ANTI‐INDIVIDUALISM: MIND AND LANGUAGE, KNOWLEDGE AND JUSTIFICATION
2009; Wiley; Volume: 50; Issue: 2 Linguagem: Inglês
10.1111/j.1468-0149.2009.00480.x
ISSN1468-0149
Autores Tópico(s)Hermeneutics and Narrative Identity
ResumoIn Anti-Individualism,11 S.C. Goldberg, Anti-Individualism: Mind and Language, Knowledge and Justification (Cambridge University Press, 2007, xiii + 265 pp. £42.75). Sanford Goldberg presents a theory of testimony and testimonial knowledge, and considers the implications of this theory for philosophy of language, philosophy of mind, and epistemology. According to Goldberg, plausible assumptions about testimonial knowledge provide important motivation for the view that linguistic meaning is not determined exclusively by purely internal states and processes, and that the same is true of the representational content of propositional attitudes. Instead of being determined by purely internal factors, such as the states and processes that are considered in cognitive neuroscience, or the states and processes that are studied by computational psychology, meaning and content depend, in part, on relations between individual agents and the linguistic communities that they inhabit. In short, according to Goldberg, considerations pertaining to testimonial knowledge provide motivation for semantic externalism. He also maintains that this motivation is largely independent of the arguments for externalism that are familiar from the writings of Kripke, Putnam, and Burge.22 S.A. Kripke, Naming and Necessity (Harvard University Press, 1980); H. Putnam, “The Meaning of ‘Meaning’,” reprinted in H. Putnam (ed.), Mind, Language, and Knowledge: Philosophical Papers, Vol. II, (Cambridge University Press, 1975), pp. 215–71; Burge, T., “ Individualism and the Mental”, Midwest Studies in Philosophy, 4 (1979), pp. 73– 121. The line of thought leading to these conclusions occupies the first half of the book. In the second half, Goldberg turns his attention to epistemological issues. What is it, he asks, for a testimonial belief to count as epistemically justified, and what is it for a testimonial belief to count as epistemically rational? He argues that these epistemic properties involve reliability in at least two ways. First, in order for testimonial beliefs to exemplify the properties, the relevant believers must be equipped with, and make use of, a reliable ability to determine whether testimony is called into question by defeaters—that is, by considerations that challenge testimony, suggesting that it may not be trustworthy. And second, the beliefs must derive from testimonial sources that are in fact reliable. Now, as is customary, Goldberg takes reliability to be intimately related to truth, and therefore to depend on factors that are external to the believer. It follows that the epistemic status of a testimonial belief is doubly dependent on external factors. According to Goldberg, then, when we reflect on the justification and rationality of testimonial beliefs, we gain a new appreciation of the plausibility of epistemological externalism—new in the sense that it is independent of the arguments for externalism that appear elsewhere in the literature. In sum, it is the author's view that plausible assumptions about testimony provide a foundation for fresh and largely autonomous arguments for semantic externalism and epistemological externalism. The book is an extended defense of this idea. The first chapter sets the stage for later developments by arguing for an initial characterization of testimonial knowledge. The characterization is based on four conditions. First, to acquire testimonial knowledge, an agent must form a belief that is testimonially grounded. In other words, the agent must form a belief in response to a piece of testimony, and in forming the belief, the agent must assume or presuppose that the testimony is reliable. Second, the agent's presupposition concerning the reliability of the testimony must be correct. Third, it must be the case that the agent arrives at a semantic interpretation of the testimony by a reliable process. That is to say, in Goldberg's words, the agent must recover “the proposition attested to by a reliable process of comprehension” (p. 31). And fourth, it must be true that the agent's acceptance of the testimony is “the upshot of a reliable capacity for distinguishing reliable from unreliable testimony” (p. 31). In all, then, reliability figures in Goldberg's initial characterization of testimonial knowledge in four different ways. Should this be a cause of concern for readers with strong internalist intuitions? Goldberg thinks not. One reason is that he is claiming only that the four reliability-involving requirements are necessary conditions of testimonial knowledge. He is allowing room for additional requirements, including requirements with a strong internalist flavor. Another reason is that, in chapter 1, at any rate, he is concerned only with knowledge. He thinks that even