Conversation, Epistemology and Norms

Transcription

1 Conversation, Epistemology and Norms STEVEN DAVIS Abstract: It is obvious that a great many of the things that we know we know because we learn them in conversation with others, conversations in which it is the intention of our interlocutor to inform us of something. It might be thought that only assertoric acts are informative. I shall argue that there is a range of conversational interventions that have this characteristic, including speech acts, presuppositions and conversational implicatures. The main focus of the paper is a discussion of the different norms, both moral and epistemological, that entitle us to believe what we learn from conversations. I compare our entitlement to believe what we learn from conversation with our entitlements to believe what we learn from perception. In providing an account of our epistemic warrant for our knowledge gained in conversation with ours, I draw on the work of Tyler Burge (1993 and 1997). Relying on others is perhaps not metaphysically necessary for any possible rational being. But it is cognitively fundamental to beings at all like us. (Burge, 1993, p. 466) Successful communication entails understanding. We might even take understanding to be one of the necessary goals of individual participants in a conversation (Davis, pp ). With understanding comes knowledge, knowledge at least of what has been communicated. Much has been made of testimony as a primary source of knowledge. I shall try to show that other conversational interventions are similar sources of knowledge, presuppositions, implicatures, and other types of speech acts besides testimony, for example. I shall draw on the work of Tyler Burge in attempting to provide the grounds for our knowledge in such cases (1993 and 1997). Burge is primarily concerned with what he calls something that is presented as true (1993, p. 467). It is not clear that this applies to the full range of conversational interventions that I shall consider. 1 Part of my goal is to extend his theory to other conversational interventions. I shall, however, part company with Burge on several points. I shall not follow him in his defense of his views about beliefs acquired from others being a source of a priori knowledge or in his claim that the rationality I would like to thank Jonathan Berg, David Davies, Christopher Guaker, Jérôme Pelletier, the members of the Jean Nicod Institute, espcially François Recanati and Dan Sperber for help with this paper. Address for correspondence: Philosophy Department, Simon Fraser University, Burnaby, B.C. V5A 1S6. Canada. sdavis@sfu.ca 1 Although Burge explicitly includes within his range of presentation as true, presupposition and conversational implicature, it is not clear that Burge s account applies to the full range of conversational interventions, asking a question, for example, that can be a source of knowledge (Burge, 1993, p. 482 n. 20). Mind & Language, Vol. 17 No. 5 November 2002, pp , 108 Cowley Road, Oxford, OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

2 514 S. Davis of the source of a conversational intervention alone can provide the epistemological grounding for conversational knowledge. 2 I shall argue that an appeal must also be made to the morality of the source. 3 Let us imagine a simple conversation in which Sam says to Alice, 1. There is a party tonight. If Alice understands what Sam said, she understands that, 2. Sam said that there is a party tonight. In this case, we can say that Sam has successfully communicated what he meant to Alice. In cases of successful communication, there is the possibility of belief arising out of understanding and knowledge arising out of the belief. In our simple example, Alice might come to know that there is a party tonight, because Sam told her that there is. But it does not follow from Alice s knowing what Sam communicated, that she believes what he said, nor does it follow from her believing that there is a party tonight that she knows that there is. This much should be obvious. Understanding that p does not imply believing that p and believing that p does not imply knowing that p. I wish to pose two questions. What are the grounds for our beliefs that arise through communication and what more has to be added to our having grounds for our beliefs for us to have knowledge? 4 The range of cases in which we have knowledge because a belief has been communicated to us goes well beyond the simple example in 1. Think about how we know many things that are the result of scientific investigation. It is not because we have done the experiments or constructed the theories, but because we have been told about the results of the experiments and the conclusions of the theories. Or consider what we know about the past, or about current events, plane crashes in the United States or tropical storms in India, for example. It is not because we witnessed the events, but because we were told about what happened or is happening. In these and similar cases, we take and are entitled to take the word of others as the unquestioned source for our knowledge (Burge, 1993, p. 485). This is not to suggest that in these and similar cases we might not have justification in addition to our having been told what is or has happened. We might see pictures of the events on television, 2 I do not take issue with Burge s claim that there are cases of communication that gives rise to a priori knowledge, although doubts have been raised (see Bezuidenhout, 1998). My concern is to examine the kinds of conversational interventions, other than testimony, that give rise to knowledge and to show what entitles us to this knowledge. 3 I shall part company with Burge on other points that I shall indicate as they arise in the course of my argument. 4 I shall refine the notion of grounds in what follows drawing on a distinction between justification and entitlement.

