Abstract
When a group does harm, sometimes there’s no obvious individual who bears moral responsibility, and yet we still intuit that someone is to blame. This apparent ‘deficit’ of moral responsibility has led some scholars to posit that groups themselves can be responsible, and that this responsibility is distributed in some uniform fashion among group members. This solution to the deficit, however, risks providing a scapegoat for individuals who have acted wrongly and shifting blame onto those who have not. Instead, this article argues that, in most deficit cases, moral responsibility is borne not by the group but by specific individual members. When an individual acts within a group, she gains an increased potential for doing harm – and, accordingly, heightened duties of care toward others. These duties can, depending on the individual’s position, require amending the group’s rules, procedures, and norms. In most deficit cases, it is individuals who have failed to fulfill these duties who are responsible.
Keywords
Say that a dozen pilots from a nation’s air force, in the course of a just war, bomb an enemy town containing a military stronghold. 1 Their orders specify an exact amount of damage to be inflicted: as it happens, the minimum amount necessary to achieve justified military objectives. But, due to a series of accidents and oversights, the pilots end up decimating the town. Thousands of its civilian residents perish.
Suppose also that each plane carried a full bomb payload, even though the planners of the mission anticipated that they would carry one-third that amount, because of a typo on an administrative order that was not double-checked because the officer in charge of doing so suddenly took ill. Standing orders for each pilot were to release all bombs on each mission. Punishments for disobeying those orders were harsh. An unexpected fog over the town during the raid obscured visibility, limiting the extent to which each pilot could see the overall damage done, while enemy technology on the ground jammed all of the planes’ radios, making it impossible for the pilots to double-check their orders. All of the pilots dropped their full payloads. But no individual intended for the town to be destroyed; it is not even clear that any individual carried out their task negligently, aside from (perhaps) the typist.
The result is what some have called a ‘deficit’ of moral responsibility (Pettit, 2007: 194): we intuit that someone is responsible to a certain degree, but we struggle to identify any particular involved individual who is responsible to that degree. 2 Describing the bombing campaign as a typo and a dozen pilots lost in the fog seems to understate culpability.
For most, the deficit is merely prima facie. One possible explanation is that, intuitions notwithstanding, no one is responsible after all (Valasquez, 2003). The more common explanation from the literature is that, while the individuals involved might not be responsible, the group as a whole can be (Goodpaster, 1983).
I defend what might be thought of as a middle ground. I take seriously collective responsibility theorists’ concern that responsibility is difficult to identify in apparent deficit cases. However, I believe that at least in the paradigmatic cases that trouble these theorists, the deficit can be closed by existing individualist accounts of responsibility, without positing theoretically unfamiliar entities like collective moral agents. On my view, individuals can be subject to higher standards of moral care when operating as part of a group than when acting individually. They can be responsible not just for their individual acts within the group but also for how their individual acts contribute to the group’s acts. Some members of groups, in virtue of their knowledge and authority, can also be responsible for redesigning the rules and organization of the group in order to ensure that the group routinely acts rightly. I call this latter responsibility ‘caretaker responsibility’.
In addition to its ontological parsimony, a strictly individualist account of responsibility in groups may yield superior moral outcomes. In an increasingly organized and interdependent world, group action is a fact of life. Without one mind to superintend a group, it can be easier for moral considerations to slip through the cracks, and at a much higher moral cost. This fact makes questions of institutional design and policy not just a subject of study for business and organizational leaders but an urgent moral concern for us all. A too-ready retreat to collective responsibility in the wake of a tragedy may give the comfort of someone to blame but obscure the fine-grained analysis required to identify specific institutional defects and motivate their correction.
This article is organized as follows. In the first part, after briefly describing what I mean by moral responsibility, I develop the idea of a responsibility deficit. In the second part, I reject collective responsibility as an unsatisfactory solution to apparent deficits, arguing chiefly that collective responsibility leads to an unfair distribution of responsibility among members. In the third part, I propose my preferred solution – heightened individual responsibility, including caretaker responsibility – and explain how it closes key responsibility deficits. In the final part, I address two counterarguments.
