case of inference (as in the case of gestures), not of Linguistic Communication. What do you think?
For further posts on Linguistic Communication, see here. For further comments on the listener, see this post.
- 1. highlighting the disagreement between Mīmāṃsakas and Naiyāyikas on the interpretation of āptopadeśaḥ (the Nyāya definition of testimony as found in the Nyāyasūtra). It means "statement of a reliable [person]" according to Naiyāyikas, who want to stress the importance of the source and "reliable statement" for the Mīmāṃsakas, who want to stress the independence of language as an instrument of knowledge.
- 2. the shift between ancient Nyāya and Navya Nyāya, since Gaṅgeśa acknowledges the possibility of testimony being successful even in the case of an unreliable speaker, if only she is transmitting a true belief (a typical example might be a liar who believes that B is the case and, thus, says A, but does not know that in fact A is the case) and the listener understands it (see Sudipta Munsi's article here (scroll to the button of the page) for further notes on the competence of the listener).
- 3. the similarity between this position and J. Lackey's one.
- 4. Dasti further includes among the possible witnesses also clocks, thermometers, and similar instruments which are not in themselves reliable (since they do not want to be sincere, nor do they know what they convey).
I wonder, however, whether a Sanskrit philosopher would not answer that one infers the hour from a clock because of the position of the hour hand and the minute one. Thus, this would be a case of inference (as in the case of gestures), not of Linguistic Communication. What do you think?
For further posts on Linguistic Communication, see here. For further comments on the listener, see this post.
15 comments:
Regarding the 2nd point, I would like to mention that the example comes close to the one given by Śrīharṣa in the Śabdalakṣaṇakhaṇḍana section of his Khaṇḍanakhaṇḍakhādyam. It is as follows:
“atha nirdoṣasya vākyaṁ tatheti cen na. sadoṣasya ‘nāsti ghaṭa’ ityabhitsato ‘'sti ghaṭa’ iti daivān nirgatayathārthavākyāvyāpteḥ. tatpramāṇaṁ na bhavaty eveti cen na. pūrvamuktottaratvāt. pravṛttisāmarthyena prāmāṇyāvadhāraṇasambhavād āpātataḥ sandehe 'py adoṣāt. sāmānyato nirdoṣatvasya bhīmāgraje 'py abhāvāt, viśeṣatas tathātvasya asādhāraṇyaparyavasāyitvāt”–
“[Objection:] (If it is said that) the sentence spoken by a flawless person is that (verbal testimony)? [Reply:] Not so. For such a definition would not extend to the case of a person, endowed with flaw, who intends to say that ‘there is no jar’ [on the ground], but accidentally pronouncing a valid sentence in the form of ‘there is a jar’. [Objection:] (If it is said) that it is not valid at all? [Reply:] Not so, for the reason which has been said before, (it entails as its object something that is not already known through any other means). Despite being open to doubt (since it is pronounced by a person, endowed with flaws), it (the statement) is not flawed in as much as it leads to a comprehension of validity through successful activity. In general, flawlessness is impossible with regard to even Yudhiṣṭhira, especially since such a thing (flawlessness) would, in any specific case, culminate into extraordinariness.” (My translation)
Two quick comments: I would still say that clocks, etc. are reliable, since they reliably produce accurate statements. It's just that their reliability is not governed by knowledge or sincerity.
Also: I am not necessarily of the opinion that Naiyayikas would say that clocks give us testimony, but that is where I try to creatively work with the philosophical picture in view and develop it. I think clocks give us testimony!
@Sudipta, thank you for the comment. As already mentioned, Gaṅgeśa's sources tend to be forgotten since he is better known and better studied.
@Matthew, I appreciate any effort to think creatively along Indian texts. But, let us go back to the definition of śabdapramāṇa/testimony. I feat that it is much easier to include thermometers (and clocks) within instances of anumāna, on the basis of the invariable concomitance between a certain position of the hour hand and the time of the day. Don't you think so?
As I note in the paper, my worry about that strategy is that it plays straightforwardly into the Vaisheshika reductionist strategy.
Think about it this way: If you had a clock that was programmed to emit correct utterances about the time in English on the hour, would it be testimony or inference? If you say that it's testimony, I would suggest that there isn't anything importantly different between that and our reading the face of a clock to learn the time. In either case, we are depending on our own syntactical and lexical competence to gain knowledge from "words", so to speak, which accurately convey information.
