A Pluralist Approach to Joint Responsibility

Abstract

Sometimes reactive attitudes target groups rather than individuals.1 This suggests that groups are sometimes morally responsible. A growing number of moral philosophers agree with this, yet fail to recognize just how diverse the set of morally responsible groups is. Dominant approaches to group responsibility thus suggest that only one type of group is morally responsible, namely, those groups that have emergent capacities that mirror the agential capacities of rational and normatively competent individuals. Let us call these fully-formed groups. However, some of our reactive attitudes target not only fully-formed groups but also what I will call aberrant groups, namely, groups that lack one or more of the capacities possessed by fully-formed groups. The full repertoire of reactive attitudes might not be suitable for aberrant groups, but it nonetheless seems justified that we disdain a Neo-Nazi rally, feel gratitude toward the group of strangers who spontaneously cooperate to save a child, or disapprove of a greedy corporation. These reactive attitudes are appropriate even if the Neo-Nazi rally lacks the structure that would make it capable of entering a moral conversation, if the spontaneously cooperating group of strangers is unable to form group-level judgments and decisions, and if the corporation is normatively incompetent. These aberrant groups are not fully-formed moral agents, and yet, intuitively, some reactive attitudes are appropriately directed at them. In which sense, if any, are they morally responsible? Discussing a range of what he calls marginal agents (people suffering from mania, autism, mild intellectual disability, and psychopathy), David Shoemaker has argued that there are three distinct types of responsibility—attributability, accountability, and answerability—and that each of these requires different agential capacities.2 In this paper, I argue that a similar moral responsibility pluralism can help us make sense of the reactive attitudes with which we meet aberrant groups and, hence, shed new light on the moral responsibility of groups. Section II outlines the problem of aberrant groups by distinguishing between three different types of groups—expressive groups, interpellated groups, and group psychopaths—that arguably warrant specific group-reactive attitudes. In Section III, I connect this with Shoemaker's pluralism, which suggests that the way in which an entity is morally responsible depends on the extent to which that entity possesses different agential capacities because different agential capacities make entities susceptible to different kinds of normative claims. Sections IV and V clarify in which ways aberrant groups can have emergent agential capacities and how they can be targets of reactive attitudes. Sections VI–VIII, VI–VIII then analyze the three types of aberrant groups in more detail. I argue, in short, that expressive groups, interpellated groups, and group psychopaths possess the emergent capacities necessary for attributability, accountability, and answerability respectively, and that this is why we meet them with the reactive attitudes that we do. Section IX considers two objections, namely, that aberrant groups only warrant what Strawson calls the objective attitude and that it is unfair to blame group members for being parts of aberrant groups. I conclude in Section X by highlighting some of the advantages of adopting a pluralistic approach to group responsibility. The dominant approaches to group responsibility examine which groups, if any, possess the capacities that would make them proper moral agents, that is, which groups possess the capacities that would make them moral agents in the allegedly only way in which something can be a moral agent. Individual agents possess these capacities by virtue of their psychology; groups must have them by virtue of their organization. Different accounts take different capacities to constitute proper moral agency and thus emphasize different kinds of organization, but most assume that there is only one way for groups to be morally responsible.3 These accounts have a taken-for-granted commitment to moral responsibility monism. But given the fact that social life is populated by many different groups, it is hardly self-evident that only one type of group is morally responsible. Intuitively, we meet many of these with morally charged responses. And according to P. F. Strawson, such morally charged responses, what he calls the participant reactive attitudes, are constitutive of moral responsibility.4 This gives us prima facie reasons for believing that many different types of groups are morally responsible. I suggest that this is so because different types of groups are morally responsible in different ways. Expressive groups are capable of having and expressing a largely coherent set of cares or commitments. They typically lack the capacity for communication (e.g., a spokesperson) and for group-level decision-making (e.g., a board) and the empathic and coordinative capacities that would enable them to understand and act on the way in which other agents depend on them. Examples of expressive groups are mobs, spontaneous uprisings, and disorganized political rallies that, despite their lack of formal structures, seem to have a joint evaluative outlook. Interpellated groups have the ability to cooperate to satisfy the moral demands imposed upon them by another agent. The members of interpellated groups can understand how their joint effort can prevent some harm, they can understand how other agents depend on them, and they can track and respond to each other's behavior in pursuit of (what they take to be) a joint goal. Interpellated groups typically lack the capacity for communication and group-level decision-making and, beyond their joint capacity for empathy, the members of interpellated groups typically do not share cares and commitments. Examples of interpellated