Exportation, Counterfactuals and Firmness of Belief * I shall attempt to sketch an intuitive solution to the problem of exportation (the transition from de dicto to de re beliefs). My search for this solution was stimulated by Kripke's famous puzzle about belief. I address the question: under what conditions can we ascribe de re beliefs based on the evidence of de dicto beliefs? For reasons of space, I won't present the details of Kripke's well-known article, which I assume to be familiar to the reader. One might be tempted to say that exportation cannot be allowed in the case of Kripke's puzzle, since if Pierre's beliefs that Londres est jolie and London is not pretty were interpreted de re, this would lead to a contradiction1. I am opposed to this temptation. The first part of my essay criticizes one argument against exportation. The second and third parts aim at providing a constructive solution, specifying how exportation should proceed. A terminological note. Much will be written here about de dicto and de re beliefs. In most cases, when I report a belief in ordinary language I shall try to mention whether it is de dicto or de re. But in all cases I shall respect the following convention: a de dicto belief is rendered as believes that Fa, while a de re belief is rendered as believes about a that it is F. I. Pierre in Kripke's London-Londres example is not irrational, yet he assents to two sentences that, unbeknownst to him, seem to contradict each other, namely that 'London is not pretty' and 'Londres est jolie'. If we ascribed to him, on this basis, two de re beliefs, the city London-Londres would be such that Pierre believes it is pretty and that Pierre believes it is not pretty. This is impossible, if Pierre is rational. We might be tempted to conclude that Pierre does not believe anything about the city which is named London-Londres: his beliefs are de dicto and not de re. A related reason for this conclusion is that when he reaches the city London, he cannot recognize it as Londres, so he is not familiar enough with it. This answer is wrong for a simple reason. In Kripke's story, Pierre started to believe that London is not pretty at a definite moment of time (call it t), long after he had started to believe that Londres est jolie. But we have all the reasons to hold that before * I thank Teodor Gugiu from the University of Bucharest and, of course, the anonymous referees of its first version for their valuable suggestions concerning this paper. This is the solution advanced by Patricia Hanna (2001), who rejects Kripke’s Principle of Translation (Kripke 1988, p. 114). An opposite direction is pursued by Joseph Moore (1999). I do not intend to reject the usefulness either of the Translation Principle or of the Disquotation Principle; I shall only hold that none of them is universally acceptable. 1 1 moment t, Pierre really believed, or at least might have believed, about LondonLondres that it was pretty – at least if we are ready to ascribe the de re belief about London that it is F to any competent speaker of French who assents to the French sentence that Londres est F and knows about London everything that Pierre does (for example, that Londres est la capitale d'Angleterre, that it lies on such-and-such coordinates on the map), even if she has never seen London2. If we denied to Pierrebefore-t the (de re) belief about London, then we should deny such a belief to all Frenchmen who had various beliefs that F(Londres) and are as linguistically competent and knowledgeable as Pierre. But it is counterintuitive to deny such a belief to all these Frenchmen. I suppose that most persons who heard of Jack the Ripper believe that he was a detestable person and they had a gentle neighbour on their floor, whom they believed to be admirable for his benevolence. But how can they know that Jack the Ripper dis not hide exactly on their floor, disguised as the gentle neighbour of whom they believed was admirable for his benevolence? Can any of us honestly claim to have a 100% guarantee that she or he is not and will never be in such a situation? No, unless one knows and would know in reasonable details the biography of all one's neighbours. And still, this is not a reason to deny that most people of London believed about Jack the Ripper that he was detestable. Even if he had been disguised as the gentle neighbour whose benevolence was admired by everybody, they would still have believed Jack the Ripper to be detestable, even during the time when they believed that the gentle neighbour was admirable for his benevolence3. Therefore: the mere fact that one has a de dicto belief that Fa and is unable to recognize a as the reference of 'a' when acquainted with a does not preclude one from having a de re belief about a.4 (I have never been to the South Pole and probably would not recognize it as the South Pole if I were abducted in sleep and suddenly woke up there; nevertheless I believe about the South Pole that it is a very cold place, and that it was discovered by Amundsen. If this were denied, it is hard to see with what de re beliefs I