3 Conversation, Epistemology and Norms 515 for example. Nor is this to suggest that the source of our knowledge might be other than its having been communicated to us. We might be the scientist who has done the experiments or constructed the theories. Of course, there are cases in which our prima facie entitlement can be overridden. If we ask someone the time who says 5:00 p.m., with a dazed look and without consulting his watch, we might not believe him. In addition, there are types of discourse in which we do not take the word of others as an unquestioned source of knowledge, when, for example, we are involved in heated political or philosophical debate, when a policeman grills a suspect, or when professors ask students questions (Burge, 1993, pp. 468 and 484). But in many of our normal transactions with others, whether in conversation or in something that we read, the word of others is what entitles us to our belief. 5 What then is the nature of this source that gives rise to knowledge? One view, advanced by C.A.J. Coady (1992), is that where others are the source of knowledge, the source is more specifically their testimony. Coady takes testimony to be a speech act that can be made explicit with a performative verb formula, I testify that or with I bear witness to the fact that or more simply with I tell you that (Coady, pp ). According to Coady, someone telling me the time would be testifying as to the time of day. Coady recognizes that testimony is usually associated with formal court proceedings and does not apply to normal talk exchanges. Despite this, he uses testimony for both what he calls formal testimony and natural testimony. Formal testimony is the testimony that occurs in court proceedings or similar institutionalized settings; natural testimony is what occurs in our normal conversations with others. I believe that even if we accept Coady s distinction, he has mischaracterized the range of conversational interventions that are a source of knowledge by assimilating all such interventions to speech acts. Undoubtedly Coady is right to point out that someone telling us something can be a source of knowledge. Suppose that Sam knows 1 to be true and communicates it to Alice. Further, suppose that she understands and believes what he said. Alice would know, thereby, that there is a party tonight, unless she had an overriding reason to doubt Sam s word. I shall call Sam s speech act a conversational intervention and the knowledge that Alice thereby gains conversational knowledge. 6 Let us accept Coady s claim that Sam s saying 5 In what follows, I shall concentrate my attention on speech and conversation, but I think that without undue distortion what I say about knowledge gained from others can be carried over to the written word. 6 There is a range of what might be considered to be conversational knowledge that I shall not consider here. For example, in understanding 1, Alice might come to know that Sam is speaking English or that he is referring to the evening of February 13, 2001 in uttering. I think that the second, but not the first can be accommodated in the account that I shall give of conversational knowledge, but I shall not take it up here. The first sort of knowledge does not turn on the hearer s understanding what has been said, but rather on her understanding something about the sentence uttered. It is the former sort of understanding and the knowledge that it can give rise to with which I am concerned here.

4 516 S. Davis what he said is a case of testimony. 7 Testimony does not exhaust the range of conversational interventions that are a source of conversational knowledge. Other speech acts should be included. Warnings, predictions, etc. are, as well, sources of knowledge. We can divide speech acts, such as assertions, warnings, predictions, etc. into their force and propositional content (Searle, 1969, p. 31). 8 In these sorts of acts, the knowledge is knowledge of the content of the speech acts. In addition, speech acts can give rise to knowledge not only of their propositional contents, but also of their felicity conditions. Suppose that Sam promises Alice to be at the party and she understands that he has done so. Alice would thereby come to know that Sam has placed himself under a prima facie obligation to be there. In addition, she would know that he believed that she wanted him to be there. Or take another case. Suppose that Sam congratulates Alice for arranging the party and she understands that he has congratulated her. Alice would thereby come to know that Sam believes that she has arranged the party and that he thinks that is a good thing for her to have done so. I can come to know something not only because I understand the speech acts performed, but also because I understand what is conversationally implied. Suppose Alice arrives one hour late for an appointment and Sam archly says, 3. Thanks for being on time. In so saying Sam implies that Alice is late for their appointment and that he is somewhat miffed by her lateness. Moreover, if Alice understands and believes what Sam implies and Sam knows what he implies to be true, she comes to know that she was late for their appointment and that her being late upset Sam, if she did not know it already or does not have any overriding reason not to believe what Sam implies. How does Alice come to know this? She does not know it because Sam has told her, but because he conversationally implied it. So we can add conversational implicatures to speech acts as a source of knowledge from our conversations with others. Speech acts and conversational implicatures, however, do not exhaust the conversational interventions that are a source of knowledge. We should also include presupposition. Consider: 7 I think that Coady is mistaken in characterizing Sam s speech act as testimony. The reason for the mistake lies in Coady s analysis of testimony, the details of which I do not have the space to consider here (Coady, 1992, p. 42). 8 According to Searle, all speech acts can be divided into force and content. For example, if someone were to ask the question, Is John coming to the party? the resulting act can be divided into the propositional content [John is coming to the party] and the force, a question. But in this case, a hearer understanding the question would not come to know that John is coming to the party. So it is only a subset of speech acts that can give rise to knowledge of their propositional content.