The moral responsibility deficit
I begin by explaining how we attribute moral responsibility to individual agents for their acts’ consequences, and the significance of these attributions. Since my aim here is not to offer a full account of moral responsibility, I confine myself to the few, mostly ecumenical observations that will be necessary in order to build my argument. Based on these observations, I explain more thoroughly the intuition that group acts can produce ‘deficits’ of individual moral responsibility.
Attributing moral responsibility
To say that an agent is ‘morally responsible’ can mean many things. Sometimes the phrase refers to general fitness for moral decision-making, which requires possession of certain cognitive faculties (Fischer and Ravizza, 1998). Nearly everyone agrees that children, dogs, and robots, for example, lack responsibility in this general sense. Here, I am interested, instead, in agents’ moral responsibility – sometimes referred to as culpability – for specific acts. This is the sort of responsibility that opens an agent up to moral assessment of her will or character, and licenses others to either blame or praise her for an act’s consequences. This article focuses only on moral responsibility for acts with bad consequences.
An agent 3 is morally responsible for a bad state of affairs only if she (a) commits an act or omission that causally contributes to that state 4 with (b) inadequate moral care. 5 A great deal of the debate over collective responsibility has centered on the question of whether groups, or at least their members, can act in a way that meets the causal condition (a). I will, however, largely bracket this question, because my argument takes a different tack, challenging attributions of collective responsibility even assuming the causal condition can be satisfied.
I will focus solely on the moral care condition (b). This is the condition that transforms mere causal responsibility into moral responsibility. Moral care is the value or significance an agent demonstrates in her practical judgments to the interests (welfare, choices, and/or projects 6 ) of other agents – that is, the extent to which she treats these interests as ends to be protected and promoted by her own actions. She meets the standard of adequate moral care so long as she does not violate any moral obligations she has toward others without satisfactory excuse or justification. While the term ‘inadequate moral care’ is not commonly used in the literature, it is deliberately broad in order to capture a range of concerns from the moral responsibility literature about how wrongdoing reflects an agent’s will, character, or deep self (Björnsson, 2014; Scanlon, 2008: 141; Shoemaker, 2015).
Our moral care, or lack thereof, is reflected in an act or omission to the extent that our valuation of others’ interests contributes to our acting or not acting as we do. A sadist reflects the opposite of moral care (the valuation of others’ pain) when he kicks children because their crying thrills him. An agent may also fall just short of moral care in his acts, usually when he pursues his own interests in spite of small but unjustifiable harms to others’ interests. This devaluing of others’ interests may happen consciously, when the agent clearly predicts the harmful consequences of his act, shrugs, and acts anyway. It can also happen subconsciously, when he simply fails to think through his act's consequences for others. He might underestimate the probability or severity of those consequences, or the consequences might never even occur to him. When we care about someone, we think carefully about the consequences of our actions for them; neglecting to do so reveals a lack of care. So, a young, able-bodied bus passenger who fails to surrender her seat to a nearby elderly passenger clearly struggling to stand might show inadequate moral care even if she never observed the other’s need.
An act can of course cause harm without any failure of moral care. Perhaps our young bus passenger is unobservant because she is blind, having an epileptic fit, or distracted by an explosion out the window. 7 Or perhaps she observes the elderly passenger’s need but fails to heed it only because she is severely ill, or holding a heavy child, or trying to escape the notice of an assassin on the bus who wants to kill her. In the former cases, she has an excuse that makes her lack of moral care no failure at all; absent the excuse, she might have shown great care. In the latter cases, she may have been – consistently with adequate moral care – justified in privileging her own interests over her fellow passengers’. 8 It’s also possible that the passenger may have had no obligation to help the elderly passenger in the first place – say, if her seat was on the other side of the bus and she was getting off at the next stop. In none of these cases, then, would the young passenger have been morally responsible for the older passenger’s discomfort.
