1 Thanks are due to Carole Rovane for shrewd and patient criticism on matters of substance and style.

2 To avoid any misunderstanding, it should be emphasized that this claim does not imply that the inquiring agent should regard himself as justified in suppressing the views of those with whom he disagrees. We may tolerate the public expression of dissent while expressing our own contempt for it. But we are not obliged to register contempt either. Indeed, the agent might sometimes listen to the dissenter's view even though those views seem absurd. Teachers are often obliged professionally to pay attention to points of view which they judge to be patently false and, indeed, to pretend that their minds are open when they are not. Such insincerity may be justified if it is effectively employed to induce students to question their views. And where professional obligation does not support such dissembling, the demands of civility in discourse may. We should, however, distinguish between contemptuous toleration both when the contempt is overt and when it is disguised for pedagogical purposes or due to considerations of conversational etiquette and genuine respect for dissenting views. Respect for a dissenting view arises when an agent confronted with dissent recognizes a good reason for genuinely opening his or her mind by ceasing to be convinced of the view initially endorsed which is in conflict with the dissent. liberals of the sort I admire tolerate dissent but their toleration is often contemptuous. To confuse toleration with respect for the views of dissenters can lead advocates of toleration to urge upon us the skepticism of the empty mind.

3 John Dewey, Logic: The Theory of Inquiry (New York: Holt 1938), 118-20; and ‘Propositions, Assertibility and Truth’ (1941), reprinted inS. Morgenbesser, ed., Dewey and His Critics (New York: Hackett 1977) 265-82, at 270-2. According to Dewey, every case of knowledge ‘is constituted as the outcome of some special inquiry. Hence, knowledge as an abstract term is a name for the product of competent inquiries’ (Dewey, Logic, 8). In my ‘Doubt, Context and Inquiry,’ in L. Cauman, et al., eds., How Many Questions (New York: Hackett 1983), 27-8, I suggested that Dewey should have abandoned this conception of knowledge. He should not have insisted on equating knowledge with warranted assertibility if heal so wished to acknowledge, as Peirce had, that to be settled, a conviction did not require a prior justification. It may, perhaps, be thought that an appreciation of the context sensitivity of Dewey's conception of ‘warranted assertibility’ would mitigate the difficulty. But if a proposition used as background information in an inquiry is not a warranted assertion relative to some prior inquiry, it is not knowledge even in the current inquiry in Dewey's official sense. Since Dewey is quite clear that his sense of ‘knowledge’ is honorific, withholding the epithet is tacit attribution of some deficiency. But it is precisely the allegation of the existence of such a deficiency that Peirce sought to rebut in The Fixation of Belief.’ Knowledge and warranted assertibility cannot coincide as Dewey suggests. There can be propositions used as background information and, hence, qualifying as knowledge which are not products of prior properly conducted inquiries and, hence, are not warranted assertions relative to any historically conducted inquiries. Moreover, it can even happen that knowledge fails even for items that are warrantedly assertible relative to one inquiry provided the result claimed has been legitimately called into question in subsequent inquiry. It may, perhaps, be worth mentioning that Dewey tended to think of conjectures or hypotheses as what I have tended to call ‘potential answers’ to the question under investigation. For me the distinction between settled assumptions and what is open to doubt or conjectural is presupposed by the deliberating or inquiring agent when facing a problem. It is one of the tasks of inquiry addressing a certain question, to identify potential answers to the question under investigation. This is the task of what Peirce called ‘abduction.’ Three minimal demands should be imposed on a potential answer: (a) Its truth should be consistent with what is taken for granted, (b) it should be a relevant answer to the question raised, and (c) whether it is to be accepted or rejected should be decidable through inquiry. Condition (a) indicates that potential answers or conjectures in Dewey's sense presuppose a distinction between certainties and conjectures in my sense.

4 The utility function representation is to be understood in this setting as characterizing the agent's values and goals in the deliberation and not as representing the net of pleasure over pain or his or her desires. The utility function could represent moral, political or cognitive valuations. The core principle involved is the principle of maximizing expected utility conditional on the act chosen. See my Enterprise of Knowledge (Cambridge, MA: MIT Press 1980), ch. 4, for a survey of core features of Bayesianism.

5 For further elaboration of these qualifications, see my Enterprise of Knowledge and Hard Choices (Cambridge: Cambridge University Press 1986).

