Group conflict continues to be a salient issue. Problems that rational choice theory has in other areas are brought to prominence in group conflict, because the conflict often revolves around differences in people’s positions and beliefs, not just around their interests. The fact that people’s beliefs are closely associated with their group allegiances is the basis of some tactics of influence and manipulation. These tactics raise a variety of ethical concerns, ranging from concerns about hypocrisy to concerns about the harm they can do to individuals. The ethical issues cannot be resolved by rational choice theory, but will require us to address the ethical problems of reconciling belief commitments with commitments to peace and harmony.

In last year’s Presidential Address I tried to tie together some thoughts about ethics and dialogue in religion and public policy. I suggested that in the contemporary world it is important to consider ethical issues about group conflicts, and that some of the approaches to ethical issues which are shown in much recent public policy development fail to get to grips with some key ethical issues about conflict between groups.

In particular, I suggested that much public policy development focusses on issues to do with the interests of individuals and groups. One problem with that focus is that it does not deal effectively with group differences that are embodied in different ways of seeing the world, ways of seeing which may not correspond to factual differences but which are cognitive differences nonetheless. These differences are not easy to conceptualise as differences in interests, if interests are cashed out in terms of preferences, to be contrasted with cognitive differences.1 I suggested that we need to give some thought to the ethics of dialogue about different ways of seeing things. This year, I want to consider some different but related points still about ethical issues to do with group conflict.

First, however, it is worth noting that the issues are just as salient now as they were last September. In Iraq, we see continuing violence which is at least to a significant extent associated with sectarian religious commitments. In many other parts of the world we have seen riots or repression tied to religious or ethnic commitments. Even here, in Sydney, the Coogee riots of last December were notable enough to make front-page news in much of the world. One result was an announcement from the Federal government that nearly half-a-million dollars would be spent promoting respect and understanding amongst ethnic groups.

In what follows, I deal with such issues in abstract terms and with reference to some historical examples. The historical examples help to make the general points, I think, but limitations of space and time mean that I cannot also relate those general points to concrete examples from the contemporary world. I hope that it will not be hard for anyone to see how that can be done.

Groups, Decisions and Roles

When I noted last year that differences of religious belief may be an example of differences that sustain group conflict but which cannot easily be analysed in terms of interests, I used this as an example of the way that group “positions” may be important as a focus of conflict. Although it can often be good advice for a negotiator to focus on “interests” rather than “positions”, there are some conflicts where “positions” are very important. I noted the concerns people have and the decisions they make and the actions they carry out which emerge from values and ethical commitments, and that there is evidence that these are bound up with people’s positions and social identities.

It is not only differences of religious belief that may be associated with different group positions. The idea of “group identity” is well known, and so are such related ideas as “in-group” and “out-group”. They involve some significant cognitive effects, such as perception of oneself and one’s own group as different from another group, and failures to notice variations within the other group.3 These and related processes of “social categorisation” and “social identification” include forms of misperception that can sustain group conflict.4

The fact that such identity can play a significant role in group conflict can be seen as an instance of a more general claim that an individual’s social identity can play a significant role in their action and decision, and this is at odds with the view of rational choice theory that behaviour is to be explained primarily in terms of self-interest, or utility maximisation. A useful discussion about the difference is given by James Montgomery. He notes that rational choice theorists tend to model mutual cooperation and trust in terms of a repeated Prisoners’ Dilemma, where self-interested actors sustain cooperation through “calculative trust”: “I trust you because I calculate that your short-run benefit from an opportunistic defection is outweighed by your long-run benefit from continued cooperation.” However, he notes, “experimental evidence raises doubts whether cooperation in PDs is based upon calculative trust.” 5 He notes that the outcomes of Prisoners’ Dilemma trials are significantly affected by the instructions given to participants, even when the payoff matrix is not affected. If behaviour is primarily to be explained in terms of participants calculating how their interests are best served, it is difficult to see how factors other than the payoff matrix can have such a substantial effect.