fairly radical internalists will be willing to allow that knowledge has a significant externalist dimension, though they would no doubt reject companion claims about epistemic justification or epistemic rationality. At all events, Goldberg offers justifications for his four conditions. The conditions are needed, he says, in order to honor the intuition that if a testimonial belief is to count as knowledge, then it cannot be a mere accident that the belief is true. Each of the second, third, and fourth conditions rules out a form of accidentality. Goldberg also uses examples to motivate the conditions. It is important to him that the conditions be accepted as correct, especially the third and fourth ones, for the book is in effect a lengthy meditation on their meaning and implications. Goldberg's argument for semantic externalism spans several chapters and is quite complex. I cannot do justice to all of its components here. In place of a complete account, I offer the following summary, which seems to capture a sizable portion of the core ideas: First premise: We have a great deal of testimonial knowledge. Second premise: Testimonial knowledge is possible only if the users of a language are equipped with procedures that enable them to recover propositions reliably from asserted sentences. (pp. 2–3) Third premise: In a wide range of cases, including a great many that arise in everyday life, we are confronted with the task of recovering a proposition from an asserted sentence in a context in which we have no real grasp of the beliefs and intentions of the speaker, and in which, therefore, we would be unable to determine which psychological states led the speaker to assert the sentence. (p. 41) Fourth premise: If, in many cases, we are unable to recover propositions from sentences by theorizing about the psychological states of speakers, then, assuming that we have the ability to recover propositions reliably, it must be true that we apply public linguistic norms in interpreting testimony—that is, fixed rules or principles that assign meanings to expressions, and that can be applied independently of assumptions about the purely psychological states and dispositions of specific speakers. (p. 41) Fifth premise: Since public linguistic norms can be applied independently of assumptions about the purely psychological states and dispositions that animate individual speakers, norms must be metaphysically independent of such states and dispositions, and it is therefore possible to imagine situations in which the operative norms are different than the ones that obtain in the actual case, but in which all purely psychological states and dispositions are the same. (pp. 104–5) Sixth premise: If the norms that assign meanings to sentences can vary while all purely psychological states and dispositions are held fixed, then semantic externalism is true. (p. 105) Conclusion: Semantic externalism is true. Although this line of thought touches most of the key bases, it also misses a few. Thus, while it is true that the fourth and fifth premises do justice to one of Goldberg's main reasons for maintaining that meaning is determined by public linguistic norms, he also has a second reason that is developed at some length. (My summary emphasizes chapter 2 at the expense of chapter 3.) In addition, I should note that Goldberg extends the present argument, which is concerned only with linguistic meanings, in such a way as to obtain the additional conclusion that the representational contents of propositional attitudes are determined externalistically. And there is also a third important omission. I will get to it in a moment. As I have represented him, Goldberg is concerned to derive semantic externalism from considerations having to do with testimonial knowledge, together with the observation that we are often unable to recover propositions reliably from asserted sentences by theorizing about the psychological states of individual speakers. If this is correct, then his line of thought is quite different than the arguments for semantic externalism that we find elsewhere in the literature. The other arguments make no mention of testimony, or of the risks that would be involved in interpreting assertions in terms of hypotheses about the psychological states of the speakers who produce them. Rather, they appeal directly to examples in which there is a dissociation between the descriptive content that a speaker has in mind in using an expression and the semantic meaning of the expression. Thus, as the reader will recall, Putnam begins one of his arguments by informing the reader that he lacks sufficient descriptive knowledge to distinguish elms from beeches, and would therefore need to obtain information from an expert in order to apply the terms “elm” and “beech” appropriately to specific trees.33 Putnam, “The Meaning of ‘Meaning’ ”. He then asks whether this fact precludes our crediting him with the ability to make true and false claims containing