could be credited at all). Therefore it is unacceptable to refuse to ascribe beliefs about London to all the Frenchmen who have various beliefs expressed in French by Londres est F, even if they never saw London and who would not have recognized London if they saw it. But from all we know about Pierre, he is one of these Frenchmen. Therefore: Pierre-before-t might have had a belief about London. (If he didn’t have it, the reason is not that he was, and perhaps would have been, unable to recognize the city of London as the reference of 'Londres'). Nothing prevents us, in principle, from ascribing to Pierre-before-t the de re belief about the city named London and Londres. I follow here Kripke himself: "It seems undeniable that Pierre once believed that London is pretty – at least before he learnt English. For at that time, he differed not at all from countless number of his countrymen, and we would have exactly the same grounds to say of him, as of any of them, that he believes that London is pretty" (Kripke 1988, p. 120) 2 3 It can be replied that “Jack the Ripper” is not really a proper name but a description (after all, the real name of Jack the Ripper has never been discovered). But for the purposes of the current example, we may imagine that “Jack the Ripper” was his real name, or that his real name was known. 4 This does not reject, but rather qualifies what Evans calls Russell's Principle: "It is not possible for a person to have a thought about something unless he knows which particular individual in the world he is thinking about" (Evans, p.44) 2 I emphasize that I do not imply that all those who believe de dicto that Fa and are unable to recognize a as the reference of 'a' when acquainted with a have a de re belief about a. I only want to point that they may have such a de re belief. If we want to exclude some beliefs that Fa from the club of beliefs worthy of exportation, the mere fact that the believer could not have recognized the object a is not a sufficient reason. Can we say that it is only after t, when Pierre started to believe that London is not pretty, that exportation is not allowed from any of his beliefs involving names of London-Londres? If we say this, it follows that a de re belief ceased to be de re once a(nother) de dicto belief was acquired. Pierre's acquiring the (de dicto) belief that London is not pretty had the power of cutting the bridge between the real object London-Londres and the name Londres as it occurred in Pierre's previous belief expressed by Londres est jolie. A de dicto belief caused a de re belief to stop being de re and to become de dicto, it 'unexported' it. I shall argue in the next lines why this cannot be a correct solution. If we allow the phenomenon of unexportation somewhere, we risk losing control of it. How can we be sure we are not in the situation of Pierre before t? Given a name 'a' that I believe refers to an object a and which occurs in a de dicto belief of mine that Fa, how can I be sure that I won't learn another name, 'b', of it, in another language, such that I won't recognize that a=b and I will believe that ~Fb? This question is different from the skeptical question “How can we be sure that we are not, all of us, in Pierre's situation?” Even if (as I shall argue) we are right in believing that we are not in the situation of Pierre-after-t, at least not most of the time, with respect to most of our beliefs, there is nothing that keeps any of us from coming to be in Pierre's situation in the future. If we find ourselves in Pierre's situation, it is plausible that, like Pierre, we won’t recognize it on the spot. If the possibility of unexportation is allowed, then for any belief of mine that Fa it is, in principle, possible to discover that it is no longer about a; that it stops being de re and becomes de dicto. The situation may be generalized. I believe (like most of those who know English and have rudimentary notions of history) that Napoleon was short: this is a firm conviction of mine, based on everything I read about Napoleon. But nothing logically prevents the possibility that I will learn some exotic language, spoken in a former French colony, in which Napoleon is called by a name unrecognizable to us as 'Napoleon', for example ‘Numburburu’. Let's assume that all speakers of that exotic language know all true things about Napoleon: that he was the first French emperor, who won the battle of Austerlitz etc, except for the information that he was short. They believe that Numburburu was tall (they cannot conceive that French emperors could be short). If I learn their language and adopt their beliefs about Numburburu, as expressed in their language, my previous belief that 'Napoleon was short' will cease – if unexportation is allowed - being about Napoleon. (This belief will be unexportated by the belief containing the name 'Numburburu'). If I come to be in such a situation, I will be with respect to Napoleon-Numburburu like Kripke's Pierre with respect to LondonLondres. We can assume that if I am in such a situation I have the means to discover it and correct my