Keywords

1 Introduction

The ontology of any domain ought to give an account of what there is in that domain, in particular of what there is that counts as interesting from one or another point of view. What counts as interesting from one viewpoint, of course, may not count as interesting from another. The farmer will give one account of what there is to be found in a field, the botanist another, the painter a third. The farmer will focus on the plants in the field; the botanist on the different vegetative life-forms, weeds as well as plants; the painter on the varieties of texture and color that those plants and weeds display against the background of soil and sky.

The domain of social ontology comprises the interactions of individual human beings together with the patterns that constrain those interactions or that emerge from them. The interactions relevant, so I shall assume, are those that involve the intentional attitudes of participants: that is, the attitudes that bulk large in the psychology of persons—whether scientific or common-sense psychology—such as belief and desire, judgment and evaluation, intention and policy, emotion and mood (Mellor 1982). Thus the interactions and associated patterns relevant to social ontology include our interactions as friends and foes, consumers and producers, compatriots and foreigners. But they do not extend to interactions that are explicable in wholly sub-personal terms: for example, interactions of epidemiological contagion, pheromonal stimulation or competition for oxygen.

But if this is the domain of social ontology, what is the viewpoint that informs it, making some questions salient, others not? I think that in the traditional and contemporary literature of the discipline—so far as it has a recognizable profile as a discipline—the viewpoint is shaped by an interest in the significance of our social interactions, and of the groups we form in social interaction, for our status as minded agents, guided by intentional attitudes.

There are three main questions that this interest has stimulated and, using terms in a somewhat stipulative sense, I describe them in turn as the individualism issue, the atomism issue and the singularism issue. In this paper I focus on each of these questions in turn and, drawing on earlier work, gesture at some arguments in favor of the positions I adopt. The presentation is excessively condensed but it may offer a useful overview of the field as a whole. I conclude with a brief discussion of the significance of these issues.

2 The Individualism Issue

2.1 History

The individualist question, which came into prominence only in the nineteenth century—and has perhaps lost its hold on our contemporary sensibility—is whether the forces associated with social life, in particular the forces that social science is liable to unearth, entail that the intentional attitudes posited in personal psychology are not always the forces that move us to action. On at least some fronts we are pawns of unrecognized social forces, so anti-individualists suggest, not the intentionally guided or autonomous agents we take ourselves to be.

Ian Hacking (1991) argues that as social science began to make an appearance in the nineteenth century, it was shaped in great part—and perhaps even called into existence—by the plethora of social statistics that began to appear as a result of the rise of the administrative, bureaucratic state. From about 1820 on the state in various European and other countries began to record and publish figures on, for example, the aggregate incidence of crime, insanity and suicide, poverty, illness and mortality. And they thereby revealed the rates at which these statistics changed across time and place, if indeed they did change, as well as their correlation or lack of correlation with one another. Many assumed that such rates and correlations would vary more or less at random, given the presumptively random way in which individuals resolve intentional issues and exercise free will. But the data gave the lie to that assumption, revealing unexpected constancies and unexpected degrees of predictability in people’s social behavior.

According to Hacking’s narrative, the discovery of these unexpected constancies led a great variety of European thinkers to the conclusion that there was a hidden hand at work in social life. This was not the invisible hand of the market that Adam Smith (1976) had charted in the eighteenth century; that is, not a mechanism whereby individual interactions, psychologically intelligible in themselves, would reliably give rise to certain aggregate patterns. And it was not the iron hand of the state: that is, not a mechanism of control intentionally exercised from above. The hand that these thinkers saw at work in the statistical constancies they espied was a much less obvious, and ultimately a much more ominous, force. It was a source of pre-determination in people’s behavior that put in question the minded status, and the intentional autonomy, posited in our ordinary psychology and experience of ourselves.

The imagined source of pre-determination was sometimes compared to the silent force of gravity that shapes the movements of astronomical bodies. The idea was that just as the heavenly bodies are forced to move in the patterns that gravity dictates, without any evidence of active push or pull, so we ordinary human beings may be subject to equally silent and equally inescapable forces, being driven unwittingly to display certain socially ordained patterns of behavior. This sort of social determinism was endorsed in a variety of forums. It shaped T.H. Buckle’s History of Civilization in England, published to great acclaim in 1857. And it assumed a vivid, theatrical form in the vision presented in Tolstoy’s War and Peace, written between 1863 and 1869. Tolstoy (1972, 1313) writes: ‘Ever since the first person said and proved that the number of births or crimes is subject to mathematical laws, that certain geographical and politico-economical laws determine this or that form of government, that certain relations of population to the soil lead to migrations of people—from that moment the foundations on which history was built were destroyed in their essence’. It became impossible, so he suggested, ‘to continue studying historical events, merely as the arbitrary product of the free will of individual men’.