5 4. Fred has stopped drinking. Conversation, Epistemology and Norms 517 If Sam knows 4 to be true and he communicates it to Alice and she believes it, Alice would thereby come to know that Fred had been drinking, although Sam did not tell her that Fred had been drinking. Rather, in saying that Fred had stopped drinking, Sam presupposed that Fred had been drinking, something that Alice can come to know if she understands what Sam said. These three cases do not fully capture the range of conversational interventions that can give rise to knowledge. Suppose that Sam is talking to Alice about his recent divorce from Ann. To indicate how he feels he wanly says, 5. I still miss her. Alice can thereby come to know how Sam feels about his divorce, but it is not knowledge that can be encapsulated propositionally (Sperber and Wilson, 1995, pp ). There is nothing that Sam can tell her that would sum up how he feels or that Alice can say that encapsulates her knowledge of how he feels. Nevertheless, Alice knows something that she did not know before Sam talked to her about his divorce and she knows this because he expressed how he feels by saying 5. This case is similar to 3 in that Sam has conversationally implied how he feels, but dissimilar in that what is conveyed is not propositional. Hence, Alice s knowing how Sam feels about his divorce, although not propositional knowledge, should not be excluded from what people can come to know by a conversational intervention. It is, however, not necessary to expand the list of conversational interventions, since in this case Sam s indicating to Alice how he feels about his divorce can be subsumed under conversational implicature. What this shows is that some conversational implicatures are not propositional. To sum up, we have three kinds of conversational interventions as sources of conversational knowledge, speech acts, conversational implicatures and presuppositions. Conversational knowledge, then, is knowledge that arises from the understanding of a conversational intervention. Not all knowledge we have that involves conversational interventions is conversational knowledge. There are two sorts of cases that I would like to distinguish from the examples above. In the first, the hearer draws on knowledge that he already has and uses it in conjunction with his understanding of a conversational intervention to draw an inference that provides him with further knowledge (Coady, 1992, p. 51). Sam and Alice are chatting about a party that was held last night and Sam, knowing 6 to be true, says, 6. Everyone at the party was drunk. Suppose that she believes what Sam said and she knows that,

6 518 S. Davis 7. Fred was at the party. From this and 6, she deduces that Fred was drunk. So Alice comes to know, 8. Fred was drunk at the party last night. Sam s saying 6 is in part the source of Alice s knowledge, but the epistemic warrant for Alice s knowledge of 8 is different from what I have called conversational knowledge. In these examples, Alice s entitlement to believe what she does rests on her understanding of Sam s conversational interventions. In coming to know 8 part of what entitles Alice to her belief is that Sam told her 6, but also on her being warranted to believe that Fred was at the party, a belief the warrant for which might not have arisen through a conversational intervention. Suppose that her entitlement is grounded in her having seen him go into the party. On this assumption, her knowledge of 8 contrasts with her knowledge of 1. In the latter example, the entitlement is that Sam told her; in the former example, it is Sam s telling her 6 and her perceptual belief that she saw Fred go into the party. What 1 5 have in common and what differentiates them from this case is that the entitlement to the belief is Alice s understanding that Sam Ø s that p, where Ø is a placeholder for a conversational intervention. 9 So we can say that conversational knowledge is that knowledge the sole warrant for which is the understanding of a conversational intervention. Had Alice s knowledge of 7 come from a conversational source, then it might appear that this example should be classified with 1 5. There is, however, one additional element involved here. From her knowledge of 6 and 7 Alice draws an inference to 8 and this inference is part of the warrant for her knowledge. Consequently, I shall exclude from the class of conversational knowledge knowledge obtained by drawing conclusions from premises, the knowledge of which is gained by conversational interventions. 10 The reason for excluding these sorts of cases is that if they were included in the class of conversational knowledge, so too would mathematical knowledge that depends on proofs using premises learned from others. But such knowledge is not justified solely by our understanding conversational interventions, but by the proofs we give. Thus, such knowledge falls outside the class of conversational knowledge. It might be claimed that our conversational knowledge is inferential and thus, our entitlement rests not only on our understanding 9 The example in 5 does not fit the form here, since Alice s entitlement is her understanding how Sam felt. But what Alice understands does not have the form ø-ed that p. This case could be included here, but it complicates the discussion unduly and it is not necessary for my purposes. See Sperber and Wilson for further discussion (pp ). 10 This is meant to exclude, for example, Alice s knowing that someone at the party was drunk, a conclusion she draws from 6.