beliefs; but even so, I shall not be able to diagnose such a situation immediately. And during the interval between the time at which I acquire my belief that Numburburu was tall and the time at which I diagnose my epistemic situation, I will be having no belief about Napoleon-Numburburu at all. If so, then I am in the 3 danger of discovering that my current beliefs about Napoleon will have ceased to be about Napoleon at all. But what happens to ‘Napoleon’ may happen to any other name which occurs in any sentence I believe. If it is allowed that such a phenomenon can happen to all our beliefs at the same time (and there is prima facie no reason why this cannot be allowed), then I risk discovering that I will have stopped, for some time, having beliefs about anything at all! I think we should respect the intuition that a de re belief cannot become de dicto: if a belief is de dicto, it has been so during all its lifetime5. But if this intuition is combined with the thesis that exportation is not admissible from any of Pierre's beliefs, then both of Pierre's beliefs were de dicto from the moment each of them was acquired. The situation is now even more dramatic. Since it is – as I showed in the above paragraph – possible that everyone of us will some time be in Pierre's situation and not recognize it immediately – none of us has any de re belief. We do not risk that all our de re beliefs should be unexported, but risk instead to discover that none of them is ever, has ever been, de re! If none of Pierre's beliefs could be interpreted de re, then no human belief could safely be interpreted de re. This conclusion must surely be rejected. II. If we have enough time to acquire empirical information, we can arguably hope to recognize whether we are in the situation of Kripke's Pierre. All we need to perform is a two-stepped task: (1) to find all pairs of de dicto beliefs, (in different and in the same language6), that contain contradictory predicates and then (2) to check whether the names contained in these beliefs are coreferential or not. Scrupulously to complete this task is within the intellectual power of anyone who does not have an infinite number of beliefs and is linguistically competent. Let's admit that we do not have an infinite number of beliefs - at least not of explicit beliefs - and that we can complete this task. But to complete it takes time. Therefore nothing guarantees that we won't be, for some time, in Pierre’s situation. And if Pierre's situation is one in which two beliefs that seemed de re turn out to be actually de dicto, and (as argued above) they have always been de dicto, we cannot guarantee for any of our beliefs that it is de re. If we want to have any de re belief at all, we have no reason to deny this privilege to Pierre. This is, in my opinion, a sufficient reason to say that exportation is allowed in Pierre's case. If both of Pierre's beliefs were de re, he would contradict himself. But we want Pierre to be, and to have been, a rational person. Therefore, at most one of Pierre's Londonese beliefs may be exportated. The obvious question is: which one? Let's pay attention to the following three counterfactuals, only one of which may be true: (1) If Pierre had known that London is Londres, he would have maintained his belief that Londres est jolie, would have abandoned his previous belief that London is not pretty and assented to the English sentence London is pretty. 5 It can be replaced by a de dicto belief, but this is another issue See Paderewski’s case (Kripke p. 130) 6 4 (2) If Pierre had known that London is Londres, he would have maintained his belief that London is not pretty, would have abandoned his previous belief that Londres est jolie and assented to the French sentence Londres n'est pas jolie. (3) If Pierre had known that London is Londres, he would have abandoned both his beliefs and suspended judgement; he would have said in English I don't believe either that London is pretty or not pretty and in French Je ne crois ni que Londres est jolie ni que Londres n'est pas jolie7. These are all three possibilities counterfactually open to Pierre if he is rational and these three counterfactuals are mutually exclusive: one of them must be true if Pierre is rational. I hold that exportation depends on which of the above three counterfactuals is true in the following way: If (1) is true, then exportation proceeds from the 'French belief': Pierre believed about London-Londres that it was pretty. Exportation is not allowed from the English belief. If (2) is true, then exportation proceeds from the 'English belief': Pierre believed about London-Londres that it was not pretty. Exportation is not allowed from the French belief If (3) is true, then exportation does not proceed from either belief. Pierre did not believe anything at all about London-Londres. This seems to me the most reasonable solution that allows exportation, allows Pierre to be rational and avoids the total skepticism that threatened us if we did not allow exportation. To make this thesis more intuitive, let's remember the example of Jack the Ripper and the gentle neighbour. I held that most of those who heard about Jack the Ripper still believed about Jack the Ripper that he was detestable even during the time when they believed that their gentle neighbour (who was actually Jack the Ripper in disguise) was admirable.8 In this example, we don’t believe about our neighbour that he is admirable (de re), although we believe (de dicto) that our neighbour is admirable9 and, moreover, we can recognize our neighbour. 