This sort of social determinism sponsored the appearance in late nineteenth-century France of a science of society—a sociology, in the name given it by the philosopher, Auguste Comte—that would reveal the laws governing social life. The great protagonist of this movement was Emile Durkheim, one of the founders of sociology as it we know it today. While developing a body of work that is valuable by almost any lights (Lukes 1973), Durkheim nurtured the aspiration to replace the sense of ourselves present in commonsense psychology—and in many scientific versions of the discipline—by displaying the social forces at work amongst us. He took those forces to operate on us coercively, in a way that bypasses our sense of what we do and why we do it, via a variety of what he called social facts. These include features of our society like the density of population, the norms and rules institutionalized there, the currents of opinion that prevail at any time, and the enthusiasms that occasionally sweep across a group. ‘A social fact’, he says in an account of sociological methodology, ‘is to be recognized by the power of external coercion which it exercises or is capable of exercising over individuals’ (Durkheim 1938, 10).

Durkheim’s 1897 study of suicide—a classic of sociology—illustrates nicely the sort of determinism in which he believed. The statistics on suicide may be wholly unpredictable on the basis of physical or biological or indeed psychological facts, he thinks. But they display a constancy in their relations to ‘states of the social environment’. ‘Here at last’, he says, ‘we are face to face with real laws’ (Durkheim 1951, 299). Asserting the relentless operation of these laws across different cultures and institutions, he comments in conscious irony that ‘Each society is predisposed to contribute a definite quota of voluntary deaths’ (Durkheim 1951, 51). The irony in the use of ‘voluntary’ is underlined by an explicit recognition that his approach is bound to scandalize ‘the zealous partisans of absolute individualism’. ‘For those who profess the complete autonomy of the individual’, he says, ‘man’s dignity is diminished whenever he is made to feel that he is not completely self-determinant’ (Durkheim 1938, 4).

2.2 The Issue

How likely is it that the laws which social science has discovered, or is liable to discover, might give the lie to our sense of ourselves as autonomous agents? Might they suggest that it is a mistake to think we are more or less successfully interpretable in the common psychological terms that we use to make sense of ourselves?

In order to make intentional or psychological sense, we must generally hold attitudes of belief and desire and the like that are rationally intelligible in light of the evidence at our disposal and we must generally act in a manner that is rationally intelligible in light of those attitudes. But we need not be unfailingly rational in these ways. It is part of our psychological understanding that there are various factors, some perhaps yet to be discovered, that cause us to be temporarily irrational, such as when we are preoccupied or upset, or subject to inertia or idees fixes. And neither need we see very deeply into the conditions that give rise to psychology-shifting effects as when we fall in love or are shocked by a traumatic experience. In order to be intentionally interpretable to ourselves and one another—in order to count as conversable agents (Pettit and Smith 1996; Pettit 2001)—we need only preserve a general conformity to rational expectations and a capacity, at least in the case of certain failures, to recognize and correct them.

According to anti-individualism, some of the laws of social science—some of the laws actually discovered or liable to be discovered—are downright inconsistent with the intentional or conversable image that we hold of ourselves. They require people to behave in certain ways and, in particular, to behave in ways that are intentionally unintelligible. The idea is that when social laws require people to respond in a certain manner, then regardless of whether this would make any psychological sense—regardless of whether it would cohere with our belief in their intentional intelligibility—people must respond in that manner. On some interpretations, including Durkheim’s own, the laws envisaged may have to be satisfied if the society is to survive and flourish: they are socio-functional necessities. And so the idea is that people are liable to be pushed by socio-functional requirements into performing in a manner that makes little or no psychological sense. They go on the blink as they put themselves, zombie-like, at the service of such necessities.

2.3 For Individualism

There are a number of considerations that argue against anti-individualism, as it has been characterized here. A first is that if we assume that the intentional laws assumed in predication of rationality would hold true in the absence of the social laws envisaged by Durkheim and others—and nothing he says suggests that they wouldn’t—then we have to think of the social laws as issuing from a novel sort of force. Vitalists argued that over and beyond the chemical laws governing living things there is a vis vitalis—a vital force—that explains why some chemically constituted organisms satisfy biological laws that are chemically unintelligible. And in the same way anti-individualists would have to argue that over and beyond the intentional laws governing agents like you and me there is a vis socialis—a social force—that explains why we psychologically organized agents satisfy social laws that are psychologically unintelligible. But just as parsimony argues against vitalism in biology, so it argues against anti-individualism in sociology. We ought to be driven to become anti-individualists only in the presence of undeniable data that cannot be explained in individualist terms.

A second consideration against anti-individualism is that there are no such data available. Even candidate laws that have a Durkheimian cast, and that make a claim to advance our understanding, can be fitted easily within an individualist picture. Assume for argument’s sake that it is a social law, for example, that an increase in unemployment gives rise to an increase in crime. If true, this law would tell you something important about the social world. No matter how fully you understood the psychology of individual participants in the society you might not have noticed the regularity it underlines. But the law would not reveal that there is a super-intentional force at work. All that it need posit is that a rise in unemployment, no matter how it is psychologically realized—no matter who lose their jobs and no matter how they feel and think—is likely to give rise in a psychologically intelligible way to an increase in crime.