7 Conversation, Epistemology and Norms 519 conversational interventions. It could be argued, for example, that Alice s knowledge of 1 arises by inference from 2. Consequently, by ruling out inference as part of the warrant for conversational knowledge I have ruled out what I have given as paradigmatic examples of conversational knowledge. I shall return to this issue below. I have claimed that successful communication involves knowledge, knowledge of what is communicated. It would seem, then, that this knowledge should be included within conversational knowledge, since it clearly arises because of conversational interventions. I shall, however, exclude this knowledge from conversational knowledge. Let us consider an example of successful communication. In understanding what Sam means in uttering 1, Alice comes to know 2. Alice s knowing 2 does not imply her knowing 1. She might know the former without knowing the latter, since she might doubt that what Sam says is true. Hence, understanding that someone means that p does not entail knowing that p. 11 I shall restrict conversational knowledge to the knowledge that we have that is over and above the knowledge we have by understanding what someone means and is knowledge that we have that arises from this understanding. It might seem that from my discussion of conversational knowledge it is a necessary condition for someone to have it that he must be able to cite as his justification that he learned it from a conversational intervention. To put this point more precisely, it might seem that I am committed to the following thesis about conversational knowledge. 9. H has conversational knowledge of p only if were H asked how he knows that p, he would be able to provide as a justification for his belief that p that someone Ø-ed that p. Let us consider an example. 10. (a) Alice: Is Fred coming to the party? (b) Sam: He is. (c) Alice: How do you know? Did he tell you? (d) Sam: No. Betty told me. In the context of the conversation, Sam s saying that Fred is coming to the 11 Again, the example in 5 does not fit the required form. In this case, Alice means to indicate how she feels about her divorce, something that she can communicate to Sam. In addition, it does not fit the cases in which by understanding a speech act the hearer comes to know one of the felicity conditions of the speech act. For example, if Sam asks what time it is and Alice understands him, she can come to know that Sam wishes to know the time. I shall postpone full generality. At this point, it is not necessary, since as I shall show 9 and the related 12 are false.

8 520 S. Davis party implies that he knows that he is. On being queried about how he knows this, he gives as his justification for his belief that Betty told him. The question is whether being able to give justifications of this sort is necessary for someone to have conversational knowledge. I shall argue that it is not. There is an important distinction to be drawn between our being entitled to believe something and our being able to give a justification for our belief. To return to the example in 10, although Sam is able to say how he knows that Fred is coming to the party, his knowing that Fred is does not require him to be able to say how he knows this. Much of what we know we know, despite having forgotten the warrant for our knowledge (Burge, 1997, p. 38). Suppose that someone asks me about the construction materials of Kensington Palace. I say without hesitation, 11. It is built out of red brick. In this case, I have no memory of how I came to know this. Since I have been in London, I might well have seen the Palace and come to know 11 because of my perceptual experience. Equally as well, I might have read about it in a guidebook or someone might have told me. All trace of how I acquired the knowledge is lost to me. Suppose that I had been told 11 and it is this that was my warrant for 11, about both of which I have no recollection. All that I remember is 11. Despite having forgotten how I learned 11, it still qualifies as conversational knowledge, since my warrant is that someone told me. This conclusion might be resisted. It could be claimed that I do not have knowledge in this case, since I am not able to give a justification for what I know. I believe that if cases of this sort were excluded, then much of what is ordinarily taken to be knowledge would have to be ruled out as genuine knowledge. As I pointed above, most of what we know we learn from others conversational interventions. And I think a moment s reflection would reveal that in most cases of this sort we have forgotten the source of our knowledge, even that the source was a conversational intervention. There is another way to resist the conclusion that conversational knowledge does not require someone s being able to give a justification for his knowledge. The example above turns on my having forgotten the source of my knowledge, and hence, not being able currently to give a justification for my belief. This does not rule out, however, revising the requirement in 9 so that it applies to some time or other. We have, then, for 9: 12. H has conversational knowledge of p only if there is some time at which if H were asked how he knows that p, he would be able to provide as a justification for his belief that p that someone Ø-ed that p. The thought is that I have conversational knowledge of 11 only if at sometime or other I am able to say, if queried about how I know it, something like,

9 Conversation, Epistemology and Norms Someone told me that it was made of red brick. My current warrant for 11 would then be that in the past I was able to give a justification for my believing it and this justification is preserved as a causally connected tag to my belief, although I am not currently able to invoke it. This, however, seems implausible. There is a stage in the development of children when they have mastered a suitable enough amount of a language for them to have conversational knowledge without their having mastered the practice of being able to justify what they thereby know. We do not deny children knowledge because of their inability to give justifications for what they come to know through conversations. Imagine for a moment someone taking a hard line on this and claiming that if a child is not able to give a justification, then he has no knowledge in these cases. This would, I believe, rule out many clear examples of what we regard to be knowledge. I know, for example, that my name is Steven Davis, but I probably learned my name at an age at which I was unable to say, if asked then, how I know it. 12 Imagine further that in the intervening years, no one calls me Steven, but rather they use Sonny, a nickname. So there is no ongoing use of my name that would provide me with a warrant for my belief that my name is Steven Davis. 13 Let us suppose that someone were to deny that I then knew my name. It would follow from this that I do not now know my name, since currently, there is nothing I can say in answer to the question, How do you know it? and when I learned my name, there was nothing I could then say in answer to the question. What this shows is that someone s being able currently or in the past to give a justification for believing p is not a necessary condition for his conversational knowledge that p and hence, that 9 and 12 are false. Since there is a justification for my believing 11, this calls for a distinction to be made between my being able to give a justification and my having a justification. To distinguish the two I shall continue to use justification for the former, but following Burge, I shall call the latter entitlement (1993, p. 458). Before turning to a consideration of what entitles someone to believe what they understand, I would like to return to a question I raised earlier, namely, whether conversational knowledge is inferential. Let us assume that there is 12 Someone might think that this is a special kind of knowledge, knowledge of language, and it is different in kind from the sort of knowledge that I am considering. There is no reason, however, to think that my knowing that my name is Steven Davis or the knowledge that I have of anyone else s name is different in kind from the knowledge that we can have through conversation. 13 I have chosen an example from childhood, but I could have as well chosen knowledge that I have acquired as an adult through conversations. Much of this knowledge, I believe, is not mediated by my being aware of what sort of conversational intervention is the source of my knowledge. I might not pay any attention to how the beliefs are delivered to me. Was it a speech act? Or a conversational implicature? But to show there are such cases would take a psychological experiment and is not something that could be based merely on intuitions about conversations.