the words “elm” and “beech.” If, for example, he was to assert “Elms are trees,” would we deny that he had expressed a true proposition? More generally, does his lack of relevant descriptive knowledge preclude our crediting him with the ability to use “elm” and “beech” with their normal meanings? He predicts that his readers will answer “no” to both of these questions, and in most cases, this prediction has turned out to be correct. Putnam's other main arguments are also based on examples of psychological–semantic dissociations, though they tend to involve thought experiments rather than actual cases. The same is true of Kripke and Burge: They describe examples in which descriptive knowledge and meaning come apart, and then point out that these examples mandate an externalist account of meaning. It is clear that Goldberg's strategy for establishing semantic externalism is quite different than this one. Goldberg's line of thought is top-down: It relies on theoretical considerations about testimony and the elusiveness of Gricean speaker meanings. The standard arguments are bottom-up: They begin with intuitions about possible cases, and proceed from them to general morals. Goldberg's line of thought is interesting and suggestive. It deserves more attention than I can give it here. It seems unlikely, however, that it is comparable to the familiar arguments for externalism in strength and scope, even when various supplementary considerations that I have not described are added in. Among other things, the familiar arguments imply that semantic competence has an important social dimension, and more particularly, that individual agents can be credited with a full grasp of the meanings of words simply on the basis of dispositions to defer to other members of their linguistic communities. As far as I can see, Goldberg's line of thought fails to deliver these results. His fourth premise tells us that agents must rely on public linguistic norms in recovering propositions from assertions; but this actually tells us very little, because very little is said about the nature of public linguistic norms, or what is involved in conforming to them. As far as I can determine, the premise is compatible with the claim that agents must have a full and autonomous mastery of all of the concepts that would be involved in making the norms explicit, and also with the claim that agents must have a full and autonomous mastery of the meanings that linguistic norms assign to expressions. A similar point applies to the fifth premise. It tells us that it is possible for linguistic norms to vary while purely psychological dispositions remain fixed; but this is compatible with the claim that agents must have a full and autonomous grasp of the meanings that norms pair with words. (The fifth premise claims only that different pairings can be imagined. This is compatible with the claim that agents must have an autonomous grasp of all of the meanings that figure in any of the pairs.) In view of these facts, it seems that the conclusion of the argument is much weaker than the conclusions of the familiar arguments for externalism. It might be replied that there is considerable evidence in chapter 4 that Goldberg intends for his argument to have the same implications as the familiar arguments, and that, accordingly, there are grounds for doubting the adequacy of the foregoing interpretation. There is a great deal of truth to these charges: Goldberg clearly hopes to establish a fairly robust version of semantic externalism. (This is the third caveat that I mentioned above.) But this observation brings a dilemma into view. Goldberg must choose between embracing the foregoing argument as doing justice to the preponderance of his intentions, and changing the argument in such a way that it clearly and explicitly implies the familiar claim about the sufficiency of deference for a grasp of meaning, and the attendant claim that an agent's grasp of meanings can be partial and vicarious. If he embraces the foregoing argument as it stands, Goldberg can claim to be providing a rationale for semantic externalism that is independent of the familiar arguments, but he must also acknowledge that his conclusion is much weaker than the conclusions of the familiar arguments, and also much weaker than he would like. And if he changes the argument by adding premises that explicitly allow for a partial and nonautonomous grasp of linguistic norms, then he can claim that it establishes a fairly robust conclusion, but the argument will no longer be independent of the familiar ones. It will have to make an explicit appeal to the idea that deference is sufficient for a grasp of meaning, and Goldberg will have to defend this idea by examples like the ones used by Kripke, Putnam, and Burge. Thus, no matter which option he chooses, Goldberg will be unable to achieve one of his main goals. Or so it seems. I turn now to part II of the book, which is devoted entirely to epistemological