Whatever individuals become unemployed, and whatever their psychological profile happens to be, the increase in unemployment means that there are more people than previously with a novel motive to commit crime—to make up for the loss of income—and with a novel opportunity to do so: the increased leisure that unemployment ensures. The increase in unemployment programs for the increase in crime, as we might say, since it means that things are psychologically organized so that, under plausible psychological assumptions about the interaction of motive and opportunity, an increase in crime becomes likely. The increase in unemployment does not produce the increase in crime in a manner that engages a novel sort of force and bypasses people’s intentional make-up (Jackson and Pettit 1992a, b; Pettit 1993).

A third consideration against anti-individualism is that not only are its claims inherently implausible and explanatorily unnecessary, they would also run into conflict with psychological tenets that lie at the very center of our web of belief and that it is very hard to imagine giving up. I have in mind the assumptions about our more or less rational character that we mobilize in interpersonal interaction, as we assume that in general we are each conversable—each capable, at least in the normal run, of being reached in conversation. This assumption shows up in our practice of talking to one another about what we ought to believe and desire and do, only despairing of this exercise with the rare individual whom we take to be out of their mind. It is particularly salient in our disposition, absent recognizable excuses, to feel resentment or indignation towards people who fail to register or respond to salient, other-regarding considerations and consequently do harm to us or to third parties (Strawson 1962). It is hard to imagine how we could continue the patterns of exchange and conversation essential to community—and maintain the parallel patterns of self-reflection and self-interrogation in which thought consists—if we gave up on the intentional, conversable image of members of our kind.

2.4 Qualifications

The account given of anti-individualism is motivated both by the history of the approach and by the fact that on this account, anti-individualism has important implications for our status as minded creatures. But I should add that there are many other doctrines that might reasonably claim to be anti-individualist and that no considerations rehearsed here are meant to challenge them.

One is the claim that social science can expand our psychological understanding of ourselves, revealing factors that perturb our normal functioning: this fits comfortably with the commonsense recognition that there is an open variety of emotional and cognitive blocks to optimal performance. Another is the claim that the social entities that come into existence as a result of individual interactions can themselves figure in people’s awareness, reciprocally influencing what they do; the appearance of money, for example, can elicit novel sorts of attitude and generate novel sorts of activity. And yet another is the common, if not altogether persuasive claim that for any grand developments associated with particular individuals—say, the Napoleonic reforms in early nineteenth-century Europe—those developments would have materialized, even in the absence of the individuals involved.

There is clearly no reason in principle why individualists in the sense relevant to our discussion might not endorse such doctrines. But there is another doctrine that may seem to challenge individualist assumptions more directly. According to this theory, there are social laws that are not psychologically intelligible, even if there are none that are psychologically unintelligible. The claim is that certain social laws cannot be derived from psychological laws—strictly, from psychological laws as they operate under various circumstances—not that they require various psychological laws to be false. They transcend psychological laws but do not confound them.Footnote 1

Strictly speaking, this doctrine need not be a challenge to the central individualist claim that social laws do not threaten to compromise our intentional or psychological sense of ourselves. But in any case it is hard to identify a persuasive social law that would resist psychological derivation in the sense required by the theory.

Suppose, by way of constructing such an example, that at an early stage in our evolution whole groups survived or perished in group-group competition; that the groups that survived were ones in which members were disposed under external threat to put aside internal divisions and fight as one against enemies; and that consequently we current human beings have almost all inherited this sort of disposition. Assume that as a result of such a group-selectional history, it is a social law that the members of a society unite against external threat. Might it be plausible to claim that that law is not derivable from psychological laws, as they operate in this or that circumstance? Might it be plausible to hold that this is so, because the law depends on the presence of a disposition that, by hypothesis, is not psychologically intelligible?

While I have no principled objection to the possibility illustrated, and while it does not really threaten the individualist position I hold, I think that the answer to this question must be, no. Any disposition that we inherited from our evolutionary history in the manner illustrated is almost bound to have been registered within our psychological sense of ourselves as a fully intelligible trait. Our intentional psychology has been formed in light of our experience of ourselves and it is surely likely that any evolutionarily established disposition to form certain attitudes under one or another circumstance would have been long identified as typical of our species. This is obviously true, as it happens, with the disposition cited in the example. No one is likely to think that a disposition to make common cause against an external enemy is psychologically underivable and surprising.