10 522 S. Davis an inference, albeit unconscious, from what I have been told to my believing 11. This is not sufficient, however, to show that my knowledge is inferential. What is required is that the inference must play a role in warranting the knowledge. So the question is not whether inference plays a role in delivering 11. For the sake of the argument, we shall assume that it does. It is whether the inference to 11 is part of its warrant. It does not follow from the fact that inference plays a role in delivering a belief that it serves to warrant the belief. Let us consider a similar case. Memory plays a role in long mathematical proofs. In a proof of this sort, the role that it plays is to preserve the various steps in the proof, but it does not serve to warrant the conclusion given the premises. It is the proof that provides the warrant (Burge, 1993, p. 463). In my believing, the inference does not play a warranting role; it is my understanding what I have been told about Kensington Palace that plays this role. 14 It might be thought that inference plays a warranting role in a hearer s knowing, for example, that someone who makes a promise places himself under a prima facie obligation or that someone who congratulates someone gives out that he thinks that the person he congratulates has done something commendable. In both these cases, I would argue that what is involved is understanding. For the hearer to understand that a speaker has made a promise he must understand that the speaker has placed himself under a prima facie obligation and for the hearer to understand that the speaker has congratulated someone he must understand that the speaker has given out that he thinks that the person he congratulated has done something commendable. Hence, in both cases understanding the speaker s conversational intervention involves understanding the felicity conditions that apply to the speech acts that the speaker performs. It might well be the case that in both instances inference is involved in the understanding. But it does not follow from this that the hearer makes an inference that is involved in his entitlement to believe that the speaker has placed himself under a prima facie obligation or that he gives out that he thinks that the person he has congratulated has done something commendable. I would now like to turn to what entitles us to rely on the word of others. I believe that comparing perceptual knowledge with conversational knowledge can illuminate what is required. Why can we rely on the deliverance of our senses for knowledge? Suppose that I believe that: 14. I see a cup. What entitles me to believe 14? It is because I am entitled to rely on the 14 This might seem to run counter to my claim above about children. But it doesn t. Children understand what is said to them without its being the case that they are able to say what they understand. What this shows is that the KK-principle is false, since there are cases in which children know what they told without their knowing that they know.

11 Conversation, Epistemology and Norms 523 deliverance of my senses. I am so entitled because there is a reliable causal connection between the way the world is and belief formation mediated by our sensory apparatus, a causal connection about which I might well be totally unaware. Not only might I not know this, but also everyone in my community might be ignorant about the causal connection and about the operations of our perceptual apparatus. This was certainly the case early on in human history and is still the case, I imagine, for most people. Lack of such knowledge is no impediment to having perceptual knowledge. Our ancestors, although they lacked such information, surely had perceptual knowledge. This is analogous to conversational knowledge. To have conversational knowledge, one need not be in a position, I shall argue, to know what the grounds are for our conversational knowledge. In the case of perceptual knowledge, it is the reliability of a certain causal mechanism that provides the grounds for our knowledge, a causal mechanism about which we might be ignorant. As we shall see, in the case of conversational knowledge, it is not the reliability of a causal mechanism that grounds our knowledge, but the source of our conversational knowledge, a rational and moral source that I shall claim provides the grounding and about which we might be ignorant. It might be thought that the existence of a reliable causal connection between the way the world is that is delivered to us by sense experience and the formation of perceptual beliefs does not entitle us to these beliefs. Although we might well be entitled to rely on our senses, it does not follow that the seeming deliverance of the senses is free from the possibility of error (Burge, 1993, p. 470). There are, after all, hallucinations and illusions. But this has no bearing on our entitlements. To be entitled to rely on what we take to be sensory input, we do not have to rule out the possibility of error. For if this were necessary, given the range of skeptical arguments that have turned on this possibility, we would not have perceptual knowledge. All that is required is that there is no good reason to think that we are currently in error for us to be entitled to rely on what we take to be sensory experience. It is not that we have to have such a reason. Again, we do not have to assay the perceptual state that we are in and to rule out the possibility of error. Given a perceptual experience, our having knowledge of what it delivers is the default. Similarly, we have conversational knowledge despite the possibility of lying and insincerity, but we do not have to rule out that these possibilities obtain to have such knowledge. What this shows is that both our perceptual and conversational knowledge claims are defeasible, but that they are defeasible does not show that they are false. Entitlement to rely on the word of others is similar, although the grounds are not the reliability of some causal mechanism. Let us imagine that Moses goes up the mountain and instead of hearing God s voice, he meets someone on the mountain who introduces himself as Bill. Bill tells Moses that Israel is a land of milk and honey and that he should take his people there. Moses descends the mountain believing that:

12 524 S. Davis 15. Israel is a land of milk and honey. Is Moses entitled to believe this? It turns out that Bill is omniscient and a compulsive truth teller and thus, would not lie or mislead Moses. Let us suppose that Moses is ignorant of the talents of his source. It would seem that since Moses source is Bill, he is entitled to believe 15, although he does not know the nature of his source and thus, does not know that he is entitled to believe it. Of course, the example does not tell us about ordinary conversational interventions. Normally, our interlocutors are not omniscient and compulsive truth tellers. Let us weaken Bill s powers. Suppose that Bill is an honest man who is cognitively prudent. He would not tell anyone anything he did not believe for which he did not have good evidence. Bill tells Moses 15 and that it is true, but Moses does not know about Bill s virtues. Despite this, Moses believes what Bill tells him. In this case, Moses would be entitled to rely on what Bill told him without knowing that he is so entitled. Of course, Bill can make mistakes and even honest people can lie. If this were the case, Moses would not be entitled to believe what Bill tells him. But Moses does not have to rule out that this possibility obtains to be entitled to rely on what he has been told. So, we can say that Moses has a prima facie entitlement to believe what Bill tells him, an entitlement that can be overridden. The point of the examples is to show that if the source of a belief is of a certain sort, then we are entitled to the belief, even though we might be ignorant of the epistemic nature of the source. Let us suppose that S Ø s [p]. 16. If H understands that S Ø s [p] and thereby, comes to believe q, then H is prima facie entitled to believe q. 15 In 16, q is identical to p or is conceptually related to Ø-ing [p] gives sufficient conditions for entitlement to believe. Are they necessary as well? Let us change the case and imagine that Moses does not believe what he has been told; he thinks Bill is just an old fool. What would we say about Moses, if we knew that Bill was omniscient and a compulsive truth teller? I think we would say that Moses ought to have believed and that it was a mistake on his part not to believe it. It was, after all, someone who is omniscient and cannot prevent himself from telling the truth who told him. Hence, Moses was entitled to believe Bill. He has this entitlement even if he 15 Ø s [p] is meant to capture speech acts such as asking whether it is raining or congratulating someone for their fine performance. [p] then is a placeholder for whatever is the appropriate sentential complement for any Ø, where Ø is a placeholder for a conversational intervention type. 16 p and q are identical in the example that turns on 1 and 2 and they are conceptually related if in the same example, Alice were to come to know that Sam believes 1. In the latter case Sam s believing 1 is related conceptually to 2, since his believing 1 is a felicity condition for 2.

13 Conversation, Epistemology and Norms 525 did not believe what he was told. This shows that although sufficient for entitlement, the conditions are not necessary does not capture the full range of beliefs to which we are entitled through conversational interventions. Let us consider the following case. Sam is a private in the army. Bill, who is not an officer and is not in any position to give Sam an order, is wearing an officer s uniform and he looks and acts like an officer. He says to Sam with an authoritative voice, 17. Sam, I order you to clean up the barracks. Sam has no reason not to think that Bill is not an officer; in fact, given the uniform, Bill s manner and his tone of voice, Sam takes 17 to be an order. Because of this, Sam comes to believe: 18. I am under an obligation to clean up the barracks. 18 Is he entitled to believe it? He is, if he is entitled to believe that Bill has ordered him to clean up the barracks. Given the way the situation is described, Sam is entitled to take Bill to have ordered him to clean up the barracks. Hence, he is entitled to believe that he is under an obligation to clean up the barracks. But since Bill is not in a position to order Sam to clean up the barracks, it is not the case that Sam understands that Bill has ordered him to clean up the barracks. For if Sam had understood that it was an order, it would have been an order. Hence, we can get by with a weaker condition than understanding. In our example, Sam only seemed to understand that Bill had ordered him to clean up the barracks. Given the circumstance, this entitles Sam to believe that Bill ordered him to clean up the barracks. This in turn entitles Sam to believe that he is under an obligation to clean up the barracks. Hence, Bill s seeming to understand that it was an order is sufficient for Bill to be entitled to believe that he is under an obligation to clean up the barracks. We can revise 16 to take account of this. 17 The example might seem to count against what Burge (1997, p. 45 n. 4.) calls The Acceptance Principle. A person is [a priori] entitled to accept a proposition that is [taken to be] presented as true and that is [seemingly] intelligible to him, unless there are stronger reasons not to do so (Brackets in original, ibid.). Whether the example counts against the Acceptance principle depends on how there are stronger reasons not to do so is construed. If it is sufficient for there to be stronger reasons that someone believes that there are, then we have a counter example to the Acceptance principle. But there is no counter example, if it is taken to mean that for there to be stronger reasons for not believing, the reasons must actually count against the truth of the belief. Although Moses believes that Bill is an old fool, he is in fact no such thing. Hence, what Moses believes is not a reason for his not being entitled to believe Notice that if in fact Bill were an officer, Sam would have conversational knowledge of 18.

