ISSA Proceedings 2006 – Duets, Cartoons, And Tragedies: Struggles With The Fallacy Of Composition

A fundamental problem arises concerning much of our language about groups. The problem is this: we apply to groups the intentional language of emotions, attitudes, and beliefs. Such language is paradigmatically individual in application and yet we apply it to groups of all sizes – small, medium, large and very large – and of varying degrees and kinds of organization. In important contexts, we refer to groups not only as doing things and being accountable for what they do, but as having attitudes and intentions related to their actions. Groups may be said not only to undertake actions but to be resentful, hateful, generous, compassionate, accepting, suspicious or trusting. They may be said to hold beliefs and make value judgments, and reach decisions on the basis of these. Corporate boards and parliaments, for example, are organized groups empowered to act for still larger groups. They take decisions and act – and when they do so, it is on the basis of beliefs and attitudes which underpin their intentions and actions. Suppose, for instance, that a corporate board reaches a decision to spend millions on exploratory drilling in some area of the Arctic. Why? Its decision is made intelligible on the grounds that it knows the price of oil to be high and rising, and has evidence implying that the area in question contains oil. Or a parliamentary body might reach a decision to send peacekeeping troops to a particular country, on the basis of beliefs about the risks and needs of the people in that country, and the feasibility of its troops making a constructive difference in that context.

For those who contest the observation that intentional language is commonly applied to groups, I suggest a reading of journals and magazines containing commentary about economic and political affairs. You will find many attributions of actions to groups and you will find that these actions are rendered intelligible in much the way we make individual actions intelligible, namely by attributing beliefs, attitudes, and values to groups. My particular interest in this area stems from work on challenges of political reconciliation, and from seeing how questions about compositional attributions arise in that context. However, as the preceding examples will show, compositional attributions are by no means restricted to that sort of context.
For convenience, let us call the application of intentional language to groups the compositional phenomenon. The compositional phenomenon strikes many people as highly problematical. Many have raised difficulties about it, saying that it cannot possibly make sense for groups to think, feel, believe, and decide. Why not? Because groups are not conscious; there is no group mind. Some go even further, contending that groups cannot do anything, qua groups, and cannot properly be held accountable for their actions. (Miller 2001) This claim strikes me as implausible to the point of perversity, and I will not explore it here. I will assume that groups, small or large, organized or not, can do things. In fact there are some things that can only be done by groups – performing choral works, reaching a jury decision, winning a soccer game, and passing laws in parliament being obvious examples.
In discussions of group conflict and its resolution, the compositional phenomenon is quite conspicuous. We find, for example, allusions to distrust, trust, apology, forgiveness, and reconciliation as phenomena in politics, in the relations between groups. (Govier 1997) Does such discourse make sense? Can we engage in it without systematically committing mistakes of logic and metaphysics? These questions will be the focus of this presentation.What I have in mind here is the Fallacy of Composition, in which we mistakenly infer conclusions about wholes or groups from premises about parts or individuals.

In this presentation, I consider a number of themes related to the compositional phenomenon. First, I consider several responses that would purport to eliminate it. I then move on then to set it in the context of theory of argument. The view I will take is that there really is a problem here, the Fallacy of Composition is genuinely a fallacy, and an important one – but that the gap underlying this fallacy can be plausibly bridged in some cases.