issues. As I understand him, Goldberg is there principally concerned to defend the following four doctrines: AR: A hearer is justified in accepting (has the epistemic right to accept; is epistemically entitled to accept) another's testimony so long as there are no undefeated good reasons not to accept the testimony. (p. 144) AI-J: Whether S counts as justified in accepting a piece of testimony—as having the epistemic right or entitlement to accept the piece of testimony—depends on facts regarding one (or more) of S's social peers. (p. 150) AI-R: A complete account of the rational standing of S's (testimonial) belief that p depends on facts regarding one (or more) of S's social peers. (p. 194) AI-P: The process that eventuates in young children's consumption of testimony (and hence in the child's testimonial belief) extends to include aspects of the child's social environment—in particular, the coherence-monitoring that is performed for her by her adult guardian(s). (p. 234) According to AR, there is always a prima facie epistemic presumption in favor of accepting testimony; testimony is innocent until proven guilty. Moreover, the presumption is nonconditional. It does not depend on one's having inductive reasons to trust the testimony of a particular speaker, nor on one's having general inductive grounds for trusting all speakers. The justification to which AR refers is foundational in character. AI-J and AI-R are alike in that they both imply that the epistemic status of a testimonial belief has an important social dimension, and more particularly, that it depends on the reliability of the person who is the source of the testimony. They differ only in that AI-J is concerned with epistemic justification, while AI-R is concerned with epistemic rationality. (AI-R is therefore the more radical claim, because many epistemologists who hold externalist views about justification are inclined to adopt an internalist posture toward rationality.) As for AI-P, it is part of a package of views that can be summarized by saying, first, that the testimonial beliefs of young children often count as knowledge, and second, that they do so because in forming the relevant beliefs, children are protected from unreliable testimony by the endeavors of their parents and other responsible adults. AI-P emphasizes the most striking part of this package, the idea that in the case of children, the procedures for determining whether testimony is called into question by defeaters are not internal to believers, but are rather housed in the social environment. All of these claims are important, and Goldberg's discussions of them are quite interesting. In each case, his discussion repays reflective examination. Unfortunately, because of space limitations, I will not be able to consider all of the claims here. Because they seem to raise the most fundamental questions, I will focus on AR and AI-P. AR enjoys a certain initial plausibility, but it is not hard to think of cases that call it into question. Suppose that Jones has had very little exposure to the world of baseball, and is not in a good position to say whether the average baseball fan is well informed about ethological issues. Suppose that Jones somehow finds herself at a baseball game, and that the person sitting next to her begins to discourse on baboons, urging, among other things, that there is good reason to doubt that baboons have a theory of mind. Suppose further that in addition to being uninformed about the interests and abilities of baseball fans, Jones has no grounds for an opinion about the distribution of ethological expertise in the general population. And suppose finally that, unlike most people, she is not in possession of a well-developed statistical profile that can be used to determine whether conversational behavior is reflective and responsible. (Most of us can determine whether an informant presents the gestalt of a reflective, responsible person. Jones is less gifted in this area than the average person.) Is Jones entitled to trust the stranger sitting next to her? It would appear not. In fact, it seems that it would be quite irresponsible for her to trust him, just as it would be quite irresponsible to place one's trust in a handwritten discussion of baboons that one found lying on a sidewalk. Now, Goldberg is well aware of examples of this sort, and is also well aware that, in the view of many people, they seem to support the following principle, which is of course abruptly inconsistent with AR: RD: A hearer is not epistemically justified in accepting (does not have the epistemic right to accept; is not epistemically entitled to accept) another's testimony unless she has (inductive or a priori) reasons, ultimately not themselves based on further testimony, for regarding the testimony she confronts as credible. (p. 143) Goldberg does not accept this principle, but he appreciates its plausibility. Moreover, as the reader will recall from my earlier summary of chapter 1, Goldberg