6 Alan Gibbard, Wise Choices, Apt Feelings (Cambridge, MA: Harvard University Press 1990).

7 C.S. Peirce, Review of Venn's Logic of Chance reprinted in M. Fisch, eta!., eds., The Writings of Charles 5. Peirce (Indianapolis: Indiana University Press 1984-86), Vol. 2, 98-102; ‘The Probability of Induction,’ reprinted in The Writings of Charles 5. Peirce Vol. 3, 300-1; and my ‘Chance,’ in L. Nissen, ed., Philosophical Topics: Philosophy of Science 18 (1991) 95-121, at 99-103

8 John Dewey and John H. Tufts, Ethics (New York: Holt 1932); and my Hard Choices, ch. 1

9 It may be pointed out that Dewey himself would have regarded the distinction I am making between norms of rationality and other norms as an untenable dualism. All norms are in some context or other objects of criticism and inquiry. But if our aim is to offer a systematic account of the rational conduct of inquiries concerned with the revision of beliefs, goals, values, and other attitudes, the proposed accounts we offer will perforce draw a distinction between the norms which are held fixed in these accounts and those which are open to revision. I grant that disputes can arise as to which norms should qualify as the fixed norms of rationality. But any specific proposed account of inquiry acknowledges is committed to some distinction between norms which are fixed and norms which are open to revision. Rival proposals should, if they entertain the ambition of being systematic, share in common recognition of some sort of distinction between the fixed minimal norms of coherence or consistency and norms which are open to revision even if they differ concerning how the line is to be drawn. I propose to restrict the norms of rationality relative to an account of inquiry to the fixed minimal norms specified by the theory.

10 Neither Dewey nor Peirce explicitly claim this. Principles of reasoning are habits or leading principles or the like. The rhetoric, however, could often be read as blueprints for rational automata or as principles applicable in self criticism. Yet some passages suggest the latter reading fairly clearly. Thus Dewey writes: ‘A postulate is also a stipulation. To engage in an inquiry is like entering into a contract. It commits the inquirer to certain conditions. A stipulation is a statement of conditions that are agreed to in the conduct of some affair. The stipulations are at first implicit in the undertaking of inquiry. As they are formally acknowledged (formulated), they become logical forms of various degrees of generality. They make definite what is involved in a demand. Every demand is a request, but not every request is a postulate. For a postulate involved the assumption of responsibilities …. On this account, postulates are not arbitrarily chosen. They present claims to be met in the sense in which a claim presents a title or has authority to receive due consideration’ (Dewey, Logic, 16-17). Dewey continues later to observe that when a specific person engages in inquiry, ‘he is committed, in as far as his inquiry is genuinely such and not an insincere bluff, to stand by the results of similar inquiries by whomever conducted. ‘Similar’ in this phrase means inquiries that submit to the same conditions or postulates’ (Ibid., 18). The postulates Dewey is talking about here and which he regards as the terms of the contract an agent enters in undertaking an inquiry are clearly, for this reason, normative or prescriptive. As terms of a contract, they are intended to formulate prescriptions which the party to the contract endeavors to meet. When the undertakings are explicit, Dewey regards them as postulational. I take this to mean that postulates are principles the agent can explicitly recognize and use in evaluating the extent to which he is fulfilling his contract.

11 The assumptions which the deliberating agent makes concerning his abilities resemble in certain important respects the assumptions made when an inquirer judges that a stochastic experiment is to be implemented. If a die is about to be tossed once, the inquirer presupposes that exactly one of six kinds of outcome is about to occur. These six possible outcomes or points in the sample space represent abilities of the die to respond in these six ways on a toss. The die is also presupposed to have a sure fire disposition to land in exactly one of these ways on a toss. In deliberation, the agent makes analogous assumptions. He takes for granted that he has the ability to make true each of a variety of propositions through his deliberation (through his choice) and that he is constrained by deliberation to make exactly one of these propositions true. The space of objectively feasible options is like a sample space. See my Hard Choices, ch.4.

12 F. Schick, ‘Self Knowledge, Uncertainty and Choice,’ British Journal for the Philosophy of Science 30 (1979) 235-52, at 243. See my Hard Choices, 4.3 for further discussion of views on this issue.

13 Appeal is being made here to accounts of contraction which recommend violations of the so called ‘Recovery Postulate’ under the conditions envisaged in the text. The Recovery Postulate is discussed and defended in Peter Giirdenfors, Knowledge in Flux (Cambridge, MA: MIT Press 1988), 3.4. Violation of this postulate is defended in my The Fixation of Unbelief and Its Undoing (Cambridge: Cambridge University Press 1991), 4.5.

14 Akeel Bilgrami, in Belief and Meaning (Oxford: Blackwell1991), mounts a convincing attack on theories of so called ‘wide content,’ ‘direct reference’ and the like because of their inability to show how content so construed can be deployed in the explanation of human behavior. The weak link in Bilgrami's argument, so it seems to me, is his assumption (shared with many of those he criticizes) that the primary function of appeals to beliefs and desires is in explanation of behavior. However, it seems to me that this reservation with Bilgrami's argument does little to damage it. Even if principles of rationality are norms rather than laws explaining behavior, we should want to claim that were we rational agents completely satisfying the dictates of such principles, self criticism would be otiose, rational angels would be rational automata and our behavior would be explainable by these principles. Indeed, were this not the case, we would regard the principles of rationality as somehow defective as norms for use in self criticism.

15 See my ‘The Demons of Decision,’ The Monist 70 (1987) 193-211; ‘Consequentialism and Sequential Choice,’ in M. Bacharach and S. Hurley, eds., Foundations of Decision Theory (Oxford: Blackwells 1991); and ‘Feasibility’ (forthcoming).