What is especially notable is that the sorts of effects in question are produced by changes to instructions which seem relatively minor. For example, there are significant effects from labelling one course of action “cooperation”, rather than just identifying it by a number or a letter. In one notable experiment, some pairs of subjects were told that they were playing the “Wall Street Game”, and others that they were playing the “Community Game”, with no other differences between the situation they were presented with:

Given the usual game-theoretic presumption that labels are irrelevant, the experimental results are striking. While only one-third of the subjects in the Wall Street condition cooperated in the first period, more than two-thirds of the subjects in the Community condition did so.6

In some respects, this reminds us of work about the psychological phenomenon of “priming”, where “recently and frequently activated ideas come to mind more easily than ideas that have not been activated”.7 Some research has found that self-interest can be “primed”: when people were exposed to stimuli like conversations that highlighted considerations of self-interest for those present, they were more likely later to cite self-interest as a reason for a particular political position.8 However, Montgomery’s discussion may also remind us of the original work that was done on group identity. It came as something of a surprise to researchers to learn that inter-group rivalry and hostility did not arise only from competition for scarce resources or from incompatible goals. In “minimal group” studies, they found clear tendencies for people to favour members of their own groups in matters of resource allocation, even when the “group” in question was essentially a fictional construct, identified only some minor difference in description.

In both minimal group experiments, and in the Prisoners’ Dilemma simulations referred to by Montgomery, we see significant effects on people’s behaviour produced by apparently small variations in the way the situation is described. Montgomery suggests that the difficulties that rational choice theory has in explaining such phenomena might lead us to consider sociological “role theory” as a better alternative for conceptualising them. On this view, individuals view themselves as embodying socially constructed roles, partly because of the recurring need we have to understand our place in the world. We might then expect small cues to prompt us toward understanding ourselves and our position in the world in one way rather than another, and have a disproportionately large effect on our behaviour.

The idea that role theory can give a better theoretical account of some behaviour than rational choice theory does, is tied to reasons why rational choice theory may not give a good explanation of group conflict. In particular, it is tied to the fact that beliefs and positions can have effects which are difficult to explain by reference to calculation of interests. Small variations to instructions significantly affect people’s behaviour in simulations of group behaviour, just as they do in Prisoners’ Dilemma simulations. They seem to do so because they affect the extent to which people accept an identity which is tied to their membership of the group. But the result is that variations in belief which to outsiders seem quite small differences can be tied to significant differences in group membership. Such differences do not seem to be clearly related to utility-based calculation, but to do with differences in self-understanding and social identity. These differences are often reflected in differences of belief.

Group beliefs and politics

Because differences in description and belief are associated with differences in group membership, differences in belief can be of major significance in group politics. We know that there are substantial influences on members of groups to conform to norms and expectations about behaviour by group members.10 If there are pressures on individuals who are group members to conform to group norms, and if group membership is tied to certain kinds of belief commitments, then it is not surprising that group conflict provides fertile ground for persuasion and manipulation that are ethically problematic.

Ruth Grant argues that hypocrisy is inevitable in politics. Part of her argument effectively revolves around the fact that public politics involves group allegiance, together with espousal of the norms and standards that go with identity as a group member.11 Rather than showing genuine normative commitment, espousals of norms and values are likely just to signal group membership and acceptance of what it entails. The same can be true for beliefs: espousal of a belief can function primarily to show group membership, rather than genuine commitment to the content of the belief. Bar-Tal suggests that “group beliefs serve as a foundation for group existence”12, and that “group beliefs provide the psychological framework that allows group members to structure their social reality about the group”.13 The fact that members of a group share a belief not widely held by others may help them to see themselves as members of the same distinctive group.

Sometimes, as a result, a position on an issue is like a flag, or rallying-point, for group members, something for them to gather around, and defend. Such a symbolic rallying-point can be important because it is thought important, however separable it may really be from the life of the group. In that situation, it can be hard to discern whether affirmation of a belief shows sincere commitment to it. There is an old saying that “If the Archbishop of Canterbury says he believes in God, then that’s by way of business, but if he says that he doesn’t believe in God, then you can believe him.” The basis for the aphorism is that people do on occasion affirm beliefs for the sake of acknowledging group membership, and that when they do so it can be hard to evaluate their sincere commitment to the belief. On the other hand, if they affirm a belief contrary to what is expected of their group, the only plausible explanation is genuine commitment to the belief.