7 It is possible that Pierre, once he learns that London=Londres, acquires two new beliefs: that the neighbourhood of London in which he lives is not pretty and other neighbourhoods, seen in pictures while he was in France, are pretty. This assumes (3) 8 This example proves that a believer may believe that Fa and be able to recognize a as the reference of 'a' and still fail to believe anything de re about a. So having the de dicto belief and being able to recognize the relevant object is not a necessary and not a sufficient condition for having the de re belief To avoid complications about the status of the definite description “my neighbour’, we might imagine that Jack the Ripper lives next door to you under a false name and you know him under that name 9 5 The reason is that if we had known that Jack the Ripper was the same as the gentle neighbour, we would have stopped believing that the gentle neighbour was admirable and would have started to believe instead that the gentle neighbour was detestable. Thus, we can formulate the desired criterion for exportation from beliefs that Fa where 'a' refers to a real object a: a belief de dicto that Fa deserves to be exportated if for any belief ~Fb we could have, if we discovered that a=b, then we would continue to believe that Fa and abandon the belief that ~Fb in favour of the belief that Fb10. True, we are still in the danger of discovering that some of what we took to be our de re beliefs were actually de dicto; since we may become or even be in Pierre's situation. Indeed Pierre, if he learns that London is Londres, will discover that some of what he took to be de re beliefs were actually de dicto .But my solution has a note of optimism in that this danger is unlikely to concern all one's beliefs11. In other words, it is, most likely, not the case that it could be discovered that all the beliefs that one took to be de re were actually de dicto. Such an unpleasant discovery would be made only if I am ready to suspend judgment as to whether Fa or ~Fb, whenever a and b are discovered to be coreferential. But I think that for most of our de dicto beliefs that Fa it is (1) either the case that we would continue to believe that Fa even if we discovered that we believed that ~Fb and a=b; or that (2) we would stop believing that Fa and would replace this belief with the belief that ~Fa once we discovered that we also held the belief that ~Fb and that a=b. III. It remains to present a method to assess the truth-value of these counterfactuals. One could think that the happiest situation is that in which Pierre, at some time u, actually learns that London is Londres and modifies one of his beliefs accordingly. If this happens, this should, so it seems, allow us to ascribe to him, retroactively, the belief about London that it was pretty or the belief that it was not pretty. But this is wrong. We cannot assume that if any moment u-n it is true that p>q12, then at the later moment u it will be false that p & ~q. For the following situation might obtain: at moment u-n Pierre would have abandoned the “French belief” if he had learnt that London is Londres, then at u he learns that London is Londres and abandons the “English belief” while preserving the French belief. A counterfactual p>q (with a false antecedent) may be true even if p becomes actual at some later time when q is false. Such situations are perhaps rare, but they are nevertheless possible. To think otherwise implies that the truth value of a counterfactual varies in time. Therefore: when trying to find what Pierre believed about London-Londres, we should be sensitive to what Pierre would have believed if he had known that London is Londres, 10 The solution to the Paderewski puzzle (Kripke 1988, pp. 130-131) is similar: If Peter had learnt that the two Paderewskis are one and the same, he would either have started to believe that Paderewski-thepolitician has musical talent, or (if his conviction that politicians cannot have musical talent is very strong), that Paderewski-the-pianist has no musical talent (since he is a politician), or would have suspended judgement. Exportation works as I suggested above 11 I hope the last section of my paper will show why it cannot concern all one's beliefs 12 I abbreviate the counterfactual if p had been the case, then q would have been the case as p>q 6 but should be rather insensitive to what he believes if he actually learns that London is Londres. The empirical test that I propose is that of the firmness of one's de dicto beliefs. This test can work only if the concept of firmness of belief is not understood