I will assume that the fact of an agent’s responsibility for an act also justifies other agents in holding and expressing certain reactive attitudes toward her like blame. 9 This is in keeping with a large swath of scholarship in moral philosophy. Indeed, many philosophers analyze moral responsibility as blameworthiness (Scanlon, 2008: 124–126). Definitions of blame vary across the field. All can agree that blame involves at least a negative appraisal of its object’s moral character – perhaps as malicious, selfish, callous, or weak of will (Arpaly, 2006; Scanlon, 2008: 126–127). Blame usually also involves the holding of ‘reactive’ attitudes, or attitudes attuned to the responsible agent’s own bad attitudes. On Scanlon’s influential account, the reactive attitudes justified by responsibility are adjustments in one’s intentions toward the agent: I may cease to trust the agent less, engage with her less, or express less special warmth toward her, depending on her particular wrongdoing (Scanlon, 2008: 145–147). Some philosophers go so far as to suggest that the appropriate reactive attitudes also include hostile emotions, such as resentment or even anger, toward a responsible agent and/or the hope that things will go badly for her (Feinberg, 1968; Wallace, 2011; Rosen, 2015). 10
Moral responsibility can also trigger two sorts of remedial obligations. The first obligations are backward-looking obligations to ‘make right’, as far as possible, the harm the responsible agent has caused. The agent may need to apologize and/or may bear certain liabilities, such as to repair damage or compensate his victims (Watson, 1996; Zimmerman, 2018). 11 There is good reason to believe that liabilities are not automatically triggered by moral responsibility (e.g., Scanlon, 2000), but moral responsibility can be a significant factor in their assignment. 12 The second type of obligations are forward-looking ones to prevent similar misdeeds in future. 13 The responsible agent may need to reform his moral character, change his valuation of others’ interests in the abstract or his habitual behaviors that neglect those interests in practice, or restructure his environment to aid in better moral decision-making. Of course, the agent likely already had and failed to fulfill these obligations of reform, but being morally responsible for a specific harmful act can trigger further specifications of those responsibilities: the agent may learn new facts about the nature of his deficiency or how to remedy it.
Moral responsibility can also motivate an agent to actually ‘take’ responsibility. An agent takes responsibility for an act when she acknowledges her wrongdoing, apologizes for it, and fulfills any related remedial obligations. The motivations for taking responsibility come in two varieties. First are internal motivations prompted by the agent’s own first-personal reactive attitude of guilt. While not every agent will experience guilt, those who do will often take actions to alleviate it. This is most easily accomplished by fulfilling the forward-looking or backward-looking obligations triggered by the wrongdoing. Even in the absence of guilt, a responsible agent may seek to improve her own self-narrative through reform and restitution. Second are the external motivations that arise when an agent is ‘held’ responsible by others – meaning she is blamed or when her remedial obligations are enforced. Feeling the weight of blame, the agent may be motivated to take responsibility in order to change others’ evaluation of her, or to increase her chances of being forgiven. If blame alone is not enough, the agent may also be moved to take responsibility when others pressure her to do so.
Everything I have said so far in this section enjoys relatively broad agreement among moral philosophers: that an agent who is morally responsible for a bad act had a morally defective will with respect to that act; that her moral responsibility licenses others to blame her and in some cases to enforce remedial obligations against her; and that her responsibility may, often due to her own or others’ corresponding reactive attitudes, prompt her to take remedial action. I will add one, somewhat less ecumenical contention, which draws on the idea that, as P.F. Strawson famously argued, the concept of moral responsibility cannot be understood fully prior to our social practices of holding each other responsible (see, e.g., Arpaly, 2006: 28; Strawson, 1962). The contention is this: that the social consequences of holding agents morally responsible may matter in determining who is morally responsible in the first place. This line of thinking will serve as the basis for much of my critique of collective responsibility in the next section.