If what I describe above is an inference, then so is the rest of putatively testimonial knowledge.
And as Nyaya (and I personally) hold to a testimonial anti-reductionism, I would say that learning from calculators, clocks, and (for some) the impersonal Veda(!) is knowledge from testimony, not inference.
This is a case where the Sanskrit term is better than the current philosophical term of choice. "Testimony" is so tied to the notion of a testifier (even "attestation" is as well) that it prejudges the case toward the need for a personal speaker. But learning "from words" much more nicely captures the fact that we can learn from both personal (honest and knowledgeable) and non-personal (because of being non-sentient) conveyors of information.
Thank you, Matthew. I completeley agree with the inadequacy of the Western term, śabdapramāṇa (which I translate with "Linguistic Communication") is much more precise.
As for the Vaiśeṣika reductionism, it is not the case that I *want* to avoid it. Rather, I think that it does not work, because I cannot reduce śabdapramāṇa to anumāna (you will surely know John Taber's illuminating essay about it). By contrast, in the case of a speaking clock, reduction seems possible, in the form of "whenever the clock says 'It's ten o'clock' it means that the time is 10 am". This is not the case for human testimony, since one has to take into account their āptatva. There is, in other words, no mechanical correspondence between words and arthas in human testimony. Nor can you claim that there is such a correspondence in the case of the Veda, since the Veda needs to be interpreted and there is no mechanical correspondence between śabda and artha.
Suppose I had a parrot in a smal cage. If you were to pass by it, you could infer from its singing that it is alive. Now, suppose that I had taught him to say "jīvāmi". Would you know that it is alive because it told you so? Or would you not just infer it out of the fact that you have heard its voice?
Dear Elisa,
Here is a passage from the Citsukhī, which I think supports your position –
शाब्दप्रमाणमपि दुर्निवारम्। तथाहि शास्त्रं शब्दविज्ञानादसन्निकृष्टेऽर्थे विज्ञानम् इति मीमांसकाः। शब्दविषयं विज्ञानं शब्दविज्ञानं तस्मादसन्निकृष्टेऽर्थे श्रोतुरनधिगतेऽबाधिते च विषये यद्विज्ञानं तच्छाब्दप्रमाणमिति यावत्। उक्तं हि वार्तिककारैः –
असन्निष्टवाचा च द्वयमत्र जिहासितम्।
ताद्रूप्येण गृहीतत्वं तद्विपर्ययतोऽपि वा॥ (श्लो. वा. अनु. ५५, ५६)
यद्वा शब्दाज्जनितं पदार्थविषयं यद्विज्ञानं तस्मादसन्निकृष्टेऽर्थे बुद्धिः शाब्दमिति। तदयुक्तम् शब्दलिङ्गकवक्तृविशेषानुमानेऽतिव्याप्तेः। भवति हि वहिष्ठस्याध्ययनध्वनिश्रवणाद्भवनान्तर्वर्तिपुरुषविशेषानुमानम्। द्वितीयेऽपि धूमादिपदात् तदर्थविज्ञाने धूमध्वजादौ जायमानानुमानादावतिव्याप्तिः।
Linguistic Communication (or Testimony) can be conceived as a cognition originated from words with regard to a referent, which is remote in being neither seized in its own form nor understood otherwise'. Or as a cognition about a remote referent arising from the knowledge of things generated by words. It is not so. Because in the first case it would over-extend to the case of an inference made of the speaker from the (inferential) mark of word. For, a person, standing outside a building, hears many people reading loudly, and infers (the existence of) someone among them from the sound heard. In the second case, it would over-extend to the case of the inference of fire made from the knowledge of smoke arising from the word, ‘smoke’.
Short response:
First a small point: Regarding what we "want", the point wasn't that you would want to protect a non-philosophical attachment to testimony but rather, that you (as do I) have philosophical commitments that would make the reductionist strategy unattractive.
Second, I think we may be caught in a holding pattern here . . .
I would say that you know it's 10 o'clock precisely because of clock's "testimony". just as children learn from parents' testimony just by trusting, so too to we learn from clocks. It's not that we learn of a vyApti by some independent means. Perhaps a better case, though, is a calculator. We learn from calculators, even in cases where we have no idea of a vyApti, so to speak, between certain inputs and outputs. With a powerful enough calculator, indeed, you can enter inputs which no one has probably ever even entertained and lean output truths which no one has ever entertained either.