14 526 S. Davis 19. If H seems to understand that S Ø s [p] and thereby, comes to believe q, thenh is prima facie entitled to believe q. 19 Applying 19 to the case under consideration yields, 20. If Sam seems to understand that Bill ordered him to clean up the barracks and Sam comes, thereby, to believe that he is under an obligation to clean up the barracks, then Sam is prima facie entitled to believe that he is under an obligation to clean up the barracks. There are then two steps in the argument that must be filled out before we can establish what it is in cases like this that entitles one to believe what one comes to believe through interlocution. 20 First, it must be established that, 21. If H seems to understand that S Ø s [p], then H is prima facie entitled to believe that S Ø s [p]. and second, that, 22. If H is prima facie entitled to believe that S Ø s [p], then H is prima facie entitled to believe that q. 21 Let us begin with 21. The question is what entitles us to go from seeming to understand that a conversational intervention is, for example, an assertion that p to being prima facie entitled to believe that it is an assertion that p. The grounds for the entitlement are the sort of grounds that we have for understanding what is conveyed to us by conversational interventions. These are not what are at issue here, since I have assumed as an epistemic starting point that we understand or seem to understand conversational interventions and that we are entitled to believe that we understand or seem to understand them. This is not to suggest that there is no epistemic problem, since there is an epistemic gap between what we seem to understand and our being entitled to rely on what we seem to understand. But I shall not try to bridge the gap here. 22 So the problem is to show that is true. I shall begin with an account of how assertions give us a prima facie entitlement to believe what is asserted, 19 The same qualification that applies to 16 applies here: q is identical to p or is conceptually related to Ø-ing [p]. Also, see footnote 14 above and 22 are not translations of 19. Rather, they are two principles that have to be established to substantiate our entitlement to our beliefs that arise because of conversational interventions. 21 See footnote See Burge 1993 and 1997 for a discussion of this issue.

15 Conversation, Epistemology and Norms 527 relying in part on the work of Burge. 23 I shall then try to show how this account must be modified to apply to other conversational interventions. Consider 1 again. Sam says to Alice, 23. There is a party tonight. We shall assume that Alice is entitled to believe that Sam asserted and that she believes 23. What, then, is the connection between her being entitled to believe that Sam has asserted 23 and her being entitled to believe 23? The argument has two parts. First, there is a connection between someone s intelligibly asserting something and his being rational and his being rational and what he asserts being true. Second, someone who asserts something represents himself as telling the truth, that is, not lying. For this reason, he can be relied on to say what is true. Hence, the entitlement to believe what someone asserts will rest on an appeal to a prima facie presumption that the speaker is rational and moral. 25 Now for the details. Asserting is an act in which a speaker, in this case Sam, implies that he believes what he says. This is carried by one of the felicity conditions of the act, more particularly, its sincerity condition (Austin, 1962, pp ). Evidence for this is the oddness of saying, 24. There is party tonight, but I don t believe it. Hence, to assert that p is conventionally to imply that one believes that p. And to imply that one believes that p is to imply that one believes it to be true. So an assertion of p is a presentation of p as being true. Hence, an agent who asserts that p is presenting something as being true. Consequently, since Alice is prima facie entitled to believe that Sam has asserted something, she is prima facie entitled to believe that Sam has presented something as being true. Moreover, Alice has a prima facie understanding of what Sam said and thus, unless there are reasons to the contrary, she is entitled to rely on her prima facie understanding. Hence, Alice is entitled to take Sam to have intelligibly presented something as being true. The intelligible presentation of something 23 I part company with Burge in two respects. Unlike Burge, my interest is not in showing that there are a priori entitlements to believe some of what we come to believe through conversational interventions. In addition, I do not follow him in his claim that the sincerity of the interlocutor plays no fundamental role in prima facie entitlements to believe what is asserted. Burge s account can be found in his 1993 and I shall be neutral at this point about whether Alice only seems to understand that Sam asserted 23 or whether she has a genuine understanding. I have assumed that seeming to understand is sufficient for being entitled to believe that one understands. Clearly, if seeming to understanding yields entitlement, so too does genuine understanding. Moreover, genuine understanding, although required for Alice to know 23, is not necessary for her to be entitled to believe it. 25 Moral might be too strong. I consider this in what follows.