Some Preliminary Metaphysics
As discussed here, the problem of compositional attributions begins from the supposition that, with respect to intentional language, group attributions are problematical whereas individual attributions are not. This casting of the problem will seem correct to many. Nevertheless, there are several ways of resisting the dichotomous contrast between individual and group that constructs this problem. First, the individual can be regarded as a kind of plurality or collectivity. (Graham 2002) Hume, for instance, famously compared the self to a commonwealth. Seeking to understand personal identity, Hume argued that we attribute it on the basis of relations of resemblance and causation between the distinct perceptions of the mind. Stating that impressions cause ideas, which then cause further impressions, Hume said, “In this respect, I cannot compare the soul more properly to any thing than to a republic or commonwealth, in which the several members are united by the reciprocal ties of government and subordination, and give rise to other persons, who propagate the same republic in the incessant changes of its parts. And as the same individual republic may not only change its members, but also its laws and constitution; in like manner, the same person may vary his character and disposition, as well as his impressions and ideas, without losing his identity. Whatever changes he endures, his several parts are still connected by the relation of causation.” (David Hume, Treatise of Human Nature, Chapter 35.) Hume offered this comparison not as an argument from analogy, but rather as an explanatory illustration of his theory about causal relations among the distinct perceptions that constitute a human mind.
Hume, then, endorsed an account in which individual selves were compositional. As illustrated in the comparison between the self and the commonwealth, Hume argued that individuals are composite; the implication here seems to be that there is no categorical difference between the individual self and some composite entity such as a republic or commonwealth. If we were to endorse such an account, we might conclude that the dichotomy between group and individual levels of analysis be resisted.

A differently oriented approach can provide different grounds for the same conclusion. Often emotions and attitudes that are attributed to individuals presuppose interactions with other persons (Graham 2002), or are themselves the product of cultural patterns and responses. For instance, an individual who is suspicious of persons in another ethnic group may hold these attitudes because of beliefs and feelings acquired from traditions in the culture. To some extent, people believe, feel, and think as they do because of enculturation. (Govier 1997) Rather than presuming that we need to explain group attitudes by arguing up to macro from micro, one could insist that explanation goes in the opposite direction, downward from macro to micro. There are, of course, variations in individual responses to cultural traditions.While one person may inherit racial prejudice from his culture, another may find it repugnant and be motivated to struggle against it. (Cohen 2001, Moody-Adams 1997)
These broadly metaphysical considerations argue against any exclusive and exhaustive dichotomy between individual and collective. But such considerations are too general to defeat the concerns of those who find compositional attributions problematic. They do not address the specific gaps when evidence about individual persons (whatever their metaphysics) is cited as support for conclusions about groups of such persons. As we shall see, many arguments for compositional attributions are weak, falling into the well-known trap of the Fallacy of Composition.

On a Pragmatic Level: Three Disputed Responses – and a Further Proposal
Apart from these broadly oriented metaphysical arguments, there are three further reactions to the compositional phenomenon as it is commonly constructed. These are:
1. The Forbidding Response. On the forbidding view, all intentional language, as applied to groups, is based on error; compositional attributions should be banned because intentional language applies paradigmatically to individuals. It should not be extended to groups, because groups are not conscious and are thus not the sorts of entities that can have beliefs, attitudes, and feelings. (Miller 2001)
2. The Legitimating Response. On this view, intentional language as applied to groups must be legitimate because it passes the only realistic and sensible test of legitimacy – namely actual use. Along the lines of ordinary language philosophy and the later Wittgensteinian philosophy, which stated that ordinary language is all right as it is, one might simply resist any systematic criticism of standard practice. (Wittgenstein 1963) After all, we regularly employ compositional attributions when they interpret and respond to actions and events; given that they do so, compositional attributions are functional. To seek to reform ordinary language on philosophical grounds would be misguided and futile.
3. The Discriminatory Response. On this view, there are indeed contexts in which intentional language applies to groups. We know from experience of war and intense conflict that nations and groups are often suspicious of each other and harbour feelings of resentment and hostility, based on felt grievances about wrongs of the past. That groups and nations have often had relationships characterizable in these ways are established facts of history. Such considerations are part of standard lore in the so-called realist school of international relations. Distrust and fear are frequently said to characterize relations between nation states. On the Discriminatory account, such negative attributions are allowed but if we attribute such traits as compassion, generosity, forgiveness, and trust to groups, that goes too far in the direction of idealism, being too emotional and value-laden to be realistic. Positive intentional attributions must be resisted or systematically reinterpreted as manifestations of self-interest. On the Discriminatory account, it is insisted that ethically positive traits are purely individual.