Group beliefs have multiple functions, from rationalisation of action to reinforcing group solidarity, and affirming a belief can equally play multiple roles, from reassuring others about reasons for action to giving information about group strength.14 We know from many sources that affirmations of belief can play multiple roles even for individuals, and that contextual factors can play an important role in how such affirmations are to be interpreted.15 A number of authors have analysed political implications.16 The relationship between affirmation and alignment can figure in a variety of political tactics. One such tactic is to put others in situations where they will be forced to make an affirmation that will align them with one party or another. Perhaps the most famous and clearest example is the Gospel story of Jesus:

And they sent to him some of the Pharisees and some of the Herodians, to entrap him in his talk. And they came and said to him, “Teacher, we know that you are true, and care for no man; for you do not regard the position of men, but truly teach the way of God. Is it lawful to pay taxes to Caesar, or not? Should we pay them, or should we not?”17

Jesus has three options. One, refusing to answer, seems to renounce claims to wisdom and authority. The others, to answer either “yes” or “no”, each threatens to compromise him by aligning him with one party or another. For the “positions of men” referred to by his interlocutors are well-established political positions:

if he advised payment of the tax he would discredit himself in the eyes of all the nationalist groups (here perhaps symbolized by the Herodians, v. 13); for such advice would be regarded as a betrayal of the national cause, and no one who gave it could expect much support or acceptance of his messianic claims from the crowds. If he pronounced against payment he would offend those (represented in v. 13 by the Pharisees) who were prepared to tolerate the status quo, and he might even be reported to the Romans for inciting to rebellion.

In fact, of course, Jesus exits neatly between the horns of the dilemma. By getting them to show him a coin and drawing their attention to the image on it, he moves the onus of proof back to them: if they carry coins that bear Caesar’s portrait, the onus is on them to show why they do not thereby acknowledge their duty to pay taxes, and to admit the role they had themselves played in bringing about Roman rule.19

Some similar kinds of cases have been analysed by Riker. One is a similar tactic to that used by the Herodians and Pharisees, but used this time by Abraham Lincoln against his political opponent Stephen Douglas.20 In this and other cases, Riker shows how political tactics often aim to compromise opponents by forcing them into positions that will alienate some or other of their allies (a modern term for this is “wedge politics”).

Beliefs, Positions and Commitment

The link between a position on a specific issue and group beliefs in general, may be quite clear-cut, or less so. There is no difficulty about seeing the relevance of an Archbishop’s position regarding the existence of God, to the beliefs of the Church he leads. The legality of paying taxes to Caesar needs more explanation for us to understand its relationship to group allegiances. These may have been quite clear to people at the time, but the relationship nevertheless depends on their array of background beliefs and knowledge. There are a range of cases where such relationships are more subtle still, down to cases where a proposal is put only for the sake of inviting people to come down on one side or another, explicitly to test their allegiance and perhaps to offer a forum for debate and manoeuvre.

In calling attention to beliefs and commitments that are associated with group membership, influence tactics can make group membership more salient and bring forth a whole array of influence mechanisms that revolve around group membership, such as conformity effects and shared conceptions of what is legitimate. It also generates expectations in others as well as group members. In particular, it affects others’ evaluation of beliefs affirmed by group members.

Because of the role people’s affirmations play in group politics and influence tactics, we may be inclined to set aside the content of the affirmations when we take them merely to exhibit group membership and solidarity with other group members. Then, of course, there is the question which comes first, the belief or the group? Is it that people come first to different beliefs, and then form groupings around those beliefs, or is it that people first align themselves, and then embrace beliefs that distinguish them from others?

If we think about politics and group alignments in general, then the answer is, of course, that both occur, in various ways. There is no reason to believe that every case is the same. In some, the content of the positions may be most important, and in others it may be personal or group allegiances. We should note, therefore, that people’s espoused commitments are not necessarily insincere, just because they are associated with group alignments. Sometimes, group membership may emerge from prior, sincere commitments, rather than the other way round. In that case, commitments associated with group membership can be genuine commitments.

In such situations, the demands on us as observers are complex partly because of the interplay of different sources of allegiance and obligation, but they are also complex because of the difficulty of evaluating people’s affirmations and positions. Thus, to revert to our earlier example, when the Archbishop of Canterbury affirms that he believes in God, we may need to know more about him to decide whether he is truly committed to that belief. Without doubt, there have been many Archbishops of Canterbury who have been devout and sincere in their professions of faith, but it is not hard to find some historical examples of clergy and ostensibly committed lay people in various religions whose interests seem more to have been in the temporal than the spiritual benefits of faith, and whose affirmations of belief mainly signalled a political position.