groups are random groups of strangers called to stop a violent assault on a train5 or people getting together to save a child from being carried away in a runaway hot air balloon.6 Group psychopaths have the capacity to deliberate, judge and make decisions. They might have the capacity for communication but necessarily lack normative competence, i.e., the ability to recognize and respond to moral reasons. This includes both abstract moral reasons such as those provided by the categorical imperative and concrete moral reasons such as those available to empathic agents aware that other agents depend on them for their well-being. Classical examples of group psychopaths are corporations such as Monsanto, Amazon, or the fictional Soylent Corporation, or military units like the SS or the Wagner Group. Expressive, interpellated, and psychopathic groups populate the social world. But they leave us uneasy. On the one hand, they are not proper moral agents. And so, we seem to have no more reason to praise or blame them than we have to praise or blame a dog, a laptop, or a baby. But, on the other hand, aberrant groups intuitively do prompt reactive attitudes. I disdain the (expressive) group of Neo-Nazi protesters marching down the street, feel gratitude toward the (interpellated) group of strangers who saves a child, and disapprove of the (psychopathic) hedge fund that invests in weapons and child labor. The solution to the problem of aberrant groups is not to plead with and try to modify our group-reactive attitudes. Strawson famously argued that our person-reactive attitudes are “naturally secured against arguments suggesting they are in principle unwarranted or unjustified.”7 This is likely also true of our group-reactive attitudes. The attempt to reduce our group-reactive attitudes to aggregate person-reactive attitudes is similarly unsuccessful (or so I will argue). Instead, I propose that aberrant groups leave us uneasy because they are morally responsible in some ways (and thus appropriate targets of some reactive attitudes) but not in others. It only makes sense to hold an entity responsible for meeting or failing to meet a normative standard if that entity is capable of meeting that standard. A beetle cannot fail to add two and two together because the beetle lacks the capacity for mathematical reasoning. Monists about moral responsibility claim that there is only one type of moral standard and only one set of capacities that makes entities susceptible to it. Pluralists claim that there are several moral standards and that entities must possess different capacities to be susceptible to these. Recent work focuses on three such moral standards. The first concerns who an entity is; the second concerns how an entity regards or concerns itself with others; and the third concerns how that entity judges or responds to reasons. Shoemaker's tripartite theory thus claims that when we morally assess someone, we assess either their character, their regard for other entities, or their judgment.8 These domains are distinct and irreducible. One neither implies nor presupposes the other. Accordingly, there are not just one but three types of moral responsibility. Focusing on our reactive attitudes (or, as he calls them, “responsibility responses”), Shoemaker distinguishes between attributability that assesses character by way of disdain and admiration, accountability that assesses regard by way of anger/resentment and gratitude, and answerability that assesses judgment by way of approval or disapproval.9 An entity must have different capacities to be morally responsible in these ways. Attributability requires the capacity for being emotionally and evaluatively responsive to one's surroundings but also the capacity to be so in a way that is largely coherent over time. Accountability requires the capacity for empathy and the capacity for acting on what we believe is good for the other. Answerability requires the capacities for abstraction, deliberation, and communication. If we want to attribute moral responsibility to a group rather than to the individuals that constitute it, the group must have some kind of agency above and beyond the agency of its members.10 At the same time, however, we do not want to claim that the group exists independently of its members, i.e., that there are some mysterious sui generis social substances. To navigate these treacherous waters, I propose that group agency should be seen as a set of emergent properties or, more precisely, emergent capacities. Roughly speaking, something is emergent when it has properties that the components that constitute it do not have on their own. Using this terminology, the monist would hold that one and only one set of emergent capacities warrants group-reactive attitudes. My pluralist claim, on the contrary, is that there is a plurality of such morally relevant emergent capacities and that aberrant groups can possess these. To understand why this is so, we must see, first, that groups can have emergent capacities even if they are not fully-formed and, in the next section, how emergent agential capacities can be targeted by reactive attitudes. The first step of my argument is to see groups as complex dynamical systems.11 Complex dynamical systems have properties that cannot be fully explained by an understanding of their components.12 Instead, they have three important features that help us understand how they can have agency above and beyond that of their members.13 First, they consist of several interacting components. In the case of social groups, these components are agents. Second, these systems have emergent properties in the sense that the collective behavior of the components forms a pattern that could not be predicted from the behavior of the components taken individually. Third, the group is self-organized in the sense that the emergent behavioral pattern is not under the control of a single component agent but is the result of several interlinked