I submit that all three of these responses are open to criticism. An objection to the forbidding response is that it is dogmatic, inflexible, and unrealistic given standard practice. An objection to the legitimating response is that its confidence in ordinary language goes too far in avoiding explanation and justificatory argument. An objection to the discriminatory response is that it is biased toward the negative. This response is grounded more in a Hobbesian attitude to the social world than in a sound theory of logic and language. Consistency indicates that if we can make sense of a nation distrusting, we can make sense of a nation trusting; if we can make logical and epistemic sense of a group resenting, we can make sense of a group forgiving;

In this presentation, I develop a fourth approach, along the following lines.
4. Compositional Construction, or Gap-filling. On this view, compositional attributions pose questions, especially when claims about group actions and attitudes are based on evidence about individuals. Real issues arise. The challenge is to acknowledge the gap and the problem, and find ways in which the gap can be bridged.

The Fallacy of Composition: Examples and Comments
To relate this problem more specifically to issues about argument and argumentation, I move to consider the Fallacy of Composition. I might add here that this fallacy has been strikingly memorialized in a sculpture by that name at the University of Groningen. This sculpture, a lighted structure, by Trudi van Berg and Jos Steenmeijer, occupies most of a wall on the building for the Faculty of Economics.
As is well-known, the Fallacy of Composition is committed when there is reasoning from premises about parts to a conclusion about a whole. There are many interesting instances of this fallacy, and many important questions, that arise in material and physical contexts. Here, I consider primarily social phenomena, given my interest in compositional attributions to groups of people. In the social context, instances of this fallacy typically involve premises about individuals and conclusions about some group of which those individuals are members. There are, of course, many examples of such flawed arguments. I will mention several instances here.

The Duet: John is a terrific tenor and Susan is a brilliant soprano. So a duet by John and Susan will be superb.
The Cartoons: A Danish newspaper, under a particular editor, publishes some cartoons that are found to be offensive by some Muslims. Through this action, Denmark has offended Muslims.
The Tragedy of the Commons: If one farmer grazes his cattle on the commons, that will be beneficial for him; therefore if all the farmers graze their cattle on the commons, that will be beneficial for all.
The Dinner Party: No one would set out dinner for her family and exclude one member from the table, refusing for no good reason to allow the ostracized person to eat. You can see from this that it is wrong for some human beings to have inadequate food while many others enjoy good meals. Therefore the world community should accept responsibility for world hunger.
The Utilitarian: Each person desires his own happiness, and each person’s happiness is in that way a good to that person. Therefore the general happiness is a good to the aggregate of persons.

In the Fallacy of Composition, the basic mistake is not merely quantitative. It arises from the fact that there are often significant structural differences between the micro and macro level. We go astray if we reason so as to fail to consider those differences. In the social context, which is our concern here, there are significant differences between individuals as such and groups comprised of these individuals in relationship with each other. Individuals in groups stand in relationships to each other and interact – sometimes cooperatively, sometimes conflictually, sometimes when occupying institutional roles, sometimes according to various habits and expectations. (May 1987) The nature and quality of the interactions between individuals in a group affects that group – even when it is very small, as in the case of the duet. If we reason that (simply) because John and Susan are both good singers as individuals, they will be good as a duet, we have ignored the fact that to present a successful duet, these two have to harmonize and work together. We have made a mistake, ignoring complications and complexities which may arise from their need to work together.
In terms of the theory of argument, it is interesting to note that the Fallacy of Composition can appear in arguments of different types. If an argument is taken to be deductive, and the premises are about individuals while the conclusion is about a group, clearly that argument will be deductively invalid in the straightforward sense that it will be possible for the premises to be true while the conclusion is false. We may locate the Fallacy of Composition within this gap. If an argument is taken to be an analogical argument in which the primary subject is a macro phenomenon, while the analogue is described at the micro level, the analogy will be inadequate because there are relevant differences between the analogue and the primary subject. We consider the Fallacy of Composition in considering the nature and relevance of these differences. If an argument from individual to group is taken as inductive generalization, it can be criticized as hasty; the individual cases do not give sufficient evidence about the group as a whole. If it is regarded as an inference-to-the-best-explanation, there will be doubts about whether a compositional attribution to a group does, indeed, provide the best explanation of the possession of characteristics by an individual or individuals, given that individuals within the group may differ from each other and can exert a certain degree of autonomy.
Concerning the gap constitutive of the Fallacy of Composition, there are two crucial factors to be considered.
(a) The problem of less. The individuals, considered simply as individuals, are less than the group considered as such, because they do not stand in relationships to each other, do not interact, cannot be said either to cooperate or to be in conflict, and are not organized institutionally.
(b) The problem of more. The individuals, considered as such, are more than groups as such, since individuals have something every group lacks, namely consciousness. An individual can literally, by himself or herself, think, reflect, plan, choose, feel, amend her feelings and so on. No group has consciousness in the literal sense in which an individual has consciousness.
In pursuing the gap-filling approach, I will return to these basic problems of less and more.