The consequences of holding agents responsible that I have in mind are twofold. First are the burdens of blame on the responsible agent. She is likely, at minimum, to suffer the psychological harm of being judged negatively, resented, or even socially isolated. If she is pressured to fulfill remedial obligations, she may need to undertake great efforts toward rectification and compensation. It is thinking about these burdens that has led philosophers to suggest that it may not be ‘fair’ to deem certain people morally responsible when they do not seem to have done anything to deserve the treatment. The second consequences are the motivational ones mentioned above. Holding agents responsible often produces better moral conduct: either by encouraging responsible agents to take responsibility, or by deterring wrongdoing in the first place. Arguably better conduct is the central benefit of this practice. If, then, a class of agents is systematically beyond the reach of these prompts toward better action, we might think twice about whether it makes sense to say that they are responsible. 14
The responsibility deficit
The intuition of a deficit in moral responsibility occurs when human action has given rise to a bad result and (1) we intuit that someone is responsible, or should be responsible, for that result to a certain degree (i.e., with a certain lack of moral care) but (2) we cannot identify any individual agent who is in fact responsible to that degree. Often in these cases we struggle to pinpoint individuals who are responsible because of complexities in the causal chain, such as many people acting together or the use of technology. Consider, for instance, the case of the bombing campaign from the introduction. Part of the reason many intuit that someone is responsible is that the bad result seems not like bad luck but in some sense predictable, attributable to a rule or process or pattern of behavior – something analogous to an individual character trait. Other cases in which a responsibility deficit might be intuited are the manufacture of a dangerous product, the evolution of a discriminatory norm, or the misfire of an autonomous drone weapon.
Some will disagree that there is any deficit of responsibility in these cases: we intuit that someone is responsible because one or more individual agents is responsible. I share this view. I nonetheless think it is worthwhile trying to understand deficit cases better because the deficit intuition motivates some collective responsibility theorists. 15 Moreover, even if there is ultimately no deficit in these cases, there is a real complication in assigning moral responsibility in them – one that we can best understand if we adopt the view I later defend.
This article focuses on apparent deficit cases involving group action. I first want to explore why, in these cases, it can be so hard to identify responsible individuals. To begin, let us set aside group actions for which no deficit exists and identifying responsible individuals is easy. For instance, when all members of a group act together with a shared intent to achieve a certain morally objectionable goal, and achieve it, each member clearly bears responsibility. Consider a case in which a victim can be killed by pulling a lever, an action that requires the strength of 10 people. If 10 people want to kill the victim and, in order to do so, pull the lever, they are each responsible for the resulting murder. Here, each participant’s process of practical judgment with respect to the group act was similar to what it would have been if she had acted alone: she intended the full act, and had the opportunity to thoroughly assess its consequences for other persons before contributing to it. Accordingly, her (appallingly) inadequate moral care is straightforwardly reflected in the group act.
The deficit intuition arises when no individual wills the group act as a whole in this way. Instead, the act is produced by a combination of the group members’ practical judgments with respect to parts of the act, rather than the whole act. The result is that it becomes harder to determine how a given member’s moral care is reflected in the group act. It may also become harder to determine whether their care was adequate, given the unpredictability introduced by multiple agents interacting. Often calamity strikes despite every participant in the act exercising the same efforts toward moral care that they would if acting in their individual capacity.
How do groups come to act in such ways that are not intended in their entirety by any individuals? Often the answer involves common structural features of groups, of which I will discuss four here. One such feature is policies and procedures for making decisions and acting as a group. Members may be bound to follow group-imposed constraints, such as deferring to other members, prioritizing certain factors, or maintaining precedents. Groups depend for their organization, stability, and predictability on general adherence to such rules. A breakdown in that adherence can undermine the group’s ability to achieve its goals effectively. But policies and procedures, designed to cover a broad range of cases, can go wrong in unanticipated contexts. Moreover, a decision made under a policy will mix the intentions of the policymaker with the intentions of the policy implementer, which will make it harder to fully assess the moral care of either party. While it is true that the implementer, at least, retains the capacity to deviate from a policy in singular instances, they may lack the knowledge to do so. Even with the requisite knowledge, the possibility of undermining the group and/or facing penalties if they deviate may distort their moral calculus in ways that obscure the level of moral care reflected in their choice. 16
Another common structural feature of groups that obscures moral responsibility is compartmentalization. Different individuals or subgroups may be assigned to make different parts of a single multi-part group decision. For instance, some individuals may promulgate an abstract policy and others may implement it. Sometimes there is good reason for this, as multi-part decisions may be too complex for one person to monitor all parts of them or may demand different areas of expertise for different parts. But split decisions can also produce disaster. Some of the decision-makers may not have complete information about the other decision-makers’ decisions, or, even knowing their decisions, may lack the expertise to do anything but defer to them. Bonnie may decide whether to bring guns to the bank robbery, but Clyde may be the one to decide whether to bring them out; perhaps Clyde wouldn’t have brought the guns, knowing he might use them, and Bonnie never would have expected Clyde to use them. In some cases, the total number of parts of the decision may even be unknown, such that some parts are never assigned to be made by anyone.
