Abstract
Language matters to epistemology for two separate reasons (although they are no doubt connected).
My interest in testimony derives from Gareth Evans, as does my conviction that it cannot be accommodated by the sort of account of knowledge which I attack in this paper. I believe I also owe to him my interest in the sorts of case I discuss in §4 below, where knowledge is retained under the risk that what would have been knowledge if the relevant fact had still obtained is not knowledge because the fact no longer obtains. I have benefited from comments by Robert Brandom and Jonathan Dancy.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
Ludwig Wittgenstein, On Certainty (Basil Blackwell, Oxford, 1969), §141.
Language is not just one of man’s possessions in the world, but on it depends the fact that man has a world at all’: Hans-Georg Gadamer, Truth and Method (Sheed and Ward, London, 1975), p. 401. I hope to elaborate this admittedly difficult idea elsewhere.
Much of On Certainty is about the status of this sort of knowledge. Wittgenstein himself is dubious about counting it as knowledge; but I think that is inessential to his main point, which is to warn against assimilating the sort of thing in question — propositions which, by not being on the agenda for testing and confirmation, function as pivots on which our practices of looking for grounds for belief can hinge — to cases where it makes sense to look for the grounds of a belief. (Wittgenstein’s doubt about counting these propositions as known may reflect the influence of the kind of conception of knowledge that I am going to attack.)
In characterizing an episode or a state as that [better: one] of knowing,we are not giving an empirical description of that episode or state; we are placing it in the logical space of reasons, of justifying and being able to justify what one says’: ‘Empiricism and the philosophy of mind’, in Science, Perception and Reality (Routledge and Kegan Paul, London, 1963), at p. 169. I put in the qualification about rational animals in order to leave room for the concept of knowledge to be extended to non-rational animals also; but nothing in this paper need depend on that.
A principle on these lines was stated by Ernest Sosa in ‘The analysis of “knowledge that P”’, Analysis xxv (1964–5), 1–8, at p. 8. It matters that I say ‘may become available to’ and not ’is acquired by’. For one thing, the opportunity for knowledge may not be there for a hearer even if the speaker is giving expression to his knowledge; see §6 below. For another, one cannot be forced to avail oneself of knowledge that one is in a position to acquire; excessive caution, for instance, may lead one to pass up an opportunity.
See Elizabeth Fricker, ‘The epistemology of testimony’, Aristotelian Society Supplementary Volume lxi (1987), 57–83, especially at pp. 60–2, for some discussion of the different possibilities here, especially on what might be meant by requiring the knower to possess the justification. I agree with her that we lose the point of invoking the space of reasons if we allow someone to possess a justification even if it is outside his reflective reach.
This is a point at issue between Fricker, op. cit., and David Cooper’s contribution to the same symposium, ‘Assertion, phenomenology, and essence’, ibid. 85–105. I think Fricker is quite right that this knowledge is perceptual. (But that is not to say that it constitutes an absolute starting-point: see §5 below.) Cooper suggests that this phenomenological and epistemological position must miss the insights of Romanticism, but that strikes me as the reverse of the truth. However, the point is not central to my concerns here.
See Fricker, op. cit. pp. 72–3.
Willingness to say ‘No’ may seem to preclude claiming that the tourist knows where the cathedral is. But what is threatened is only one reading of the claim that he knows that he knows where the cathedral is. And it is quite dubious that someone who knows must know that he knows, in the relevant sense. (See David Wiggins, ‘On knowing, knowing that one knows, and consciousness’, in E. Saarinen, R. Hilpinen, I. Niiniluoto, and M. Provence Hintikka (eds.), Essays in Honour of Jaakko Hintikka (D. Reidel, Dordrecht, 1979), pp. 237–48.) There may be another reading of the principle that a knower knows that he knows; I have phrased my sceptical queries so as to leave it open that, if we stop looking for non-question-begging certifications of epistemic standing, we may be able to retrieve a possibility of crediting the tourist with knowledge that his informant is competent and trustworthy, as something on a level with the knowledge he acquires in the transaction, not prior to it in the space of reasons. (Compare the idea that knowledge that one is not dreaming is on a level with the knowledge of the environment that one’s senses are yielding one, not something one would need to be able to credit oneself with first, in order to be able to take it that one’s senses are indeed yielding one knowledge of the environment. See my ’Singular thought and the extent of inner space’, in Philip Pettit and John McDowell (eds.), Subject, Thought and Context (Clarendon Press, Oxford, 1986), pp. 137–68, at pp. 147–8.)