Reducing Composition to Something Else?
But first it will be useful to consider some approaches that will be resisted here.
In a version of the legitimating response to our problem, the very notion of a Fallacy of Composition may be contested. For example, one might say that there are recognized figures of speech in which one element serves to represent the whole – as when we say “all hands on deck” or “give us this day our daily bread”. The figure of speech here is that of synecdoche. And in these familiar expressions, it is quite clear what is being said. The hand represents the person of a crew member and the bread represents the nutritional needs of people. Surely these things are understood and only the most pedantic person would object to these ways of talking. Synecdoche, one might say, has been around for a while and is an unobjectionable device.

Within political discourse, consider this statement: ‘Berlin opposes Washington on Iraq.’ In this locution, we find synecdoche insofar as the capital cities are named to represent the people of nation states. Pedantically we can spell it out: to say that Berlin opposes Washington on some matter is to say that Germans, as represented by their government in Berlin, disagree with Americans, as presented by their government in Washington, on policies regarding Iraq. One might insist that what is said is surely understood and perfectly legitimate; there is no problem here, we know what is meant, and synecdoche is an established mode of speech. But wait a minute: this case, unlike that of the hands on deck, this claim about Washington and Berlin involves a compositional attribution. There is does seem to be some amount of philosophical mystery in the matter. What does it mean for a nation or collectivity (Germany, or Germans) to disagree with another nation or collectivity (the United States, or Americans)? How are we to understand such claims? What sorts of evidence would support them? This is the compositional problem. The fact that we understand synecdoche in some other contexts does not make the compositional problem disappear in this sort of context.

It is sometimes said that the Fallacy of Composition has to be judged case by case and is in this respect a ‘material’ fallacy and not a formal one. (In this context, ‘formal’ and ‘general’ should not be confused. My treatment claims to be general, but not formal.) (Govier 1987, 1999). I leave the social sphere to find a simple example here. Consider, for instance, the case of a uniformly brown cookie; say it is a peanut butter cookie and its ingredients have been well mixed by the cook so that all its visible parts are brown. If we were to reason that because all the visible parts of the cookie are brown, the cookie itself is brown, we would reach a true conclusion. Yes indeed. However this result does not mean our argument from parts to the whole avoids errors in reasoning. We got to the true conclusion by luck alone. It does not follow from the fact that we sometimes get lucky and arrive at a true conclusion that the Fallacy of Composition is material and has to be understood on a case-by-case basis. There is still something wrong with the argumentation scheme in the case; there is a problem with any general scheme reasoning from parts to whole with no gap-bridging device. That we are lucky in some cases, because in those cases the shift from micro to macro happens in this instance not to be negatively relevant to the conclusion, does not show that the general scheme is rationally defensible.
Perhaps what is going on in compositional attributions is akin to, or an instance of, stereotyping. We too easily form a ‘them’, where instead distinctions and divisions are needed. In some cases, our simplistically formed category of ‘them’ serves to buttress the polarization or even the demonization of an ‘out-group’ as contrasted with an ‘in-group.’ The basic mistake here is that a group is cast according to the attributes of some few individuals within it. Although some generalizations about groups may hold true, statistically, there are individuals within a group who do not fit the stereotype. And furthermore even a description that applies to a majority of individuals within a group may not apply to the group considered as a collectively.