In the case of the parrot which utters "jIvAmi", I would say that such is probably inference, from an animal's ability to make noises to its being alive. But this is disanalogous to the calculator or clock, where there is a systematic and reliable connection between the content of its variety of outputs and states of affairs which obtain.
Not sure exactly what you are getting at with the point about "mechanical outputs". My sense is that the Aptatva of a speaker is instrumental to her being reliable. The crucial issue is the reliability. The analogue to Aptatva for nonconscious things would be (e.g.) if a calculator is properly functioning with the right sorts of algorithms in place, etc., etc.
Hi Matthew and thanks for the rejoinder.
The point seems to me that in the case of a calculator (a clock, a thermometer and so on) *I* am unable to drive a correct inference, but someone more conversant with bytes may be. In other words, there is an invariable concomitance between what you insert and the output you give. This is not the case for Linguistic Communication, where the relation between what one has perceived and what one sees is not fixed. To name just the simplest example, if an āpta has seen a cat lying on a mat she might say "I saw a cat on a mat" or "I saw a mat under a cat" and both would be instances of Linguistic Communication (together with infinite other possibilities). This seems to me the distinctive element (and not only the fact of occurring through words) between inference and Linguistic Communication. It seems to me that you can still build a correct inference between the fact of being now 10 am and the sounds repeated by the "speaking" clock.
The case of the parrot is not thought as analogous to that of the calculator or clock since in the latter “there is a systematic and reliable connection between the content of its variety of outputs and states of affairs which obtain.” It seems that it is assumed, such a “systematic and reliable connection” is “intrinsically” present in clocks and calculators. But is it not rather the case that clocks and calculators do not have such a “connection” ‘intrinsically’, but it is “engineered” into them from outside? Given this, suppose the time is 10pm, but for achieving some unscrupulous end, I make my clock slow by 30 minutes and let others have the time reading 9:30 pm. Under such circumstances, shall my clock, which is now participating in my unscrupulous enterprise, be as “reliable” as before? Again, if my clock on comparison with all other clocks that give that correct reading of 10pm is found to be manipulated, how shall we account for the fact that this clock is unreliable while others are unreliable? So will not the concept of ‘reliability’ of the ‘clock’ here suffer from under-extension (avyāpti)? In the same way, if I teach my parrot to utter “na jīvāmi” instead of “jīvāmi” will it “reliably” convey its state of death? Thus, does it not emerge that the notion of “reliability” is merely a matter of mutual agreement and convention (puruṣatantra) and is not “self-contained” (vastutantra)?
Elisa,
I see the distinction you are making: in an inference from a to b, the relation between a and b is somehow naturally fixed. In an instance of testimony, where utterance c leads me to believe d, it is conventional. I appreciate that you are trying to make a clear distinction here and that's a good place to start.
Still not convinced, though. When my friend is in her office, she puts the ratio on. Simply by hearing her radio, I make an inference to her presence in the office. Yet the relata here are entirely conventionally and non-naturally related. On the other hand, it is possible to have a world where some terms are naturally related to their referents. My learning from persons using such terms to communicate would still be learning from words and not inference. So your definition is both avyApti and ativyApti.
If your point i merely to say that in inference, the relata must be joined, and it is less secure in testimony (which I doubt is your point), this is really immaterial. Both testimony and inference are by definition, successful; we can have misleading AbhAsas of either one.
Sudiptaji,
You are right, that for the sake of the argument, I was assuming that the clocks, etc. are well functioning and reliable. I am not sure what you are getting at, though. In a way, your point seems to go against Elisa's that in inference, things are somehow naturally connected but in testimony, it is conventional. This would support the clock/calculator being sources of testimony. Forgive my not entirely getting it.
Quick follow up:
I think that at the borders, certain knowledge sources are very similar and there are cases that could be read in either way. For example, in the kind of perception that nyAya recognizes as being mediated by prior cognition such that you see that the snow is cold, one could plausible argue that it is inferential. (I still think it's perceptual, but you could make the case, since it is so richly mediated by the deployment of interlocking concepts).