counterfactually, i.e. if we have a method of testing the firmness of a belief independently of talk about the circumstances in which that belief would be preserved or abandoned. Let’s assume that it is a primitive psychological fact that some beliefs are stronger than others, and the strong belief that Fa would continue to be held even if the believer discovered that she had believed that ~Fb and that a=b. Firm beliefs are, counterfactually, difficult to remove. Firmness of belief is a psychological concept, analogous to the logical concept of necessity. Firm beliefs are, from a psychological point of view, what necessary truths are from a logical point of view: as necessary truths are true in all possible words, so firm beliefs are held by a believer in all, or in almost all, counterfactual epistemic situations. (An extremely firm belief of Jane is a belief that Jane would hold no matter what happened, no matter how her other beliefs changed. Less firm beliefs of Jane are the beliefs that she would hold in almost all counterfactual epistemic situations: i.e., if the belief that Fa is a firm belief of Jane, then there are very few beliefs that Jane may acquire or lose so that she would stop believing that Fa. Necessary truths make true any counterfactual in which they occupy the position of the consequent: If q is a necessary truth, then for any p, it is the case that if p had been the case, then q would (still) have been the case. Firm beliefs make true almost any counterfactual in which they occupy the position of the consequent: if q is a firm belief of Jane, then for almost any p, it is true that if Jane had believed that p, she would still have believed that q. I italicized the word "almost" in order to highlight that firm beliefs are still subject to potential reversal: only an extremely firm belief is such that its believer would hold it true no matter what else he believed. The idea that is present behind this counterfactual solution to Kripke's puzzle reposes on the fact that Pierre faces a potential conflict between his "English belief" and his "French belief". These beliefs do not commit Pierre to a contradiction if they are de dicto, but they do so if they are both de re. Whenever two de dicto beliefs threaten to contradict each other if they are interpreted de re, i.e. if exportation is allowed, there are, so to speak, two opposite tendencies of exportation. If these two tendencies met each other, an actual, explicit conflict between the two beliefs would occur and this conflict would result in the revision of at least one belief - obviously, under assumptions of rationality. This conflict between beliefs ends with the victory of the more intensely held belief; moreover, it would have ended with the victory of the firmer belief, if the firmness of the two beliefs had been compared. Exportation goes hand in hand with a metric of the firmness of belief. When someone believes (de dicto) that Fa and ~Fb, ignoring that a=b, his beliefs are in most cases not of equal firmness. The less firm belief survives in the person's system of beliefs only because the object that is the reference of ‘a’ and of ‘b’ is not recognized as being both the bearer of ‘a’ and of ‘b’. If it were so recognized, then the less firm belief would be abandoned. De re beliefs are thus (relatively) firm de dicto beliefs. The reason why I introduced the concept of firmness of belief is that it seems more intuitive to compare the firmness of two beliefs than to assess whether a de dicto 7 belief may rightly be interpreted de re. Additionally, firmness of belief is a parameter that allows us to test the belief's counterfactual power to eliminate other beliefs. If - as I assumed - the relative firmness of a belief may be measured in the absence of an actual conflict between two de dicto beliefs, making exportation dependent on it is a useful, non-circular test. I shall end by making some suggestions for expanding the counterfactual solution to the problem of exportation. These suggestions are perhaps speculative, but I think that they are based on a valid intuition: that the distinction between de re and de dicto beliefs is related to the distinction between believing that p and believing that p is true¸ respectively. Let’s try to see what it would mean to generalize Kripke's puzzle to entire sentences. Imagine that I become the disciple of an Indian sect and take the oath to believe everything that my guru says. My guru speaks only in Hindi. He utters repeatedly, on an authoritarian voice, the sentence Surya dharti13. After he becomes convinced that I really trust him, he grants me a scholarship to India, where I meet people with a rudimentary knowledge of English. They ask me Surya dharti true? and I nod. Unbeknownst to me, the meaning of Surya dharti was actually "the Sun revolves around the Earth". (Perhaps there were no Hindi words in it corresponding to the English words sun, to revolve and Earth but this does not affect this mental experiment). In such a case, my Indian interlocutors will be right in ascribing to me the belief that Surya dharti is true, although