The notion of stereotyping seems to fit the case of the Danish cartoons. Initially it was one editor who chose to commission and publish the contested depictions of Mohammed. This man, Flemming Rose, commissioned the drawings for a children’s book, and did that for reasons of his own. Rose suspected that Danes were self-censoring in their comments on Islam and Islamism because they were afraid of intense reactions, including physical violence, by radical Islamists. He wanted to find whether people would be bold enough to make some drawings and send them in. Rose said, “I commissioned the cartoons in response to several incidents of self-censorship in Europe caused by widening fears and feelings of intimidation in dealing with issues related to Islam.” (Rose 2006) Flemming Rose was one individual in one particular situation, with his own quite specific goals and concerns. In the initial situation, there was little reason to deem him typical of Danes generally; nor was he in any way authorized to represent Danes as a collectivity. In their response to the distribution of the cartoons, some Muslims in some countries rioted, burned embassies, and advocated boycotts of Danish products on the ground that the cartoons were blasphemous and offensive. It is by no means clear that Flemming Rose offended Muslims in general. But even if we say that he did, a vast leap is made attitudes attributed to Rose are attributed to Danes more generally. Flemming Rose is not all Danes or most Danes; still less so did he represent the state of Denmark. (As embassies and products of Denmark were attacked, Danes began to rally to support Flemming Rose. At that point it could be more plausibly argued that ‘Denmark’ supported his actions; this scenario seems characteristic of the polarization underlying serious group conflict. )

One of the strongest objections was to a particular cartoon depicting Mohammed wearing a turban with a bomb in it. If Mohammed is represented as a terrorist and is the prophet of this religion, then, one might say, that the person who drew this particular cartoon was himself guilty of stereotyping because in his representation of the bomb in the turban, he implied that all Muslims are violent terrorists. About this suggestion, Rose commented, “Angry voices claim the cartoon is saying that the prophet is a terrorist or that every Muslim is a terrorist. I read it differently: Some individuals have taken the religion of Islam hostage by committing terrorist acts in the name of the prophet. They are the ones who have given the religion a bad name.” It did not escape the attention of commentators that violent reactions to the stereotyping of one’s group as violent only serve to confirm the very stereotype that one protests. (Fatah 2006) But then this whole matter is not, fundamentally, one where we would expect logic to reign supreme.