In the same way, maybe there's borderline cases here. I think that my suggestion for the difference between inference and testimony is that the latter depends on one's grasping language and trusting it. But this language can be very simple, like hand gestures or the face of a clock.
Dear Prof. Dasti,
My point is I dispute the very notion of reliability or āptatva, because of its flexibility and not being ‘intrinsic’ (vastutantrātmaka). No doubt I view the problem from the Advaita Vedānta standpoint, and argue against āptatva only with regard to the pāramārthika level of existence. But testimonies and āptatva do have some relevance (though a limited one) within the precincts of the vyavahārika realm. Śriharṣa and other neo-Vedāntins have refuted even the definitions of pada, vākya, sarvanāma, etc. but such an enterprise makes sense only with reference to the pāramārthika layer of existence, otherwise it would amount to committing the fallacy of ‘vadato vyāghāta doṣa’, consequent upon such statements as ‘mama mukhe jihvā nāsti’.
If jñānalakṣaṇasannikarṣa is accepted, there would be no ground for accepting anumāna. Thus says Swāmī Prajñānānanda Saraswatī in the Sanskrit introduction to his edition of Sureśvarācārya’s Naiṣkarmyasiddhi with the Bhāvatattvaprakāśika commentary of Citsukhācārya –
“अद्वैतवेदान्तिमते तार्किकाभिमतोक्तरीत्या ज्ञानलक्षणासन्निकर्षं स्वीकर्तुं न शक्यते अनुमित्युच्छेदप्रसङ्गात्। तथाहि – धूमदर्शने ज्ञानलक्षणासन्निकर्षेण वह्नेः प्रत्यक्षज्ञानसम्भवेऽनुमितिवैयर्थ्यापत्तिः।” (p. 32)
@Matthew,
you are right, I am trying to distinguish inference and linguistic communication. The distinctive point seems to me the possibility to build an invariable concomitance (vyāpti). In the case of Linguistic Communication, every essay to build an invariable concomitance seems to me deemed to fail, because there are just too many elements to relate. If you relate words/meanings, you end up needing also the reliability of the speaker, if you relate reliability and meaning, you still need to bridge the gap between words and reliability.
The linguistic nature of Linguistic Communication does not seem to me the distinctive factor, instead. The words uttered by a speaking clock (or by a speaking doll) are a definite set and are invariably connected to their meanings ("It's ten o' clock" always means that it is 10.00), so that —I think— a vyāpti can rapidly be established.
The counter-examples mentioned by Sudipta seem to me to be included among upādhis (fire and smoke are invariably connected, provided that the wood is not wet; "It's ten o'clock" and the fact that it is 10.00 are invariably connected, provided that the clock has been correctly programmed).
Dear Sudiptaji,
Thanks for the clarification!
At least amongst us non-liberated folks, I'd say that testimony is invaluable. Without it, we'd know nothing of history or, for most of us, science and world events beyond the borders of our home. We wouldn't know who Shankara or Shriharsha were either!
Thanks for that quote re: jNAnalakSaNasannikarSa. I do think, though that there may be a principled way to make a distinction between it and inference. In short, in perceptual cognition, the informational content is fused with the phenomenal content. In seeing a table, there is a phenomenological encounter with the appearance of a table; to use a certain jargon, I am appeared to table-ly. In inference, the content of the cognition is not fused with the phenomenal visual presentation. You see smoke, but through inference, you know there is a fire. You are not appeared to "fire-ly". With this distinction in place, one could say that in perception mediated by prior cognition, you "see" the coldness of fire. Now, I am not saying this is an accurate way to explain what's happening, but for Nyaya, this is, at least theoretically, is a principled way to make the distinction. The coldness is somehow apprehended within the phenomenology of the perceptual experience itself. Pace Swāmī Prajñānānanda Saraswatī, this would not make anumAna useless, since there is a clear criterion for differentiating between them.
Again, I agree that the line is blurry (that was my original point), but I don't think that on their own theory, Nyaya would make inference useless or superfluous.
My former teacher, David Sosa, has argued that testimony and many kinds of inference were really instance of what Fred Dretske called displaced perception (can't get into it here, but it's in Dretske's book *Naturalizing the Mind*). That whole project ties into this discussion in many interesting ways.
I need to be offline for a few days so I won't be able to respond for a while. Be well!
Post a Comment