Some such suggestion is implicit in Fricker’s argument, op. cit. p. 75, against the idea that one can be entitled to a presumption of sincerity and competence in the absence of special evidence to the contrary; she responds ‘I would not like to be obliged to form beliefs in response to others’ utterances in accordance with this presumption!’ But I do not want to defend the idea that Fricker is attacking here, that there is a general presumption of sincerity and competence (as if gullibility were an epistemic right, or even an obligation). In the case I am considering, I think the tourist is entitled to his belief about where the cathedral is, without taking care to rule out the possibility of a practical joke; but I do not think that is because he is exercising a general presumption of sincerity and competence. That is the sort of thing that it is natural to appeal to in a version of the conception I am attacking, one which keeps the idea that mediated standings consist in the cogency of arguments but is less optimistic than Fricker about how cogent the available arguments ares, unless they are beefed up with general presumptions of this sort. I want a more radical departure from the governing conception. This should become clearer in due course.
See Blackburn’s ‘Knowledge, truth, and reliability’, Proceedings of the British Academy but (1984), 167–87; the quoted phrase is from p. 185.
Induction can have a confusing effect here: it can seem to be a counter-example to the principle. But demanding that an argument be conclusive is not the same as demanding that it be deductive. I suggest that philosophers have been insufficiently willing to query the antecedent of the principle at n. 29 (pp. 242–3) of my `Anti-realism and the epistemology of understanding’, in Herman Parret and Jacques Bouveresse (eds.), Meaning and Understanding ( Walter de Gruyter, Berlin and New York, 1981 ), pp. 225–48.
Perhaps some will be tempted to maintain that even so one does know that the result will be red, protecting that claim from being undermined by the fact that one does not know that the result will not be white on the ground that knowledge is not closed under known implication: for which thesis, see Robert Nozick, Philosophical Explanations (Clarendon Press, Oxford, 1981), pp. 206–11. Whatever the merits of the thesis, such an application of it strikes me as desperate.
It is a familiar experience to find, some time after one has, say, missed the newspapers for a week, that things one thought one knew have for some time been no longer true. It is striking that the experience has no tendency to dislodge one’s belief that one knows a great deal else in the sort of way in which one thought one knew the thing that one has just been disabused of.
We can take the former knowledge as unproblematic, for the purposes of the example. Perhaps it was derived from broadcast coverage of Churchill’s latest birthday. (The point I am making with this example is about retention of knowledge, and is not meant to turn especially on the fact that the knowledge retained was originally acquired by testimony; see below.)
There may be a temptation to say that the same goes for the roulette-wheel case, on the basis that if we disallow it, we disallow all knowledge by induction. But it is simply not true that in the roulette-wheel case one knows that the outcome will be red. If induction is a way of coming to know things, that is not an example of it. (Knowing that the outcome will probably be red is of course quite another matter.)
A case like this one, involving retention of knowledge not originally acquired by testimony, is briefly discussed by David Braine, ‘The nature of knowledge’, Proceedings of the Aristotelian Society lxxii (1971–2), 41–63, at p. 42.
Blackburn, op. cit.,proposes that one knows when one’s informational state, conceived otherwise than as having the fact known in one’s cognitive grasp, leaves no real possibility (’chance’) that things are not as one takes them to be. (Having the fact known in one’s cognitive grasp would leave no chance at all of being wrong, but Blackburn contemptuously dismisses conceptions of knowledge on these lines.) He applies that account of knowledge to the general hypotheses on which sceptical arguments trade (that one is a brain in a vat, and so forth). The upshot is that whether one counts as knowing that such hypotheses do not obtain depends on who has the onus of proof in a dispute with a sceptic. But if my informational state, between intakes of news, with respect to who is President is not allowed to embrace the fact that Bush is still President, it surely leaves a real possibility that Bush is no longer President. Assassinations, or other sudden deaths, of Presidents are `kinds of things that happen’. So are misleading perceptual appearances (and so on; different kinds of real possibility are relevant to the different sorts of knowledge). So even if Blackburn achieves an onus-swapping standoff with the kind of sceptic who attempts to wield general sceptical hypotheses to undermine whole regions of knowledge all at once, it looks as if his picture will deprive us of pretty much the same knowledge, only piece by piece. If we deny ourselves a `guaranteeing’ conception of a putative knower’s informational state, the less rich informational state we thereby restrict ourselves to will always leave open perfectly real possibilities (not the sceptic’s arguably unreal possibilities) that he is wrong. (Blackburn simply misses this point; he concentrates entirely on the general sceptical hypotheses, as if there could be no threat to ordinary knowledge claims except from them.)