16 528 S. Davis as being true reveals two things that can be presumed about the speaker, the source of the presentation. The speaker can be presumed to be rational and moral, rational in that the intelligible presentation of something as being true, as Burge puts it, is a prima facie sign of rationality (Burge, 1993, p. 471) and moral in that the intelligible presentation of the speaker as believing what he says is a prima facie sign of his being moral. Let us assume that in seeming to understand Sam, not only is Alice entitled to believe that she has understood him, but that in fact she has understood his conversational intervention. So we shall assume that Sam has asserted 23. When we are finished our discussion, we shall unwind the assumption. Consequently, the question that we shall ask is: In what way is Sam s assertion of 23, his intelligible presentation of its being true, a sign of rationality? To answer this question, we shall look at the properties that Sam has, by virtue of his asserting 23. But first a bit about rationality. We are interested in Sam s rationality, since he is the source of Alice s belief. Rationality can be attributed to actions, abilities, capacities, dispositions, beliefs, intentions, agents, etc. There are interrelations among these. Let us look at the relationship between agent and action and the connection with rationality. There are two ways in which the rationality of the agent connects with an action. The first is that some actions require for their successful performance that the agent have various capacities and abilities that function properly. In turn, this requires that the agent be sensitive to certain norms, the sensitivity to which is partly constitutive of what it is to be rational. The second way in which actions are connected to the rationality of an agent is that the rationality of an agent is a function of the rationality of his dispositions to act and of his ability to apply these dispositions in particular situations. A disposition to act is rational, if it results in rational actions. Hence, an agent s being rational entails that when the agent performs a particular action, it will be rational. It might seem that this is obviously false, since agents who are rational can on occasion act irrationally. Certainly this is the case, but in so far as an agent acts irrationally he is not entirely rational, nor can the disposition from which such an act flows be entirely rational. But our interest is not to move downward from the rationality of an agent and his dispositions, but upward from the rationality of an action to the rationality of an agent. If an agent s act is rational, what does this say about the agent? I think that we can safely say that an agent s acting rationally in a particular situation is a sign that he has a disposition to act rational and hence, is a sign that he is rational. That an agent s acting rationally in a particular case is a sign for his being rational does not result in its being certain that he is rational, but as Burge puts it, it justifies a prima facie presumption that the agent is rational (1993, p. 476). Let us turn to Sam and his asserting 23. He has performed a speech act, containing a content, in a conversational context. The performance of the speech act indicates that he has a range of capacities, including his ability to make assertions, to enter into conversational exchanges, to apply a range of

17 Conversation, Epistemology and Norms 529 concepts in expressing the content of the assertion and to use a range of expressions in a grammatically intelligible construction that are relevant to his making his assertion. To sum this up, we can say that in asserting 23, Sam exhibits speech act, conversational, conceptual and linguistic abilities and capacities. There are connections among these capacities and abilities. The ability to make assertions, and more generally, the ability to perform speech acts, has as a necessary component the ability to understand and use a range of concepts, since a speech act requires for its performance having the concept of the speech act performed, as well as the concepts that constitute the constituents of the content expressed. 26 In our example, Sam cannot assert 23 without having the concept of an assertion and the concepts that are constituents of the content of what is expressed by his utterance of 23. Moreover, in performing a speech act and expressing the content of the act, Sam must have the linguistic means to perform the act and to express the associated content. 27 Speech acts typically occur in conversational contexts. Sam must have the capacity to address his remark to Alice, to react appropriately if uptake is not evident, to recognize the relevance of various conversational responses to his assertion, for example, Alice s telling him that she already knows 23 or that the party was last night, etc. There is a connection between these abilities and rationality. Each of these abilities, speech act, conceptual, conversational and linguistic, involves the possibility of error. They require that Sam has the capacity to recognize the relevance of remarks to the effect that he has made a mistake and the disposition to make corrections when he has made a mistake. Take having concepts. Having a concept is necessarily linked with being able to recognize correct and incorrect applications of the concept. Consider having the concept of an assertion. This is connected to the ability to recognize as being relevant criticisms to the effect that what one has asserted is false or that there are no reasons for believing it. Being able to recognize mistakes and being disposed to make corrections of the mistakes so recognized are partly constitutive of rationality. Hence, in so far as he asserts 23, Sam can be taken to be rational. 26 It could be argued that children are able to make assertions but they do not have the concept of an assertion. Having such a concept, it might be claimed, is a higher order capacity that is displayed and is necessary in understanding various sorts of remarks that are directed to the speaker who makes an assertion, for example, a critical remark to the effect that what the speaker says is false. I think that if someone who makes a statement lacks the ability to see the relevance of this and similar remarks to what he has said, then it is not clear that he has made a statement. Imagine a child who insists that his truck is broken and when shown that it is not, still insists on saying Truck, broken, and he insists on saying this no matter what he is told or shown. We might then have doubts about whether he has the concept expressed by broken and thus, whether in uttering Truck broken, he has asserted anything. 27 I do not mean to imply a tight connection between linguistic abilities and ability to perform a speech act. It might be the case that Sam has speech difficulties in which he inverts words, but in certain contexts this might not prevent him from asserting something.