How we should evaluate others’ affirmations of belief will vary from case to case, because the relationship between affirmation and group membership will vary from case to case. Religious beliefs provide vivid illustrations of complex interplay between belief and group alignment. Consider the negotiations between Catholics and Protestants at Regensburg in 1541.22 The aim of the Regensburg meeting was to attempt a reconciliation between Catholics and Protestants. Central to that had to be agreement on some contested matters of doctrine, but intermingled with it also were issues to do with the political alignments of the day: “by the time of the Diet of Regensburg in 1541 doctrinal and socio-political questions were inextricably bound up with one another”.23 The authority of Scripture is one prominent example.24

Participants in the Regensburg discussions considered such major issues of doctrine that separated Catholics and Protestants, and in doing so they focussed in great detail on the substance of the issues. But, at the same time, their deliberations were surrounded by implications of one position or another on party structures and alliances. The Emperor Charles V needed resolution of the religious schism to restore his political authority in Germany, the Papacy was concerned both about doctrine and about restoration of church lands, while German princes were concerned both about the religious principles that united their people and the preservation of a measure of independence. The religious and political issues were completely entwined. There is no need to assume that debate over doctrinal issues was simply a cloak for political bargaining, or that the debate would reach theological conclusions in an ivory tower remote from political considerations.

Commitment and Hypocrisy

We can find similar interplay between religious doctrine and political positions throughout history, from the rise of Christianity in the Roman Empire down through the role of the Catholic church in communist Poland during the days of the Solidarity movement, to present-day tensions in the Middle East. Bar-Tal notes that “group beliefs, which characterize the group, demarcate its boundaries with the out-groups”25, and the corollary is that by professing such a belief one marks oneself as a member of the in-group.

Problems then can arise about interpreting others’ affirmations, because of all the different motives people may have for their affirmations, and the different functions that such affirmations can have. The extent to which utterances can display group allegiance is often a crucial function, but a function which can be masked. Awareness of groupings, and their associated beliefs, can be important in all sorts of social and political life. Inviting an affirmation can be an invitation to disclose group membership, testing the person to whom the invitation is given. “Knowing their hypocrisy”, Jesus said to those who approached him, “Why put me to the test?” Partly, the test was to know that it was a test. Because people’s affirmations of belief sometimes have more to do with signalling membership of a particular party than with reflective commitment to the content of the belief, asking a question may ostensibly express interest in a particular issue, but also aim to lure the other party into some compromising response.

These are all instances of utterances or actions which have an ostensible purpose and one or more ulterior purposes. They all raise some similar ethical issues, potentially departing from the requirement that we treat one another as autonomous subjects and allow one another to make responsible decisions based on the best information available.

In some such cases, where a question or utterance has amongst its functions that of putting people off their guard, any ethical problem about attempted dissimulation may be treated as similar to ethical problems with deception. But there is a further issue in many cases, shown by the Gospel case. The charge of “hypocrisy” made against the Herodians and Pharisees focuses on the aspect of their question which consists in their purporting to have a genuine interest in Jesus’ answer to their question. Not only do they seek to lure him into a vexed position, at the same time they seek to display themselves in a favourable light, as earnest and humble seekers after truth.

In general, hypocrisy seems to consist in claiming moral purity or uprightness that one does not really have. In claiming rectitude we do not possess, we are like people who give counterfeit money in payment of a debt: not only have we failed to meet our obligations, what we do adds the further evil of undermining people’s faith in the currency. As hypocrites, we not only fall short of virtue, we undermine people’s belief in virtue as a goal. Further still, in some cases, we betray individuals who accept our expressions at face value. For these reasons at least, a charge of hypocrisy is a serious one.

However, that is not all that is at issue here. Consider the Gospel case once more. The hypocrisy of Jesus’ questioners lay in them purporting to have a genuine interest in the answer to their question, because if they did have such a genuine interest, that would show them to have worthy interests and concerns and aspirations towards good policy and right action. In fact, of course, their actual motivation was not a genuine interest in the answer to their question. They had their own views about that issue, and firm commitments, which Jesus’ answer would not affect. Their actual motivation was to force Jesus into a difficult position. Their question, therefore, had several different functions. One was the surface function, of seeking information or opinion on the issue, for their guidance. A second was the ulterior function, of structuring a situation where any choice by Jesus would be problematic for him. A third was to put Jesus off his guard, perhaps to take less care with his answer than he might when he knew their intentions. Insofar as it is an attempt to put Jesus off his guard in a dangerous position, and lead him to expose himself to attack, the reasons to question it may not only be to do with hypocrisy and deception, but may be like reasons we have to question uses of force and violence.