is committed to the following thesis: G: If an agent's testimonial belief is to count as knowledge, then the agent's acceptance of the relevant piece of testimony must have been vetted by a reliable capacity to distinguish reliable attestations from unreliable ones. (p. 3) Given counterexamples like the one involving Jones, and the associated intuitive pull of principles like RD and G, how can it be appropriate to embrace AR? Goldberg's answer is that AR is advanced against the backdrop of two particular empirical assumptions. The first is that testimonial belief-fixation involves (subcognitive) processes whose job it is to enable the recipient to discern cases in which trust is to be extended, from cases in which it is not. And the second is that these processes are in fact highly reliable in that, in cognitively mature and well-functioning adults, they issue in acceptance in most of the cases where the testimony is both true and reliable, and they issue in non-acceptance in most cases where the testimony is false or otherwise unreliable. (p. 148) In effect, then, Goldberg defends AR by combining it with two other principles. The resulting package authorizes us to trust testimony, but only in circumstances in which the following three conditions are satisfied: (i) the relevant agent has no good reasons for doubting the testimony; (ii) the agent is equipped with appropriately powerful procedures that enable him to detect such reasons when they exist; and (iii) the procedures in question are reliable. The defense is at least partially successful. Thus, while Jones is clearly a counterexample to the original version of AR, she is not a counterexample to the present more elaborate version. By hypothesis, she is not equipped with procedures for determining when special factors are present that call the trustworthiness of testimony into question. Or at least, she has no such procedures that can be applied to baseball fans who spin ethological tales. At first sight, anyway, the new version of AR is a well-motivated compromise between the original version of AR and the opposing principle RD. The new version differs from the original one in that it incorporates a substantial restriction on the presumption of trust, implying that trust can be extended to a piece of testimony only after it has been subjected to a vetting procedure that is reasonably powerful and that normally delivers correct results. Before testimony can be accorded trust, it must pass through a filter that is capable of detecting defeaters. Even with this restriction, however, AR seems much more liberal than RD. In effect, RD denies that there is a presumption of trust of any kind, for it implies that trust is appropriate only when an agent has an inductive or a priori reason for thinking that a piece of testimony is correct. Readers who are familiar with the literature on testimony will recognize AR as a version of antireductionism, the view that acceptance of testimony is an autonomous and fundamental way of acquiring knowledge. According to this view, the acceptability of testimony does not in any way depend upon supporting considerations deriving from such alternative modes of acquiring knowledge as perception, memory, and inductive reasoning. Readers will also recognize that RD is a version of the opposing position known as reductionism. Both of these positions command wide followings, and are debated endlessly in the literature. In effect, then, in trying to broker a compromise between AR and RD, Goldberg is trying to resolve a large-scale quarrel within the ranks of epistemologists. Will the proposed compromise have the intended result? This remains to be seen. To be sure, Goldberg's proposal makes it more difficult for reductionists to defend their position, for, as we saw, it blocks counterexamples like the one involving Jones. This is a significant advance. It is far from clear, however, whether the proposal provides the foundation for a fully general defense of antireductionism. Goldberg defends AR by invoking procedures for detecting defeaters, but he says very little about the nature of the procedures. What sorts of defeater must an agent be able to detect? And what sorts of background knowledge must an agent possess? In the absence of answers to these questions, it is hard to see whether the proposed revision will succeed in blocking all of the compelling counterexamples. (Notice that it is not ultimately satisfactory to say, as I have above, that the procedures must be “appropriately powerful” or “strong enough.” Being strong enough might entail having substantial background knowledge of the domain to which a piece of testimony refers, and being able to tell whether the testimony coheres with that knowledge. If so, then Goldberg's revision of AR would be at risk of collapsing into RD, for it would in effect require that an agent be in a position to assess testimony for truth, and this would tend to undermine the claim that testimony can