Some of these reflections suggest an inductive interpretation of the Fallacy of Composition, according to which we would assimilate it to another fallacy, that of Hasty Generalization. Leaving the cartoons and conflicts surrounding them, I turn here to a dispute regarding the South African Truth and Reconciliation Commission. Many of the TRC’s early defenders – including Archbishop Desmond Tutu himself – emphasized stories of individual forgiveness and reconciliation, and then went on to speak of national reconciliation between black and white South Africans. (Tutu 1999) The logical gap is apparent here. But what is its nature, exactly? Is the problem simply that there were not enough individual stories… the sampling of cases was not large enough, and possibly not representative, so that there is a problem of hasty generalization? To generalize to ‘most’ or ‘all,’ we need more of the some – and that is the problem? I do not think that is quite the problem here. Getting more of the some would not suffice, because it would not address the issue of level shift, from micro to macro, from relationships between individuals to relationships between large groups. For a group to forgive another group, or to reconcile with it, group processes are required. If we are to say that there is some kind of reconciliation between groups that have previously been opposed, then we have to be able to speak of the attitudes of these groups (either aggregatively or collectively) and we have to characterize them as shifting in ways that are reconciliatory. Compositional problems arise here as they do not if our concern is straightforwardly a matter of Hasty Generalization.
It is sometimes suggested that the Fallacy of Composition can be understood as involving Equivocation. On this account, there is a shift of meaning when we move from micro to macro level. If we use the same terminology in both contexts, we ignore this shift, and reason on the basis of an equivocation. (Van Eemeren and Grootendorst 1992) For example, individuals may remember things, may experience traumas, and may work through those traumas in a quest for healing. People speak, as well, of the need on the part of nations and groups to remember aspects of the past and work through traumas that have been experienced by the nation, and heal. But what does such language mean in the context of a collectivity? There has to be a shift in meaning, and when we make compositional attribution, we ignore that fact. On this interpretation the Fallacy of Composition might seem to be reducible to another fallacy, that of Equivocation.
As with the brown cookie, there are instances in which an answer to these questions seems easy to come by. Consider, for instance, the case of acknowledgement. Many discussions of post-conflict processes call for acknowledgement, by nations and groups, of wrongs committed by agents acting on their instruction and behalf. And nations and groups really can acknowledge; it is easy to see what this means. A nation can, for example, establish memorial days, commission sculptures, build and maintain museums, issue official statements of apology and recognition, and establish institutions for funding projects. It is thereby acknowledging various historical facts, and committing itself to value judgments about them. So far as policy and expressive artifacts are concerned, collectivities are likely to have greater resources and more power than individuals. Individuals can acknowledge too. They typically do so by making statements of admission expressive of their beliefs and attitudes, and in the case of wrongdoing, those admissions allow that the acts were wrong, were done culpably, and should not be repeated. Groups are not disadvantaged compared to individuals when it comes to acknowledgement; in fact, given their greater resources, they may be more able to acknowledge and memorialize than are individuals.
But the fact that in this particular case and some others compositional attributions seem unproblematical only suggests a more general solution; it does not in itself provide one. General questions about the legitimacy of the shift have not disappeared. What would it mean for a nation to remember? To forgive? To show concern and generosity? To deal with its past? To reconcile? To say that there may be equivocation, that there is an alteration in meaning when we proceed from micro to macro, remains true for many cases. But these observations about equivocation do not fully handle the problem. What is the shift? What sorts of evidence (if any) can justify compositional attributions? The gap remains and must be bridged. How do we do it?

Ways to Bridge the Gap
There are human actions that are not the actions of individuals. These actions include such things as the singing of choral works, the waging of wars, and the conducting of national electoral campaigns. These are actions and they are human actions. It is people, human beings, who do these things. And people do not and in many cases cannot do them as individuals. So how do human beings do these things? How do we manage to sing the choral movement of Beethoven’s Ninth Symphony? Conduct an election? Or dispatch troops to fight in a distant country? The answer is obvious: we do these things in organized groups, in which there are procedures and practices.
Suppose that the organizational structure is tight enough that a large group has a smaller sub-group authorized by its rules to deliberate and act in a range of cases. Let us call this sub-group the executive. Suppose further that the executive conducts deliberations in which people speak and reason together and reach decisions on the basis of its proceedings. In these deliberations, individuals put forward ideas and arguments and other individuals respond to them. Assuming even a modicum of democratic process in the case, the reasoning and decisions of the group are not necessarily those of any individual within it. There will be exchanges of information and judgment, argument, dialogue, and dialectical developments. The process in which various people make and respond to claims and arguments engages a number of people, and their arguments and responses affect each other. The decision may be said to emerge from the deliberations of the group, and may be deemed to be a joint decision. (Gilbert 1987)
Suppose, for example, that the executive of a political action group decides not to send messages out to members using the national postal system. It reaches this decision after deliberations involving considerations about possible delays and losses that its members claim to have occurred within in that system. Its decision with regard to this matter indicates an attitude that may be attributed to the executive. Its attitude is one of distrust in the postal system. If the executive decision is known to the larger group and not opposed by them, thereby being tacitly accepted, we can attribute the attitude to the larger group. To consider another illustration, suppose the executive of a judges’ organization meets to consider criticisms of a number of judicial decisions on matters pertaining to gender and its deliberations cumulate in an executive decision to organize workshops to educate judges on the matter. Let us suppose that the executive comes up with a policy and recommendation for action. Given this decision by the executive, certain beliefs and attitudes can be attributed to it. For example, if the executive is recommending educational workshops for judges, on gender themes, it must believe that judges need more information and training about gender and legal process, and that these workshops could provide them. Given its authorized role, the beliefs and attitudes attributed to the smaller group may also be attributed to the larger group, presuming that most do not object given information about this initiative. By their failure to object, they may be said to indicate tacit consent to these policies and to the beliefs and attitudes indicated by them.