Christopher Peacocke discusses retained knowledge in Chapter 10 of Thoughts: an Essay on Content (Basil Blackwell, Oxford, 1986). He defends a ‘Model of Virtual Inference’ which represents such knowledge as requiring the knower to have at his disposal a sound abductive argument to the truth of what he is said to know. I think it is significant that Peacocke considers only knowledge of standing states, such as that Hume died in 1776. Perhaps someone who finds himself seeming to remember that fact can have a sound abductive argument from his present informational state, considered as not embracing the information that Hume died in 1776, to the conclusion that this is so. But that does not carry over to retained knowledge of changeable states. From my willingness to vouch for Bush’s being President, I cannot get by abduction to his being President now, as opposed to his having been President when I last checked.
In Blackburn’s terms, seeing that…, remembering that…, and so forth are ‘guaranteeing’ epistemic positions. Blackburn suggests (op. cit. pp. 176–8) that if one traffics in the idea of ‘guaranteeing’ states, in the stamping-grounds of sceptics (knowledge of the external world and so forth), one lapses into bizarre imagery (see his remarks about ‘the glassy blob of the mind’, p. 177). But the relevant concepts belong to sheer common sense; what would be bizarre is to suggest that we do not achieve such ‘guaranteeing’ positions as seeing that things are thus and so. See, further, n. 43 below. Blackburn’s moves are skewed, I think, by an aspiration (which he tends to read into others) to answer sceptical challenges. I do think the epistemological outlook I am recommending makes sceptical challenges seem less urgent, but obviously not by answering them. (If someone is exploiting a general sceptical hypothesis in order to attack a knowledge claim, he will not be impressed if one attributes to oneself a ‘guaranteeing’ informational state with respect to the proposition one claims to know; if the sceptical hypothesis holds, the attribution cannot be true.)
What counts as excessive depends on the proposition known. If one missed the British news media for a fairly long period, and when one tuned in again there were no lingering traces of national mourning, it might not be doxastically irresponsible to take it that a greatly loved national figure like Churchill was still alive; it would be different with someone else.
The topic of doxastic responsibility is clearly complex. Note that the standards can depend on what is at stake. Consider again the case of the child at school. If nothing turns on it, we might casually credit her with knowledge of the arrangement of the furniture in the living room of her house. But if we tell the story so that something that matters a great deal to her depends on whether she is right, it may become doxastically irresponsible for her to vouch for the layout’s being as she recalls it to be. In such circumstances, it starts to be significant for her epistemic status that her parents may have moved the furniture, and she is in a position to know that that kind of thing does happen.
One does not count as seeing something to be the case, even if the fact that that is how things look to one results, in the way characteristic of seeing, from the fact that that is how things are, if one’s taking it that that is how things are is doxastically irresponsible (one has, say, excellent reasons for mistrusting one’s vision).
The idea that perceptual appearances can be absolute starting-points is one form of the Myth of the Given, demolished by Sellars in `Empiricism and the philosophy of mind’, op. cit.
The thesis that there are absolute starting-points is a way to formulate epistemological foundationalism. The traditional competitor of foundationalism is coherentism, and that label fits the position I am endorsing here. But we need to get straight how inferences (other than those involving the factiveness of the epistemic concepts) are relevant before we embrace coherentism as an alternative to foundationalism. Otherwise the coherentist alternative is the heroic position, and it is not clear that it yields any real improvement over foundationalism. (See Crispin Wright, `Facts and certainty’, Proceedings of the British Academy lxxi (1985), 429–72, at p. 469.)
Testimony figured only incidentally, presupposed to be a source of knowledge, in the examples of §4.
Different descriptions of standings are appropriate for different manners of acquisition of knowledge by testimony. One with more general application is `having learned (from such and such a source) that…’. Consider also the expression ‘I gather that…’, which claims knowledge from testimony without identifying the source. (Compare `I heard it through the grapevine’.)