be presumed to be correct.) I turn now to Goldberg's discussion of the principle he calls “AI-P.” To repeat, it runs as follows: AI-P: The process that eventuates in young children's consumption of testimony (and hence in the child's testimonial belief) extends to include aspects of the child's social environment—in particular, the coherence-monitoring that is performed for her by her adult guardian(s). (p. 234) To express the point of AI-P in somewhat different terms, if the testimonial beliefs of young children count as nonaccidentally true, and therefore as knowledge, it is because the children in question are protected from unreliable testimony by the endeavors of their parents and other responsible adults. Goldberg makes a number of interesting remarks in elaborating this important claim. I will focus here, however, on what he says in defense of it. The defense begins with the following premises: TKC: Cognitively immature children acquire knowledge through testimony. (p. 204) NT: Knowledge presupposes a non-trivial satisfaction of the No-Defeater condition. (p. 233) In interpreting Goldberg's reference to the “no-defeater condition” in NT, we should, I think, recall his earlier claim that if a testimonial belief is to count as knowledge, the testimony must have been vetted by a procedure for detecting defeaters that is both reasonably powerful and highly reliable. The defense of AI-P also makes use of a third premise, namely, LAK: LAK: Children cannot properly be described as satisfying the No-Defeater condition. (p. 233) It is not hard to discern the line of thought leading from TKC, NT, and LAK to AI-P. In brief, it goes like this: Because children acquire knowledge through testimony, the processes that produce their testimonial beliefs must at some point involve an agent who is able to detect factors that would call the reliability of testimony into question. But children are cognitively immature, and are therefore unable to detect such factors on their own. It must be the case, therefore, that other agents play an essential role in the processes that produce testimonial beliefs in children. It must also be the case that these other agents have the ability to detect defeaters and the will to protect children from defeated testimony, either by preventing them from encountering it or by discouraging them from taking it at face value. Reflection shows that there are in fact agents who fulfill these conditions—namely, parents, teachers, and other adults who assume responsibility for the welfare of children. The conclusion of the argument, AI-P, is surprising and also, as Goldberg acknowledges, somewhat counterintuitive. Why is this? The answer, I think, is that it allows for a separation between the faculties that are responsible for generating testimonial beliefs and the faculties that are responsible for monitoring testimony for evidence of unreliability. According to AI-P, the generating faculties and the monitoring faculties may be housed in different agents. It follows that it is in some ways much easier for an agent to be justified in holding beliefs than epistemology has traditionally maintained. It has traditionally been thought that the believer must have some sort of sensitivity to defeaters, and must be able to respond to defeaters in an appropriate way. AI-P contradicts this idea, allowing that agents may fail to satisfy the traditional conditions concerning defeater detection provided that there are compensatory factors in the external environment. In effect, AI-P aims to reorient epistemology by saying that it is not single agents but rather teams of agents that are the units of epistemic appraisal. This is a surprising and radical view. Although Goldberg is to be thanked for making this interesting idea available, I feel that, in the end, it is too counterintuitive to be acceptable. I think we can best appreciate its problematic character by turning our attention temporarily away from children, and focusing instead on an adult who accepts testimony uncritically. Suppose that Ming is an emperor who has little interest in or aptitude for the affairs of state. He forms many beliefs about what is happening in his realm, but he does so only on the basis of assertions by his favorite adviser, who, fortunately, is always at pains to base his judgments on large amounts of high-quality evidence. Suppose also that the emperor has no reason to trust his adviser. Rather, his trust is blind: He grew up trusting the adviser's testimony concerning conditions outside the palace, and he has found no reason to withdraw that trust. Suppose, finally, that Ming has no sense of the factors that are reliably correlated with testimony that is misleading or false. That is to say, he has no ability to recognize defeaters, at least with respect to testimony about the affairs of state. Now, as I understand Goldberg, because Ming is protected from error by the honesty and vigilance of his adviser, he must say that Ming has