This argument is not decisive, of course. It applies only to one putative example of a social law that is not psychologically derivable and intelligible, even if it does not require any psychological laws to be false. But I think that most candidates for the role envisaged are likely to fall to similar considerations. In any case, we need not concern ourselves unduly with the question of whether the theory that such examples would bear out is likely to be sound. For unlike the sort of theory associated with Durkheim and his followers, at least as I have interpreted them, it would not do anything to undermine our status as individually minded agents.

3 The Atomism Issue

3.1 History

Where the individualism question is whether people’s status as minded, conversable agents survives in the space of aggregate social laws, the atomism issue is whether, on the contrary, that status presupposes a life conducted within the constraints of social relationships. You can have such and such a height or weight quite independently of whether there are any others around but you cannot enjoy prestige or power except in the presence of others. The question here is whether any of the properties associated with intentionality or conversability are more like prestige than they are like height or weight.

Although Aristotle (1996, Bk 1) argued that we human beings are essentially social or political agents, associating this feature with our ability to relate in a deliberative, linguistically mediated way, the atomism issue really came into prominence in philosophical discussion only in the eighteenth century. It became an issue in light of the German Romantic claim, foreshadowed in Rousseau’s Discourse on the Origin of Inequality (1973), that human beings depend on language for a range of minded capacities and that they depend on society for access to language. This claim became the central theme in the work of objective idealists in the nineteenth century, figuring prominently in the thought of Hegel and his followers.

Thomas Hobbes (1994a) had argued in the early 1640s that language is essential for the appearance of distinctive human capacities, presenting it as the source of what makes human beings special. He formulated this view as an alternative to Descartes’s (1985) claim—defended in his Discourse on Method of 1637—that language was a sign, not a source, of human distinctiveness; this Descartes took to consist in the presence of thinking substance, res cogitans. Hobbes maintained that language is a human invention that changed the nature of its inventors, giving them powers of mind that made them special among animals (Pettit 2008). He argued in particular that without language people would not be able to ratiocinate or reason; they would be incapable of thinking through theoretical or practical issues. They would not have the capacity exercised paradigmatically by the hunched figure of Rodin’s Le Penseur.

Hobbes did not think that the language that is essential to performing as a minded human being is essentially social. But from the time when Rousseau began to defend that idea, the atomism issue became a staple of discussion. If language is a construct that emerges only in the interaction of human beings with one another—if, contrary to Hobbes, it could not be the invention of a single person—then human beings are going to depend on interacting with one another for the appearance of the capacities that language underlies, in particular the capacity for thought. It is no longer going to make sense to think that a solitary individual, operating within the space of his or her own consciousness, could achieve the status of a properly minded agent.

The anti-atomists of the nineteenth century rang many changes on this theme. These changes included the Hegelian claim that it is only in interaction with one another that human beings become self-conscious (Hegel 1991). But the changes rung extended more generally to observations on the artificiality of abstracting from social context and treating individuals as the primary units of mind and agency. F.H. Bradley (1876, 173–74), the English idealist thinker, argued in this spirit that ‘the mere individual is a delusion of theory’ and that to ‘know what a man is you must not take him in isolation’.

3.2 The Issue

The question that divides atomists and anti-atomists is whether there are any features essential to human beings—in particular, any feature like the capacity for reason and thought—that depend for coming into existence on the enjoyment of social relations (Taylor 1985). But in order to understand the question properly there are two construals that we should put aside, one of them causal, the other logical.

On the causal construal, the question is whether we human beings depend causally on interaction with others—for example, on interaction with parents and other adults—for the appearance of distinctive mental capacities. Since it would be crazy to deny that we do, this reading of the issue has little or no appeal; it would make atomism utterly implausible and give anti-atomism too easy a victory.

On the logical construal, the question is whether we human beings depend as a matter of logical necessity on interaction with others for the appearance of these capacities. But this reading is equally unappealing, since it would make anti-atomism wholly implausible and give an easy victory to atomism. How might anyone argue that it is inconceivable that creatures like us could enjoy the full range of mental capacities in isolation from one another? To defend such an inconceivability claim would be to maintain that Descartes’s image of minded, potentially isolated subjects is not only mistaken, for example, but logically mistaken: there is no possible world in which people conform to his model. Few if any have ever thought that this was plausible.

I favor a reading of the atomism question that avoids both of these extremes, casting the issue as one of whether we human beings depend in a contingent but non-causal manner on our interacting with one another, or on our ever having interacted with one another, for the possession of distinctive mental capacities. The mode of dependence I have in mind is contingent rather than logical in character and constitutive rather than causal.

Consider your dependence on the presence of suitable antibodies in your blood for the enjoyment of immunity against a certain disease. The antibodies that make you immune do not cause that immunity, as they might cause a distinct, temporally downstream effect; they serve rather to constitute it. Thus you do not have to wait on the antibodies to have a causal effect in order to become immune: you are immune from the moment they are present. And yet the antibodies that make you immune are not logically connected with your immunity. It is possible in principle that you might enjoy immunity by any of a variety of other biological or indeed miraculous means. They constitute your immunity but do so as a contingent matter, not as a logical necessity.