There are multiple ethical issues which arise from the way that belief figures in group dynamics and related tactics of persuasion and influence. In conclusion, I wish to focus just on one of them, to take us back to the point at which we began.

I suggested that rational choice theory may have difficulty in accounting for the dynamics of group membership by reference to calculation of utility, just as it has such difficulty in accounting for some observed dynamics of Prisoners’ Dilemma simulations. But rational choice theory has two aspects. It is to some extent an approach to explaining people’s behaviour, and so figures in substantial parts of some modern economic theory. In addition, however, it often has a prescriptive element, suggesting that we can not only explain people’s behaviour by hypothetical calculation of utility, but also that we can work out what people ought to do by calculating what would maximise utility. The role of belief in group dynamics raises a question about the prescriptive element of rational choice theory as well as about the explanatory element.

It is plausible to believe that there are many cases where utility would be maximised by letting one’s beliefs be guided by group affiliations, rather than by the evidence.26 It would often be more comfortable for individuals to persuade themselves that what most people believe is what they ought to accept, rather than to stand out against the crowd. It would often avert costly and disruptive conflict if minorities would simply give up their idiosyncratic beliefs.

However, to expect people to adopt or renounce beliefs just to accord with others’ seems enormously problematic. What is more, it seems ethically problematic. It seems clear that there is sometimes an ethical problem with espousing beliefs for reasons to do with group membership rather than reasons to do with the beliefs’ intrinsic merits. To arrive at beliefs for the reasons we can see for them, rather than because of the social pressures on us, is part of what it is for us to be responsible, autonomous agents. To expect people to do otherwise is to expect them to forsake part of what makes them autonomous agents.

In cases of group conflict which involve differences of belief, it does not seem as though we can reasonably plot a course of action by referring just to calculation of interests. In that case, however, the question remains for us, what ethics does require. What is the role in ethical decision-making of the need to come to beliefs in a conscientious, autonomous way, as opposed to the need to avert the suffering that so often arises from group conflict? Perhaps we do not want to be fanatical ideologues, but neither do we want to be guided in our beliefs and affirmations only by group affiliations. How we can combine the need for dialogue in times of group conflict, with personal commitment and intellectual integrity? These are issues which we as ethicists need to grapple with. They are not easy issues. Rational choice theory does not deal with them effectively. It is up to us to do better.


1 In addition to references given last year, and especially by way of connecting this point with what follows below, see Montgomery, J.D. “Contemplations on the Economic Approach to Religious Behavior,” The American Economic Review, 86:2 (1996), 443–447.

2 “Cultural program to ease Sydney tensions”, ABC News Online,, accessed 21 May 2006.

3 Fiske, S.T. & S.E. Taylor, Social Cognition, 2nd ed., New York: McGraw-Hill, 1991, p. 123.

4 On this and what follows, see Baron, R.S., N.L. Kerr & N. Miller, Group Process, Group Decision, Group Action, Buckingham: Open University Press, 1992, pp. 135–44 and Brown, R.J., Group Processes: Dynamics Within and Between Groups, Oxford: Blackwell, 1988, ch. 8.

5 Montgomery, J.D. “Toward a Role-Theoretic Conception of Embeddedness” American Journal of Sociology, 104:1 (1998), 92–125, p. 93, citing Dawes, R.M. & R.H. Thaler, “Anomalies: Cooperation.” Journal of Economic Perspectives, 2 (1988), 187–197; Sally, D. “Conversation and Cooperation in Social Dilemmas: A Meta-Analysis of Experiments from 1958 to 1992” Rationality and Society, 7 (1995), 58–92; and Rabin, M. “Psychology and Economics” Journal of Economic Literature, 36:1 (1998), 11–46.