knowledge of conditions in his realm, and also that Ming's beliefs about those conditions are epistemically justified. But it seems to me that these propositions are quite wrong. Because Ming is an adult, it is natural to see him as a paradigm of epistemic vice. He lacks even a rudimentary ability to detect defeaters, and (we can suppose) he has no interest in acquiring such an ability. In view of these failings, it seems inappropriate to say that Ming has knowledge of the conditions in his realm, except in some vicarious sense that is quite different than the one that we normally have in mind. As I see it, any theory that would encourage us to praise Ming's beliefs, by sharply separating knowledge and justification from epistemic competence, is misguided. To say this, however, is to invite the question of how it is possible for premises as appealing as TKC, NT, and LAK to lead to an unacceptable result. My answer is that when we reflect carefully on the premises, we find that they are less plausible than they seem at first sight. More specifically, reflection shows that both TKC and LAK are problematic: TKC: Cognitively immature children acquire knowledge through testimony. LAK: Children cannot properly be described as satisfying the No-Defeater condition. Goldberg thinks that TKC receives strong support from intuitions. In fact, however, it seems that our intuitions may only support the much weaker claim that cognitively immature children can acquire knowledge from parents and other caregivers in whom they can appropriately place a special trust. To appreciate the merits of this view, suppose that a child encounters a voluble but benevolent stranger in the street. This stranger, an adult, provides the child with a lot of highly accurate information about the Boston Red Sox—that they play in Fenway Park, that Babe Ruth was once on their roster, and so on. Suppose further that as they stroll, the stranger is at pains to make sure that the child acquires no misinformation from other sources. Thus, he shoos away a hot dog vendor who is telling lies about his product, distracts the child when they are passing a billboard that proclaims the virtues of General Motors cars, and so on. I find that I have very little inclination to say that this child is acquiring knowledge from the stranger. The reason is that the stranger is just a casual acquaintance, and the child has as yet no ability to distinguish between casual acquaintances who are trustworthy and casual acquaintances who are unreliable. Accordingly, if the beliefs that the child acquires from the stranger are true, this is just a happy accident. Even though she does not share Ming's intellectual laziness, the child is equally dependent on epistemic luck. It would be wrong to ascribe knowledge or epistemic justification to either of them. I also have a complaint about LAK. It seems to me that children often have a lot of evidence concerning the reliability of their parents and other familiar caregivers. Every day, children participate in large numbers of conversations like these: Child: Where is my shoe? Parent: Under the kitchen table. Child: May I please have some peas? Parent: There are peas in the green bowl. Child: Can Mary come over to play? Parent: Yes, I've already invited her; she'll be here in a few minutes. In a huge percentage of cases, exchanges of this sort are followed by confirmation of the parent's testimony. As a result, children have immense amounts of evidence that their parents are reliable sources of information about domestic matters. Accordingly, it is entirely appropriate for children to trust the testimony of their parents concerning such matters. But also, their evidence provides support for the view that their parents are generally reliable, at least with respect to matters that fall within their ken. Because of this, I suggest, children are entitled to trust the testimony of their parents concerning matters in the great world. Parents are generally much less reliable about external affairs than domestic ones, but it seems that the evidence available to children makes it appropriate for them to adopt a general posture of trust. This posture is of course modified as they mature. In my judgment, Goldberg did not work hard enough to make his ideas accessible to the reader. A number of passages, including some fairly important ones, are characterized by vagueness of intent, partly because certain central notions (such as the concept of a public linguistic norm) are left unexplained. On the whole, however, I found the book highly rewarding: it contains suggestive arguments for many significant doctrines; it raises several important questions that are genuinely new; and it calls attention to connections among a number of issues that seem prima facie to be independent. I found it quite stimulating, and, despite having reservations about a number of its main claims, I learned a great deal from it. I look forward to Goldberg's next book.
Referência(s)