On the construal I favor, the atomism question is whether there is a form of social interaction on which, in a similar manner, we contingently but constitutively depend for the possession of some central feature of human mindedness. If there is such dependence, then the exercise of that capacity will be inherently social in character. And in that sense the anti-atomist claim will have been established.

3.3 For Anti-atomism

Arguments against atomism have to start by picking out a feature of our minded make-up, then, and offer reasons why the presence of that feature presupposes social interaction in a constitutive role. I will sketch an argument that focuses on the capacity to reason and, more basically, to follow a rule. This is not the only sort of argument that might be put forward in support of non-atomism but it has the merit of focusing on a feature of human mindedness that is clearly important to our functioning and that appears to mark us off from other animals. While other animals can reasonably be attributed intentional states like belief and desire and the like, they give little or no evidence of the reasoning or thinking that we human beings conduct, whether on our own or in deliberation with others.

Reasoning in the intended sense may consist in determining on the basis of a rule like modus ponens whether a certain conclusion follows from premises already believed. But it may also consist in something much simpler such as wondering whether something not confronted previously is deserving of a familiar name: whether it counts under the appropriate rule of classification as an instance of this or that property or kind. I will concentrate on this latter sort of case, asking whether the rule-following involved in such a basic exercise of classification presupposes social interaction. I argue that it does, drawing on a response to the problem of rule-following raised by Ludwig Wittgenstein (1958, 1978), particularly in the interpretation of that problem offered by Saul Kripke (1985).

Suppose you grasp the meaning of a term—say, the property ascribed by a certain predicate—and aspire or intend to use the term in an appropriate way: that is, in a way that faithfully tracks the corresponding property. Assuming that you do not grasp the meaning of the term on the basis of definition in other terms—assuming that in that sense the term is semantically basic—you presumably identify the property you mean to track, and so the rule you expect to be guided by, on the basis of examples. Let the term in question be ‘regular’, as that is used of shapes. You will be introduced to the rule you mean to follow by various examples of regular shapes, where these are presented in a suitable contrastive context. Thus the examples used to cue you might be geometrical squares, circles, ellipses, triangles, and horseshoes, where these vary in color, size, font and the like, and are set in contrast to a variety of irregular shapes.

The main problem with rule-following is to explain what fixes the identity of the property or rule that such a finite set of examples is meant to present; in particular, what fixes the identity of the rule in a way that gives you access to that identity: after all, you must know which rule is in question if you are to be able to track it intentionally. There is no doubt, we may assume, that confronted with such a set of examples, and assisted by appropriate contextual priming, you are likely to catch on to the intended pattern. You will form a disposition to extrapolate to other cases, classifying candidate shapes as of a kind or not of a kind with the examples: that is, as regular or irregular. But how could just the formation of such a disposition amount to following a rule? How could it enable you to identify a rule with an indefinitely large extension; to aspire to follow that rule in using the term ‘regular’ across novel cases; and to do so, as rule-following requires, in a way that allows you to recognize that you may get that rule wrong?

The account of rule-following that I favor builds on the assumption that you are indeed likely to form a spontaneous extrapolative disposition in response to a set of examples like those mentioned. But it adds two important elements to that story, one proleptic or anticipatory, the other interpersonal or social. And it claims thereby to be able to explain how the disposition can allow you to identify a rule as something you can aspire to comply with, yet aspire without any absolute guarantee of success (Pettit 1993, 2002).

The proleptic part of the story is that the disposition elicited by the examples enables you, consciously or unconsciously, to conceive of the rule as something you can target as an object of compliance. You can think of it in anticipation as that rule, the one that you rely on your disposition to reveal in a case-by-case way.

Imagine, to take a parallel, that you know how to get between two places in virtue of, first, knowing where to go initially as you set out from one or the other end; second, knowing that when you get to that initial landmark you will know where to go next; third, knowing that when you get to the next landmark you will know where to go then; and so on. In such a case you will know that route between the two places, the one that is encoded in your disposition to move between landmarks; and of course you may know that route without being able to draw or describe it.

The idea in the first, proleptic part of the story about rule-following is that in the same way you can know the rule presented in a finite set of examples just by being able to rely on the disposition that is elicited by the examples. You can recruit the disposition to the role of identifying that rule in a case-by-case way and you can use the examples, then, to make that rule available as an object of attention: to pick it out as that rule, the one that is going to be salient to anyone with the required disposition.

The anticipatory or proleptic story will not suffice on its own, however, to explain how you can get to identify and follow a rule like that associated with the property of regularity. For all that the story involves in its first stage, you would have no reason to think that you could misidentify the rule on any occasion; you would have no obvious ground for conceiving of your rule-following as fallible. The second, interpersonal part of the story is mean to repair this defect.