6 Montgomery, “Toward a Role-Theoretic Conception of Embeddedness,” citing Samuels, S.M. & L. Ross Reputations versus Labels: The Power of Situational Effects in the Prisoner’s Dilemma Game, unpublished manuscript, Stanford University, 1993; Ross, L. & A. Ward “Naive Realism in Everyday Life: Implications for Social Conflict and Misunderstanding”, in T. Brown et al (eds), Values and Knowledge, Hillsdale, N.J.: Erlbaum, 1995, pp. 103–135; and Ross, L. & A. Ward “Psychological Barriers to Conflict Resolution”, in M.P. Zanna (ed.) Advances in Experimental Social Psychology, Vol 27, San Diego, Calif.: Academic Press, 1995, pp. 255–304.

7 Fiske & Taylor, op. cit., p. 257.

8 Young, J., C.J. Thomsen, E. Borgida, J.L. Sullivan & J.H. Aldrich “When Self-Interest Makes a Difference: The Role of Construct Accessibility in Political Reasoning” Journal of Experimental Social Psychology, 27:3 (1991), 271–296.

9 See e.g. Brown, op. cit., pp. 223–4.

10 See e.g. ibid., pp. 91–106 and Baron, Kerr & Miller, op. cit., pp. 63–69. In what follows I draw to a significant extent on Chap. 11 of my Ethics and Organisational Politics, Cheltenham and Northampton: Edward Elgar, 2004.

11 See Grant, R.W. Hypocrisy and Integrity, Chicago: University of Chicago Press, 1997, pp. 48–50, although her statement accepts that only calculation of interests is rational, and that these mechanisms are not, which discussion in the present paper contests.

12 Bar-Tal, D. Group Beliefs, New York: Springer-Verlag, 1990, p. 105.

13 Ibid., p. 106.

14 Ibid., p. 107.

15 See e.g. Austin, J.L. How to Do Things with Words, Oxford: Clarendon Press, 1975, and Grice, H.P. Studies in the Way of Words, Cambridge, Mass.: Harvard University Press, 1989.

16 E.g. Bell, D.V.J. Power, Influence and Authority, New York: Oxford University Press, 1975, and Riker, W.H. The Art of Political Manipulation, New Haven: Yale University Press, 1986.

17 Mark 12:13–15 RSV; cf. Matthew 22:15–17, Luke 20:20–22.

18 Nineham, D.E. The Gospel of St Mark, 2nd ed., Harmondsworth: Penguin., 1969, p. 315.

19 Ibid., p. 315; see also Cox, G.E.P. The Gospel According to St Matthew, London: SCM Press, 1952, pp. 135–6.

20 Riker, op. cit, pp. 1–9. The question Lincoln put to Douglas was “Can the people of a United States Territory, in any lawful way, against the wish of any citizen of the United States, exclude slavery from its limits prior to the formation of a state constitution?” In this case as in the Gospel case and many others, we need a detailed understanding of the political situation to see the significance of the question. It is explained in this case by Riker.

21 The importance of a position on an issue as a symbolic rallying-point is one reason why in negotiation it may sometimes not be a good idea to focus exclusively on “interests”, rather than positions, as Fisher and Ury recommend in Getting to YES (Boston: Houghton Mifflin, 1981): where a group is party to negotiations, its position on an issue can be important for its members’ solidarity and commitment. See Provis, C. “Interests vs Positions: A Critique of the Distinction” Negotiation Journal, 12:4 (1996), 305–323.

22 Matheson, P. Cardinal Contarini at Regensburg, Oxford: Clarendon Press, 1972. For further discussion, considering how the episode shows the difficulty there can be distinguishing “negotiation” from “argument”, see Provis, C. “Negotiation, Persuasion and Argument” Argumentation, 18:1 (2004), 95–112.

23 Matheson, op. cit., p. 5.

24 Ibid., p. 117.

25 Bar-Tal, op. cit, p. 106.

26 Although Stocker suggests that “intellectual activities done for the goals and according to the criteria inherent in them are among the greatest human goods” (“Intellectual Desire, Emotion, and Action”, in A.O. Rorty (ed.) Explaining Emotions, Berkeley: University of California Press, 1980, pp. 323–338, p. 337), and if we accept that forming beliefs is an “intellectual” activity there may be a way to pursue an idea that it is in people’s interests to form beliefs on the basis of evidence. While I think that line of argument fails, I cannot pursue the point here. On what follows see also Stocker’s “Responsibility Especially for Beliefs” Mind, 91 (1982), 398–417.