Relationships and processes affect results. I am proposing that in such cases the gap between individual attitudes and those of the group may be bridged by the facts of group process. What A,B,C,D, and E come up with after meeting together emerges from their discussion and – because it emerges in this way – is distinguishable from what any one of them would have come up with individually. There is something distinctive about the process in which the decision has been reached, because it has involved these individuals in relationship to each other. (Gilbert 1987, Graham 2002) The decision or action that results from the deliberations of the executive is a group product, attributable to the executive because it is a product of the interactions of its members, and attributable to the larger group if they tacitly consent. Because the decision or action can be attributed to this group, the intentional attitudes and beliefs implied can also be attributed to it.
The two members of a duet can speak directly to each other, but large groups cannot deliberate face-to-face. Canada cannot have a discussion except insofar as some representative persons have discussions in some contexts, and these discussions are publicized and become public. An obvious possibility is that of an explicit and authorized political process. If the context is that of the House of Commons in Ottawa, these participants are representative of the Canadian public because they have been elected in a process that is broadly accepted as legitimate. Given representativeness and tacit consent, policies adopted in the House of Commons can be regarded as those of Canada. Insofar as these policies are understood and stand unopposed, they can be attributed to Canada as a collectivity. The collectivity has engaged in deliberations and actions through its representatives.

A complication arises at this point. Where there is no group process, the problem of compositional attributions cannot be solved in this way. (May 1987, Graham 2002) What about more loosely organized groups or groups that are scarcely organized at all? It would seem that unorganized groups can act – as they do in various forms of street demonstration and protest. A recent example is that of extensive protests in Paris, with regard to the proposed law on youth employment. In some cases of street protest, people come together without there being a clear organizational structure constituting them as a collectivity. We may consider cases in which there is nothing like a designated executive enjoying powers granted by a collectivity in which persons are members or not. Suppose, for example, that 200,000 people have gathered in the center of Paris to express their discontent with a proposed law, and many of them are carrying signs and shouting slogans against that law. Given that participation in the protest is voluntary, given the context and the reasonable supposition that the meaning of signs and slogans is understood, it makes sense to attribute to these persons attitudes of opposition to the proposed law. (Indeed, the attribution of such attitudes is already implied when we describe a crowd as protesting the proposed law.’)
But suppose now that one hundred or so of these people begin to engage in property violence. Let us say that they throw stones and smash the windows on cars shops. And suppose that such persons are a minority. Should we say that the protesting youth are engaging in property violence? That they are threatening, destructive? My account here would have the implication that these further attributions cannot be justified unless there is further evidence, according to which we would have grounds for attributing these attitudes to most of the individuals present or to the group as a whole. How do those present respond to the violence? Do they indicate support by cheering and joining in? Do they indicate opposition by shouting out against the violence or trying to prevent it? Or by leaving the scene? Do they indicate ambivalence and embarrassment by standing awkwardly by? If there is no predominant pattern of response in such a case, given that there is no representative executive to speak for the group, we cannot attribute either approval or disapproval.
Clearly, my account of gap-bridging presupposes that there is organization within the group. When representativeness and tacit consent are less clear, it is difficult to justify attributions to the group as a whole or even to a majority of its members.
I have argued here that there is an important sense in which compositional attributions are problematic. When premises are about individuals and conclusions are about groups, there is gap in the argument. The existence and understanding of this gap underpin the tradition of the Fallacy of Composition. I have maintained here that this fallacy is genuine and important, and I believe there is much to learn by logically probing claims about ‘the Danes,’ ‘the West,’ ‘Muslims,’ and so on. Stereotypes, hasty generalizations, and unclear language often underlie simplistic polarization, at a cost both to accurate understanding and to decent relationships. For all the qualifications we may make about the individual/group dichotomy and the clarity of some concepts, there is a problem of compositional attributions. But I am arguing against any notion that all such attributions should be resisted. On the contrary, I have claimed that some of them are unobjectionable because they can warranted by a line of argument in which the gap is bridged. This warranting is most straightforward when groups are organized.