they won't be right in ascribing to me the belief that the Sun revolves around the Earth: indeed, my conversion to the guru's sect was not so deep as to change my firm belief that the Earth revolves around the Sun, and not vice versa. The reason why I cannot be said to believe that the Sun revolves around the Earth is the counterfactual: If I had known that Surya dharti means that the Sun revolves around the Earth, then I would have no longer believed that Surya dharti is true and - if I had learnt Hindi - I would have expressed my belief by the contradictory (in Hindi) of Surya dharti. And the truth of this counterfactual is exactly what we wanted to say when we said that my belief that the Earth revolves around the Sun is a very firm belief, and remained so even after my conversion to the guru's sect. On the contrary, if my belief that the Earth revolves around the Sun had been less firm, and my conversion to the guru's sect had had more profound effects, then the following counterfactual would have been true: If I had learnt that Surya dharti means that the Sun revolves around the Earth, I would have abandoned my belief that the Earth revolves around the Sun and would have replaced it with the belief that the Sun revolves around the Earth, as expressed in English. This situation is a generalization of Kripke's puzzle about belief. According to one's metaphysical sympathies, one may define the analogous of exportation as the transition from sentences to propositions, or from beliefs that sentences are true to beliefs that facts are so-and-so. This transition corresponds to what we usually call understanding a sentence. Indeed, my merely assenting to p does not guarantee that I understand what p means. It is however psychologically implausible to have very firm 13 I am sorry for not having been able to find the correct translation in Hindi of the sentence “The Earth revolves around the Sun”. I only found that “surya” means “Sun” and “dharti” means “Earth” 8 beliefs that we don’t understand. Aldous Huxley imagined in his Brave New World the experiment of hypnopaedia - education-through-sleep, in which children are conditioned to learn various sentences during their sleep. This had the undesired result that they learnt no more than sentences: their beliefs were, so to speak, purely Quinean14, without involving any attitude to the components of the sentences believed (and obviously without allowing existential quantification or exportation): ‘The Nile is the longest river in Africa and the second in length of all the rivers of the globe. Although falling short of the length of Mississippi-Missouri, the Nile is at the head of all rivers as regards the length of its basin, […]’ At breakfast the next morning, ‘Tommy,’ someone says, ‘do you know which is the longest river in Africa?’ A shaking of the head. ‘But don’t you remember something that begins: The Nile is the…’ ‘The- Nile-is-the-longest-river-in-Africa-and-the-second-in-length-of-all-the-rivers-of-the-globe…’ The words come rushing out: ‘Although-falling-short-of…’ ‘Well now, which is the longest river in Africa?’ The eyes are blank. ‘I don’t know.’ ‘But the Nile, Tommy.’ ‘The- Nile-is-the-longest-river-in-Africa-and-the-second…’ ‘Then which river is the longest, Tommy?’ Tommy bursts into tears. ‘I don’t know,’ he howls (Huxley, 1955 p. 20-21) By modifying a little Huxley's example, we may imagine that a hypnopaedically conditioned child learns the sentence 'The-Amazon-is-the-longest-river' and, based on the information he collects from other sources, he later believes that the Nile is the longest river and Amazon is the second longest - this without abandoning his hypnopaedically-formed belief. To the question 'What does he believe about the Amazon?', the answer is, again, dependent on the truth value of the three counterfactuals that may be formed (he would have abandoned one of his beliefs or both) and these counterfactuals reflect again the comparative firmness of his beliefs in our case, they reflect how strong his hypnopaedic indoctrination proved. Exportation is allowed from firm beliefs. We risk discovering that we don’t have beliefs about anything if we have weak beliefs. And we risked discovering that we will only have beliefs about nothing if we could not replace our weak beliefs with stronger beliefs. References: Gareth Evans, The Varieties of Reference, Oxford, the Clarendon Press 1982 Patricia Hanna- Linguistic Competence and Kripke's Puzzle, in Philosophia, June 2001; 28(1-4): 171-189 Aldous Huxley – Brave New World, Harmondsworth, Middlesex : Penguin Boks, 1955 Saul Kripke – A Puzzle about Belief, in Nathan Salmon and Scott Soames, Propositions and Attitudes, Oxford University Press 1988, pp. 192-149, originally 14 9 At least according to Quine[1956] published in Meaning and Use, ed. A. Margalit (Dodrecht: D. Reidel, 1979, p. 23983) Joseph Moore – Misdisquotation and Substutivity: When not to Infer Belief from Assent (Mind, vol. 108, 430-April 1999) W.V.O. Quine - Quantifiers and Propositional Attitudes, The Journal of Philosophy, 53 (1956), pp. 177-187 10