component processes. The idea that groups are complex dynamical systems sheds new light on how groups can have emergent agential capacities. Where monistic conceptions of group responsibility operate with a strict dichotomy between fully-formed individual agency and fully-formed group agency, the pluralist approach suggests the following: Several agents can mutually interact, regulate and constrain each other with the result that they exercise their capacities in a way that is counterfactually different from how these agents would have exercised their capacities if they were not arranged in that particular composition. This constitutes an emergent agential capacity in the sense that the capacity cannot be predicted from the behavior of the members separately and must therefore be attributed to the group as a whole. Importantly, an emergent agential capacity can be self-organized in the sense that mutual interaction, regulation, and constraint do not require top-down control.14 Building upon this conception of emergent agential capacities, it is helpful to distinguish between what I suggest we call soft and firm emergence.15 An agential capacity is softly emergent if and only if the higher-order agential capacity, A, is realized through functionally similar lower-order agential capacities, a1…an. Here, A exists only by virtue of how it regulates or constrains a1…an. An agential capacity is firmly emergent if and only if the higher-order agential capacity, A, is realized through functionally different lower-order agential capacities, b1…bn. In a composition with a firmly emergent capacity A and the functionally similar lower-level agential capacities a1…an, A can be exercised independently of a1…an. To see the difference between soft and firm emergent capacities, consider the different ways in which groups are said to share emotions. If the emotion of the group is the sum of how the individual group members happen to feel and their feelings are not altered by their group membership, there is no emergent group emotion but merely an aggregate emotion. If the emotion of the group is softly emergent, it is realized through the lower-level emotions of the individuals. This happens, for instance, when each of us feels in a specific way because we are affecting and affected by how the rest of us feel. In this case, our feelings are mutually constrained by a feedback loop so that we, as a composition, feel differently than we would outside of the composition even if our emotion is realized by or located in my and your emotions.16 Finally, if the emotion of the group is firmly emergent, it is realized by different means than our lower-level capacities for having feelings. Along these lines, Margaret Gilbert has argued that a group has an emotion if the members of that group are jointly committed to having that emotion as a single body.17 According to Gilbert, this entails that the members normatively require each other to have the thoughts and perform the actions typically associated with that emotion (e.g., a feeling of guilt) but not necessarily to actually feel that emotion. In this case, the group emotion is realized through a (cognitive and conative) “feeling rule” rather than the exercise of the lower-level capacity for emotions themselves.18 This account of emergent capacities allows that some lower-level agential capacities are overdetermined in their exercise, e.g., that a group member feels X whether he was in the group or not. The distinguishing feature is at the level of the composition that must have a different counterfactual pattern compared to the same lower-level agential capacities arranged differently. I propose that the distinguishing feature of group-reactive attitudes is that they, in contrast to person-reactive attitudes, target the exercise of morally relevant emergent capacities. Let me unfold this idea in a bit more detail by distinguishing between three different ways in which we hold groups responsible. First, sometimes our reactive attitudes simply target an aggregation of people based on how each of them exercises their individual agential capacities. Such “groups” cannot be appropriate targets of reactive attitudes because there is no emergent agential capacity that would make it possible for them to meet a normative standard. Any blame must hence be fully distributed among the individuals. Let us call this shared responsibility. Second, current literature focuses on a much stronger form of group responsibility, namely, collective responsibility. Collective responsibility suggests that a group is praise- or blameworthy independently of its members. In the words of Tracy Isaacs, “claims about the responsibility of collectives do not entail (or erase) claims about the responsibility of individual members.”19 On a reactive attitudes approach, this requires that we target the emergent capacity but not necessarily the lower-level capacities as this would imply, say, that the members share some of the blame. This seems implausible if the emergent capacity is self-organized for, as argued above, a self-organized emergent capacity consists of nothing but the mutually constrained but irreducible interaction and regulation of component capacities. Instead, it seems that when we hold a group collectively responsible the target of our reactive attitude is mainly the component exercising top-down control over the composition. Third, reactive attitudes can also target self-organized emergent capacities. In contrast to the aggregate capacities targeted in shared responsibility, the exercise of self-organized emergent capacities requires that the lower-level capacities interact, and mutually constrain and regulate each other. In contrast to the emergent capacities targeted in collective responsibility, the exercise of self-organized emergent capacities precludes top-down control. This means that if a reactive attitude targets a self-organized emergent capacity for a wrongful exercise, then