The respect of likeness is that hearing from someone that things are thus and so, like seeing that things are thus and so, is a `guaranteeing’ informational state. Of course that is compatible with all kinds of differences. In particular, I am not suggesting that in acquiring knowledge by testimony one experiences things to be the way one comes to know they are (an obvious phenomenological falsehood which Fricker, op. cit. pp. 74–5, spends some time denying). The crucial notion is that of a `guaranteeing’ informational state whose possession of that feature is not to be understood in terms of how strong a reason for believing the proposition in question is afforded by an underlying non-`guaranteeing’ informational state. There is no need to assimilate this to the idea of a direct perceptual or quasi-perceptual mode of access to the state of affairs known. (Cf. Wright, `Facts and certainty’, at pp. 443–4). On the contrary: the epistemic standing constituted by having heard from someone that things are thus and so is clearly mediated by having heard the person say that things are that way; and this mediation (unlike the mediation of seeing that things are thus and so by having it look to one as if things are thus and so) clearly precludes the idea of a direct perceptual access to what one comes to know. What I am objecting to is the prejudice that what this mediation amounts to must be that the non-`guaranteeing’ informational state yields the subject something on the lines of a premiss from which he can (with other premisses if necessary) infer the proposition he is said to know, in such a way that his epistemic standing can be made out to consist in the cogency of the argument.
Here it is important that the topic of this paper is the first of the two sorts of knowledge through language which I distinguished in §1, and not the second. The remark in the text would be quite wrong about the second; as Wittgenstein says (On Certainty,§ 143): `A child learns there are reliable and unreliable informants much later than it learns facts which are told it.’
Peacocke, op. cit. pp. 149–50, gives an example. Mary forms beliefs about whether it is raining sometimes by looking and sometimes by deduction from astrological principles. Her friend cannot acquire knowledge that it is raining from her say-so, even on the occasions when what she is giving expression to is knowledge.
Fricker, op. cit.,considers the epistemology of testimony in the context of a choice between Justificationism and Reliabilism. Reliabilism as she explains it may not require that someone who acquires knowledge by testimony even has the concept of another person speaking his mind; if there is such a requirement, it is only fortuitous. In effect, Reliabilism in Fricker’s contrast abandons the idea that knowledge is a standing in the space of reasons. I agree with her rejection of this position, but I am taking issue with her implicit suggestion that the only way to keep the space of reasons relevant to the epistemology of testimony is by adopting the sort of view that I considered in §2.
It is noteworthy that Peacocke’s example, cited in n. 31 above, has Mary talking to her friend (who presumably knows her peculiar ways of coming to believe that it is raining). Peacocke does not consider how, if at all, the case is altered if we consider someone who does not know Mary, hearing her say that it is raining on one of the occasions on which her utterance is an expression of knowledge.
J4 There seems to be a general possibility of such cases; something can be irresponsible for one person and not for another because the first knows something that the second does not know. (Such cases would be counterexamples to something one might mean by saying that knowledge is seamless. But note that they do not threaten the principle suggested by Gareth Evans at p. 331 of The Varieties of Reference (Clarendon Press, Oxford, 1982).)
Someone who acquires knowledge by testimony has some reason, independently of our crediting him with that knowledge, for believing the proposition in question. That seems guaranteed by his meeting the condition which I have suggested is necessary for doxastic responsibility: that his belief-acquisition be rationally responsive to considerations whose relevance is ensured by an understanding of how the knowledge-acquisition that he takes himself to be engaged in works. Peacocke, op. cit. p. 166 writes: ‘There is a strong intuition that a belief is not knowledge if it is acquired by testimony for which there is no inductive or abductive argument available to the believer to the truth of the testimony.’ Peacocke goes on to suggest that that intuitive requirement is not met in a case he describes, involving an isolated archaeological relic with a single intelligible sentence inscribed on it. I think that is wrong: the requirement is very weak (simply that one must not be totally without reason for belief), and surely in Peacocke’s case there is some inductive argument (from what civilizations generally do with inscriptions) for believing what is written on the relic. The important point is that the requirement is far too weak for meeting it to be intelligible as what makes a case constitute one of knowledge. (The point here is close to one that Peacocke makes himself, at p. 167, n. 9, in arguing cogently against the idea that ‘Necessarily, most assertions are true’ can play a central role in the epistemology of testimony). The intuition is no recommendation for an inferential model of knowledge by testimony, as Peacocke suggests. (I suspect that the presence of such a model in the context distorts Peacocke’s sense of whether the intuition is met in his archaeological case: meeting the intuitive requirement is only a necessary condition for knowledge by testimony, but Peacocke responds to the case as though it were sufficient.)
Compare Peacocke’s remarks about the need to explain such formulations, op. cit. p. 149.
See Robert Brandom, `Asserting’, Nous xvii (1983), 637–50.