The claim in the second part is that in using your extrapolative disposition to identify the rule you mean to follow, you assume that there is something to follow that is available to others too, answering to their dispositions as well as to yours. This means that faced in any instance with a discrepancy between your response and those of others you will naturally balk and look for an explanation that enables you—ideally, enables all of you involved in the divergence—to continue to think that there is something objective you are each meaning to track. The best explanation that is consistent with the objectivity assumption would identify some warping obstacle or oversight on the part of one or another party, so that the rule you aspire in common to follow can be cast as that rule, the one that shows up in each of your dispositions when the disposition operates in the absence of such perturbing factors: that is, in the absence of factors that would save the assumption of objectivity and yet explain the divergence.

If anything like this story is on the right lines, then rule-following consists at base in triangulating with others on a presumptively objective pattern, relying on that pattern to be available in virtue of the interplay between individually extrapolative and mutually corrective dispositions. Consistently with the story, it may often be the case that you intentionally and successfully follow a rule in isolation from others. All that is required is that you have had some experience of triangulation in the past and that you acknowledge the relevance of triangulation, if it is available, in the resolution of certain discrepancies. But that is still enough to establish the social character of rule-following. Drop the authorization of others in the identification of basic rules and you will lose any ground for presuming that the rules you mean to track are genuinely objective patterns—patterns that it is possible for you to misidentify.

The story sketched here might be replaced by a story in which each of us means to track the rule identified by our personal, extrapolative disposition, as that operates under presumptively reliable conditions (Blackburn 1984). But the substitute story faces the problem of explaining how we could individually identify such conditions. And even if it were to avoid that problem, it fails to explain how we can be licensed in assuming, as we routinely do assume, that the pattern we track in the use of a simple term like ‘regular’ is the pattern that others track too. In any event the social story answers much better to our common sense of what transpires in learning the meaning of the terms we use from others in our linguistic community. While there may be a possible world in which human beings each rely on their private, idiolectical resources to identify the basic rules they follow in reasoning, there is little or no ground for thinking that that world is the actual one.

This brisk presentation directs us to one line of argument that makes a good case for anti-atomism. The activity of rule-following and reasoning rests, it would seem, on the availability of a practice of using one another to give ourselves suitable targets of thought: suitable patterns to be guided by in working out what the world we chart in common requires us to say and think in this or that instance. We do not causally depend on the history and availability of such triangulation as we might depend on something distinct from reasoning itself; the dependence is constitutive in character. Nor do we depend on it as a matter of logical necessity; as just noted, there is nothing incoherent in the idea that we might identify and track rules on a private basis. But the sort of dependence involved is still enough to ensure that our capacity to reason and follow rules has a social character. As a contingent but constitutive matter, the ability to reason and follow rules presupposes interaction with others; it is not something that we could enjoy out of society.Footnote 2

There are serious issues raised by the adoption of such a theory of rule-following. For one thing, it means that any basic rule that we follow in reasoning—say, any property we ascribe in the use of a given predicate—will really be an equivalence class of rules that happen to coincide across instances that are in principle accessible to human negotiation. But this is not the place to explore such implications and consider their significance.Footnote 3 Let it suffice for the moment that we have found one plausible argument in support of anti-atomism.Footnote 4

4 The Singularism Issue

4.1 History

The singularism issue, as I understand it, is whether there are only singular human agents or whether certain groups can also perform in an agential role. We speak loosely of many groups as holding by certain attitudes and performing certain actions. But that need not give the lie to singularism. The issue is whether there are any such groups that constitute agents proper or agents in their own right, as it is often said. In presenting my views on this issue I follow earlier work, in particular work done in collaboration with Christian List (Pettit 2001, 2003; List and Pettit 2002, 2011, 2012).

The singularism issue has a long history, going back to a medieval debate that had been prompted, according to many accounts, by a decree of Pope Innocent IV in 1246 (Kantorowicz 1997). Arguing that a group agent could not be excommunicated—it did not have a soul and could not be sent to hell—Innocent described it as a persona ficta. Philosophers and theologians generally took this to mean that such a group was a fictional person or agent, not a person or agent in any real sense (Eschmann 1946). But lawyers developed the view that institutions like universities are examples, not of fictional persons, but of artificial persons: bodies that can act as natural persons act, at least within the context of law, and that count therefore as persons proper (Woolf 1913; Canning 1980). Thus they hailed guilds and towns, parishes and monastic orders, even the Church itself, as artificial persons that could enter contracts, own property, sue and be sued in the courts, and generally bear rights and obligations in the manner of their natural counterparts.