The gap defining the Fallacy of Composition can be bridged when group structures and relationships provide contexts for people to think together and act on the basis of their joint deliberations. We can understand how the deliberations and actions of an interactive group provide grounds for attributing to it attitudes and beliefs. By these mechanisms, the problem of too little is addressed. It does not matter that the group itself does not have consciousness, because intentional attitudes can be correctly attributed to it on the basis of interactions between its members. Then, in virtue of representativeness and tacit consent, we can see how those attitudes and beliefs can also be said to characterize a larger group. These features show how the problem of too much is resolved. Putting together emergence, representativeness, and tacit consent, we are able to bridge the gap constitutive of the Fallacy of Composition as it applies to groups and individuals.

REFERENCES
Cohen, S. 2001. States of Denial: Knowing about Atrocities and Suffering. Cambridge, UK: Polity Press 2001.
Dwyer, Susan 2003. “Reconciliation for Realists” in C.A.L. Prager and Trudy Govier, editors, Dilemmas of Reconciliation. Waterloo, ON: Wilfrid Laurier University Press 2003.
Eemeren, Frans H. van and Rob Grootendorst 1992. Argumentation, Communication, and Fallacies. Hillsdale, NJ: Lawrence Erlbaum Associates.
Eemeren, F.H. van and Rob Grootendorst and F. Snoeck-Henkemans. 1996. Fundamentals of Argumentation Theory: A Handbook of Historical Backgrounds and Contemporary Developments. Mahweh, NJ: Erlbaum Associates.
Eemeren, F.H. van and Rob Grootendorst, 2003. A Systematic Theory of Argument: The Pragma-Dialectical Approach. Cambridge, UK: Cambridge University Press.
Fatah, Tarek 2006. “What Would the Prophet Have Done?” Globe and Mail, February 5.
French, Peter 1984. Collective and Corporate Responsibility. New York: Columbia University Press.
Gilbert, Margaret 1987. “Modelling Collective Belief.” Synthese 73, 105 – 203.
Govier, Trudy 1987. Problems of Argument Analysis and Evaluation. Dordrecht, NL and Berlin: Foris/ de Gruyter.
Govier, Trudy 1997. Social Trust and Human Communities. Kingston and Montreal: McGill-Queen’s University Press.
Govier, Trudy 1999. The Philosophy of Argument. Newport News, NJ: Vale Press.
Govier, Trudy 2002. Forgiveness and Revenge. London: Routledge 2002.
Hume, David 1975. A Treatise of Human Nature, edited by L.A. Selby-Bigge, second edition revised by P.H. Nidditch. Oxford: Clarendon Press.
Johnson, Ralph 2000. Manifest Rationality: A Pragmatic Theory of Argument. Mahweh, NJ: Erlbaum Associates.
May, Larry 1987. The Morality of Groups . Notre Dame: University of Notre Dame Press.
Moody-Adams, Michele 1997. Fieldwork in Familiar Places: Morality, Culture and Philosophy. Cambridge, MA: Harvard University Press.
Rose, Flemming 2006. “Why I Published Those Cartoons,” Washington Post February 19.
Tutu, Desmond 1999. No Future Without Forgiveness. New York: Doubleday.
Walton, Douglas and John Woods. 1975. “Composition and Division” Studia Logica 36, 381 – 406.
Walton, Douglas. 1995. A Pragmatic Theory of Fallacy. Tuscaloosa, Alabama: University of Alabama Press.