that reactive attitude targets both the higher-order capacity and the lower-order capacities because these are mereologically connected. Let us call this joint responsibility. Assuming that our reactive attitudes like other emotions have transparent intentionality,20 they are partly individuated by their targets in a way that is phenomenologically clear to us. This means that we, by carefully attending to the experiential structure of our moral emotions, can distinguish between shared responsibility, collective responsibility, and joint responsibility. There is, hence, a phenomenological difference between blaming a collection of people separately (shared responsibility), blaming a group as a whole and the group members as parts of that whole (joint responsibility),21 and blaming a group without necessarily blaming its members (collective responsibility). This is true, I think, for both self-directed and other-directed reactive attitudes. For instance, I can blame myself tout court, as a group member, or I can blame a group to which I belong without blaming myself. For the remainder of this paper, I will focus on other-directed group-reactive attitudes. And for the sake of the argument, I will grant the monist that fully-formed groups are fit to be held collectively responsible. My focus is hence on how we hold aberrant groups that we do not belong to jointly responsible.22 Having thus clarified the nature of emergent agential capacities and their relation to the reactive attitudes, we can finally turn to the problem of aberrant groups. We start with expressive groups. Expressive groups are capable of having and expressing a largely coherent set of cares or commitments. In pure expressive groups, this capacity is the only morally relevant emergent capacity. Pure expressive groups can neither empathize with others nor make decisions although the individuals who constitute those groups often can. In pure expressive groups, the capacity for having and expressing a largely coherent set of cares or commitments is likely to be softly emergent. In the midst of an assembly animated by a common passion, we become susceptible of acts and sentiments of which we are incapable when reduced to our own forces; and when the assembly is dissolved and when, finding ourselves alone again, we fall back to our ordinary level, we are then able to measure the height to which we have been raised above ourselves.25 It does not matter to my argument which exact mechanism causes this emotional and evaluative regulation. What matters is that some groups are unified by an emergent capacity for having and expressing a largely coherent set of cares or commitments, that this results in a counterfactually different pattern of emotions and evaluations than a mere aggregation of the same capacities would, and that this does not require other morally relevant emergent capacities. I intend the phrase “cares and commitments” to encompass the entirety of an entity's evaluative outlook. The phrase reflects that there are two distinct ways of being an evaluator. First, you can evaluate in the sense that some of your desires are reflectively or rationally endorsed or rejected. These are what I call commitments. For example, my commitment to utilitarianism causes me to give money to charity, although I would much rather spend that money on ice cream. Second, however, you can also be an evaluator in the sense that you are emotionally disposed to do certain things, whether or not these desires are reflectively or rationally endorsed. These are what I call cares. For example, a teenager raised in a conservative religious community may have strong homosexual desires. Here we are inclined to say that the desire is more expressive of who that teenager truly is than his conscious judgment that homosexuality is wrong. Like Shoemaker, I believe that both cares and commitments are formative of an agent's character and, hence, something that can be appropriately subjected to moral assessment.26 Joint commitments require both the emergent capacity for emotions and evaluations and the emergent capacity for judgment. Joint cares, on the other hand, require only the emergent capacity for emotions and evaluations. As David Silver has argued, we sometimes morally assess the cares or commitments of groups in the same way that we morally assess the character of an individual.27 We admire or disdain an organization's culture, a crowd's atmosphere, or a community's ethos. For ease of reference, I will use the term culture to refer to a group's evaluative outlook. A group is attributability-responsible for a set of attitudes or actions if and only if those attitudes or actions express the group's culture, that is, if those attitudes or actions depend on and are harmonious with the group's emergent cares or commitments. Importantly, the attitudes of a set of group members might be attributable to the group even if these attitudes are superficially different from each other. Imagine, for instance, a political rally in which A appears fearful, B appears vengeful, and C appears agitated. At the surface level, these attitudes are different, and yet they can still depend on and be harmonious with, say, the group's hatred of what they take to be a powerful ethnic minority. To borrow a term from Bennett Helm, what matters is that the group is “a subject of import,” that is, that the group is jointly invested in some object or project in such a way that the group members realize a largely coherent emotional pattern.28 As long as the different attitudes of A, B, and C are appropriate in light of the import of the group, the group is fit to be held attributability-responsible for these attitudes. You might object to the idea that expressive groups warrant group-reactive attitudes if you take the reactive attitudes to be “forms of communication.”29 Against this common assumption, it should be noted that not all reactive attitudes, at least as conceived by Strawson, have an