The concept of the artificial person survived in legal usage down to the nineteenth century but at the end of that century it received a great boost from the work of the German legal historian, Otto Gierke, who sought to resurrect the medieval category. He had an enormous influence on English and American legal and political theorists, many of whom took up the case for treating society as an arena of interaction, not just for individual agents, but also for the corporate bodies that they constitute (Hager 1989; Runciman 1997). Those bodies were taken to include the state at the highest level of aggregation but also the guilds and unions, the clubs and associations, the churches and colleges, that individuals constitute in more intimate forms of collaboration. A commitment to the reality of such agents, and to their status as agents proper, was characteristic of a variety of political movements in the early part of the twentieth century—for example, in guild socialism—but people generally retreated from this commitment about the time of World War II, perhaps as a result of an unwanted association with Fascist, so-called corporatist thought.Footnote 5

4.2 The Issue

Most groups are collections of people united only by a common characteristic or history or location and do not present in any way as agents. They are as varied as the collection of those who are red-haired or over two meters tall, those who come of a certain ethnicity or hold by certain political or religious beliefs, those who live in a particular area or were born at a particular time. But other groups certainly do present as agents, having members who actively join or acquiesce in the collective pursuit of certain goals, for example, and in the collective selection of the means that those who act in the name of group should follow in promoting the goals. The political party that organizes itself to campaign for election, the corporation that sets out to maximize the returns to its shareholders, the church that arranges for the satisfaction of certain proselytizing goals: all such bodies put themselves forward as entities that simulate the performance of individual agents. As individuals embrace a variety of purposes, deliberate about their relative importance and seek to identify the best means for their promotion, so the same is true of the corporate bodies illustrated.

The singularism issue is whether the bodies that simulate individual agency in this way count as agents proper, agents in their own right—whether, in older terminology, they should be treated as artificial persons. There is no agreed criterion of when a corporate body might simulate agency, yet fail to be an agent proper, but I shall take the relevant yardstick to have the following, quite demanding character. A corporate agent will not be an agent proper just insofar as the attitudes it embraces—and so its associated actions—are determined, issue by issue, on the basis of the attitudes of some or all of its members; they are mechanically responsive to corresponding member attitudes. An agent that was responsive in a mechanical, issue-by-issue way to the attitudes of its members would be like an avatar of those members, not an independent agent. Its thinking that such and such or its deciding that so and so would amount to nothing more or less than its members—equally or unequally empowered—having the profile that mechanically generates such aggregate dispositions.

The member responsiveness that would deprive a corporate body of the claim to agency proper can take a variety of forms. Any corporate body will have to form attitudes on the purposes it is to pursue, the priorities that should obtain amongst those purposes, the opportunities available for pursing them, the best means for doing so in an individual case, and the like. A responsive, and so not properly agential body might fix its attitudes on such issues by majoritarian or non-majoritarian voting among the membership as a whole; by a majoritarian or non-majoritarian process of voting on different issues by different, delegated sub-groups; by one process of voting in the case of one delegated sub-group, another in the case of another; and so on. I am prepared to say that even such a complicatedly responsive group agent has no more claim to be an agent proper than the group that is controlled by a single dictator and constitutes just a front for that person’s purposes and opinions.

4.3 For Anti-singularism

My argument for anti-singularism is that any agent that is organized to simulate agency in the manner of a corporate body must be organized in a manner that rules out mechanical responsiveness and in a way, therefore, that gives it a title to be regarded as an agent proper. There are two claims essential to the argument: first, that any body that simulates agency must be robustly sensitive to the demands of rationality; and second, that the satisfaction of such rationality requirements rules out the satisfaction of the responsiveness requirements. Together those claims establish the conclusion that well-functioning corporate agencies cannot be mechanically responsive to their members and must count as agents proper, agents in their own right.

To be an agent is to have the capacity to endorse goals, to form representations of the environment in response to incoming evidence, and to act according to those representations in pursuit of the goals. To have such a capacity is to form attitudes rationally on the basis of evidence, as we say, to act rationally on the basis of those attitudes, and to maintain only attitudes that are rationally co-tenable. Or at the least it is to be sensitive to any failures in such rationality and to be disposed to put them right.

Taking up the first claim in my argument, then, a group will be able to simulate agency successfully—to mimic the performance of an individual agent—only to the extent that it can satisfy such constraints of rationality or, at the least, be suitably sensitive to failures. And not only must it happen to satisfy those constraints as things actually are; it must also do so robustly. It must be so constituted that as we imagine it being faced with novel evidence on one or another issue, or becoming disposed to embrace a novel goal, we have grounds for expecting that it will adjust so as to maintain a rational, effectively agential profile. Did a group not have this profile then it would not be equipped to act for its purposes reliably: it would often find itself disposed to act in inconsistent ways. And equally it would not be an entity with which we could do business, as in projecting the responses it will make to various overtures, negotiating with it on that basis, agreeing to enter contracts with it, and so on.