For these labels, see, e.g., Nozick, op. cit. pp. 280–3. For the idea of a mixed or hybrid epistemology, see Peacocke, op. cit. In Chapter 9 Peacocke sets out an externalist reliability condition for knowledge (see pp. 155, 157); then in Chapter 10 he argues that this must be supplemented with a condition requiring `internal rationality’ (p. 156).
We could complicate this to allow for cases where the subject is blameworthy because his restricted informational states fail to include something they should have included; he should have checked something but did not. That is an analogue to negligence in the field of practical blameworthiness. The complication makes no difference to my point: even if one exercises maximal care at achieving the right restricted informational states, one will still be at the world’s mercy in believing what they give one reason to believe. 40 The point is peculiar to empirical knowledge. If someone takes himself to have proved a conclusion or computed a result when he has not, there must have been a defect in his moves in the space of reasons; it cannot be that the only thing he can blame for what has gone wrong is the world. That is essentially the feature of proof (or computation) that Crispin Wright aims to generalize, in his account of what it is to have verified a statement (`Strict finitism’, Synthese 11 (1982), 203–82, at pp. 210–18). In a way that is very strange by my lights, Wright combines an understanding of that feature of proof (or computation) with endorsing, even in that case, the retreat to a lesser informational state, the move which I am trying to explain as motivated by the desire to find a region where thought is immune to the world’s unkindness. He writes (p. 210): ‘If arithmetical computation is to be a paradigm of verification, then to be entitled to claim to have verified a statement cannot be to be entitled to claim a conclusive, indefeasible warrant for its assertion; for the most painstaking and careful execution of a computation confers no guarantee that is correct.’ That is to retreat (in respect of what warrants one’s assertion) from `I have proved that it is so’ (whose truth surely would constitute a conclusive, indefeasible warrant) to ‘I have before me what, on painstaking and careful inspection, appears to be a proof that it is so’. But the retreat seems unmotivated, given the fact that if I am misled in such a case, the fault is in my moves in the space of reasons, not in the world. I suppose it is because Wright thinks mathematical proof and empirical verification are on a par in respect of the necessity of that retreat (and so in respect of the defeasibility of available warrants) — in effect, on a par in respect of vulnerability to the Argument from Illusion — that he thinks he can generalize that feature of mathematical verification without risking an undue concession to scepticism. I think the resulting epistemology is disastrous. (These remarks improve on my discussion of Wright’s epistemology in `Mathematical platonism and Dummettian anti-realism’, in Dialectical (1989), 173–92.)
I press this question in my `Criteria, defeasibility, and knowledge’, Proceedings of the British Academy lxvili (1982), 455–79.
On the analogous temptations in philosophical thinking about practical rationality, see especially Bernard Williams, `Moral luck’, in his Moral Luck ( Cambridge University Press, Cambridge, 1981 ), pp. 20–39.
This formulation should make it clear how wildly off-target Blackburn is (op. cit.,p. 176) in supposing that my appeal to `guaranteeing’ informational states belongs within the general framework of the attempt `to ensure that there is no element of luck, or even contingency, in the true believer’s title to knowledge’. The traditional effect of the attempt to transcend luck is that the area of known fact is shrunk ‘potentially down to an entirely subjective realm’; Blackburn takes me to offer a different option, within the same general framework, according to which, instead of a shrinkage in what can be known, the mind (the seat of these supposed luck-free `guaranteeing’ states) expands to ‘embrace’ all sorts of worldly states of affairs. No wonder Blackburn finds the idea crazy; it is crazy. A `principle of charity’ might have led Blackburn to wonder whether it can have been what I was proposing in the work he is discussing (`Criteria, defeasibility, and knowledge’, op. cit.). But he is so locked in to the thought that epistemology must centre on a luck-free zone (a role played in his own favoured epistemology by the `indicative’ states to which we are pushed back by a generalized form of the Argument from Illusion, plus what can be inferred from their content) that he cannot comprehend how I can have been questioning that framework conception; so he saddles me with the insane position which is the only interpretation that my words will bear within the framework.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1994 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
McDowell, J. (1994). Knowledge by Hearsay. In: Matilal, B.K., Chakrabarti, A. (eds) Knowing from Words. Synthese Library, vol 230. Springer, Dordrecht. https://doi.org/10.1007/978-94-017-2018-2_10
Download citation
DOI: https://doi.org/10.1007/978-94-017-2018-2_10
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-4287-3
Online ISBN: 978-94-017-2018-2
eBook Packages: Springer Book Archive