obvious communicative intent. David Beglin rightly points out that neither forgiveness, hurt feelings, guilt, nor the feeling of obligation fit this interpretation.30 So, although some reactive attitudes presuppose the capacity for moral address, this is not the case for all. As a pluralist, I happily accept that different reactive attitudes presuppose different agential capacities. Rather than being communicative, attributability responses admire or disdain their target at a distance, so to speak. Some think that blame is only appropriate when the action or attitude that one is blamed for is in principle sensitive to judgment or responsive to reasons.31 This implies, T. M. Scanlon notes, that “a collective agent can (…) be a possible object of blame, only if there are procedures through which it can make institutional decisions.”32 I am skeptical that there is a necessary connection between responsibility and blameworthiness, on the one hand, and responsiveness to reasons, on the other hand. This thesis can be interpreted as a local claim regarding a particular action or attitude or a global claim regarding the general capacities of the blamed agent. Both are problematic. Scanlon, Angela Smith, and others defend the local claim that a specific action or attitude is only attributable and hence an appropriate target of blame if that action or attitude is responsive to reasons. Yet, on my view, some attitudes are so central to who someone is that they inform our moral assessments of that person even if they are beyond their reflective reach. Suppose Mike was abandoned by his parents as a child, and due to this childhood trauma, he is prone to violent fits of jealousy where he mercilessly beats his partner. Mike knows that it is morally wrong for him to beat his partner, and he has gone through years of cognitive therapy to change his ways but to no avail. The merciless beatings are, despite his best effort, beyond his rational control. How do we react to someone like Mike? Of course, the fact that he regrets his actions and has tried his very best to become a better person will lessen our reactions to him, but I nonetheless think that people are prone—and rightly so—to hold his violent disposition against him and to blame (or, to be precise, disdain) him for acting on it. If this is indeed appropriate, attitudes and actions need not be judgment-sensitive or reason-responsive to be attributable to an agent and, hence, appropriate targets of moral assessment.33 But what about entities that are globally incapable of judging and responding to reasons? Are they wholly excluded from the moral community? Scanlon argues that when there is no procedure for institutional decisions (i.e., the group-level equivalent of the capacity for judgment or reason responsiveness), “there is no basis for attributing attitudes to such groups in anything other than the distributive sense, in which saying that the group holds certain attitudes is simply to say that most of its members do.”34 This, he concludes, “is just stereotyping.”35 But here Scanlon presupposes that some form of top-down control is necessary for capacities to be emergent. Yet, as I have shown above, emergent capacities can also be self-organized, and this gives us, pace Scanlon, a basis for attributing attitudes to groups even if they lack a centralized decision-making procedure. In short, to hold a group attributability-responsible is not, as Scanlon thinks, to assess it in terms of its responsiveness to reasons but in terms of its culture. Interpellated groups have capacities for cooperating in order to satisfy the moral demands imposed upon them by other agents in need. In the literature, the main example of an interpellated group is provided by Virginia Held: five unacquainted people witness a violent assault in a subway car, and it is obvious to each of the five people that they could stop the assault with no serious injury to themselves if two or more of them were to cooperate.36 Held argues, controversially, that the five strangers have a group responsibility to stop the assault although they lack the capacity to make group-level decisions and judgments.37 A group is accountability-responsible for a set of attitudes or actions to the extent that those attitudes or actions display the quality of regard of that group for another agent, that is, to the extent that those attitudes or actions display the group's emergent concern for another agent with whom the group has the capacity to empathize. Let us consider in more detail how a group can have the emergent capacities for acting jointly and for empathizing. First, as noted above, the vast literature on collective intentionality suggests that groups can have emergent attitudes without being fully-formed. An important subset of this body of literature focuses on joint or shared action. An influential strand of this literature, pioneered by Michael Bratman, claims to be “reductive in spirit.”38 One might expect a reductive account to explain the kind of small-scale, short-lived, and egalitarian cooperation central to examples such as Held's in terms of aggregate capacities. But a closer look shows that even reductive accounts are only plausible if they presuppose the existence of emergent capacities. Bratman's central claim is that we intend J only if (a) I intend that we J and (b) you intend that we J.39 Björn Petersson (2007) has among others charged Bratman's account for being circular since “we J” appears in both the analysandum and the analysans.40 Bratman, however, claims that there is no circularity in his account since there is a decisive but subtle difference between the two. “We intend J”—the analysandum—refers to what it is for us to plan to J. The analysans then explains that this requires that each of us have an unplanned or pre-reflective understanding of what we are capable of doing together.41 Bratman's account is reductive in the sense that the emergent capacity of the analysandum (“We