The second claim in my argument is that if a group organizes itself to be rationally compliant and sensitive in this robust fashion, and if it confronts an interconnected set of issues on which it has to judge—as any real-world group certainly will—then it cannot organize itself in a mechanically responsive manner. This claim rests on a set of results in social choice theory—specifically, in the branch known as judgment-aggregation theory—that have begun to appear over the last decade (List and Pettit 2002; List and Polak 2010). But it can be illustrated by what I have described elsewhere as the discursive dilemma (Pettit 2001), building on the work of some legal theorists on a related question in law (Kornhauser and Sager 1993).

Suppose that a group of three people, A, B and C, have to make up their views as a corporate agent on four issues: whether p, whether q, whether r and whether p&q&r. And imagine that the group is member-responsive in a majoritarian way, being disposed on any issue to form the judgment supported by a majority of members. The matrix in Table 5.1 shows that majority voting may lead them to judge as a group that p, that q, that r and—on the basis of a unanimous vote—that non-p&q&r. Thus it shows that if the group is to satisfy rational sensitivity, as the simulation of agency requires, then it must breach majoritarian responsiveness.

Table 5.1 A discursive dilemma
Full size table

This example shows that majoritarian responsiveness is not consistent with the rational sensitivity that group agency requires. In order to operate properly as an agent, the members of the group have agree that whenever a majority vote generates a position inconsistent with positions already adopted, as in this case, they should go to a second round of consideration in which, regardless of their individual positions, they decide on which of the inconsistent attitudes to drop.Footnote 6 They have to monitor the positions generated over time by the group, taking each vote initially as a straw vote, and act to ensure that in the attitudes finally endorsed the group satisfies the basic requirements of rationality. In short, they have to construct the mind of the group, independently of the minds of its members, so that it is suited for agency. The members might be led under such a procedure to hold as a group that p, that q, that r and that p&q&r, accepting the fact that on the last issue they as a group have to maintain a view that each of them individually rejects.

Our example shows that a group cannot operate on the basis of majoritarian responsiveness and must adopt something like the straw-vote procedure. The various judgment-aggregation results in the literature generalize the claim illustrated. They support the thesis that no matter which form responsiveness assumes, majoritarian or non-majoritarian, centralized or delegated, it is liable to undermine the possibility of a robust form of rational sensitivity. And those results argue for the claim that if a group is to act like an agent, then it cannot be mechanically responsive to its members. The group may not follow the straw-vote procedure; that is only one way in which members can give the group they form a mind and an agency of its own. But whatever procedure is followed, the members of every group agent have to do something parallel. They have to allow the needs of group rationality to trump member responsiveness and to prompt the formation of a corporate body that counts as an agent in its own right.

The upshot is that if a group is to simulate agency, as many groups do, then it is has to replicate agency; it has to constitute an agent proper and not just an avatar of its members. While this result is surprising, however, it is in no way mysterious. It is not in virtue of any novel force or spirit that individuals come to constitute an agent in its own right but only in virtue of the way in which they organize their collective affairs, in particular the business of generating shared attitudes. The group they form may count as a different agent from its members but it amounts to nothing more or less than the same collection of individuals.

5 Conclusion

In opening this paper I said that social ontology is naturally guided by an interest in the significance of social interactions for our status as minded agents, guided by intentional attitudes. The positions for which I have sketched a defense support, first, the individualist claim that for all that social laws imply, people are intentional, conversable agents who are sensitive to the demands of rationality and display the modified autonomy ascribed in common sense; second, the anti-atomist claim that nevertheless people depend constitutively on social interaction for the capacity to reason and follow rules that human mindedness presupposes; and third, the anti-singularist thesis that when people come together to behave like a corporate agent, they have to form a collective mind of their own: they cannot tie the attitudes they endorse and enact as a group to the attitudes they hold as individuals.

These three positions in social ontology have important methodological and indeed normative implications (Pettit 1993, Chs 5 and 6; List and Pettit 2011, Chs 3 and 7). Methodologically, individualism argues for seeking only social laws and explanations that make psychological sense; anti-atomism makes a case for grounding psychological explanation in patterns of conceptualization—perhaps displaying cross-cultural variability—established in common across a society; and anti-singularism shows that if we are to make sense of the behavior of a group agent like a corporation or church or state, then among the explanatory strategies explored, we have to make use of the intentional stance we deploy in interpreting individuals (Tollefsen 2002).

Normatively, the three positions have corresponding implications. Individualism helps to vindicate giving priority to the interests of individuals—presumptively, considered as equals—in assessing social arrangements: no institution can make for good that does not make the lives of individuals go better. Anti-atomism suggests that we should reject the traditional idea that the benefits in terms of which to justify social and political life, establishing its merits in comparison with an anarchistic condition, should be restricted to benefits that individuals could enjoy equally in the absence and in the presence of social relationships. And anti-singularism argues for ascribing real rights and responsibilities to corporate agents, though only a pattern of rights and responsibilities that, as individualism requires, best serves the interests of individuals.Footnote 7