A third common group structural feature is necessary coordination. This happens when an act must be carried out by multiple persons, each of whom agrees to do so – either because the act is otherwise impracticable or because the group’s rules require agreement. Of course, coordination permits groups to accomplish more than all of their members could, in aggregate, working alone. Coordination also poses no problem for identifying moral responsibility if it works, and, as in the lever case above, everyone acts with a shared intent. But coordination can fail. In a classic collective action problem, everyone ideally would cooperate but the consequences of being the only cooperator are fairly high – and so no one steps forward. Coordination can also fail simply because of the size of the group. Attempts to change subtle or cultural norms in groups, too, can be uphill battles. Even when coordination works, it can take time, impeding the right action at the right moment. While slowness can be advantageous, precluding reactive or hasty decisions, it can be calamitous in emergencies.
Finally, large groups often produce acts with broad scope: acts that are enacted in many iterations and contexts. For instance, group organization can permit one decision to be enacted in hundreds or even thousands of different locations – franchises, or offices, or cubicles – simultaneously. But the maker of a broad-scope decision has to consider a far greater range of possibilities than the maker of a decision that will be enacted at just one time, in one context. While I may only buy one replacement part for my car, a car manufacturer may buy one million. If each part has a 0.000001% chance of failing dangerously, then I can safely ignore that risk; if the manufacturer ignores the danger, however, then at least one of its parts is likely to fail. However, we do not ordinarily require people to consider possibilities with a one-in-a-million chance of occurring. At least, life would be difficult if every household decision required such careful study.
We can see how these structural features of groups obscure responsibility by examining a case. Say that the board of a car manufacturing company is considering whether to produce a new, cheaper ignition switch for one of the company's car lines. Of the three engineers assigned to review the switch, one insists that it is unsafe; a second believes it to be safe, but with doubts; and the third, the senior engineer, insists that it is safe. By company policy, they vote for a common recommendation and report, written by the senior engineer. Based on this report, the design executive testifies to the switch’s safety before the board. The board defers to her, as company policy requires, and votes 9-1 to use the cheaper switch. Later, research predicts unusually high future popularity for this car line, and the production executive proposes increasing production threefold. The board agrees. While further doubts about the switch’s safety circulate among company employees, none of them reach the board. The switches prove unsafe and cause six drivers to die in accidents before the cars are recalled.
Individuals’ moral care is obscured at every stage of this decision-making process. Policies and procedures obscure the responsibility of the board members (because they are required to defer to the judgment of the executive), as well as the first and second engineers (because the safety recommendation is voted on, and only one report is written). Compartmentalization obscures the responsibility of everyone, because the engineers are voting only on whether the ignition is safe, not on whether to produce the cars with the ignition; the executive and board members are voting only on whether to produce cars with the switch, without access to the empirical data. A coordination failure prevents the group from reacting to growing doubts. Finally, the decision to use the ignition switches has broad scope. As a result, no individual makes a fully informed decision to produce the number of cars ultimately produced despite the safety risk.
All of these features – policies and procedures, compartmentalization, necessary coordination, and broad-scope action – enable groups to function effectively. Yet the resulting increase in complexity makes it harder for group members to exercise moral care and thus raises the risk of morally costly mistakes. For highly organized groups, it can also be harder to figure out whose (if anyone’s) moral care is reflected when the group acts. Often, it is tempting to blame the bad policy, or the coordination problem. The difficulty is that only agents, not structural features, can bear blame.
Now that we have loosely established what a responsibility deficit is, why is it cause for concern? For some, the answer is that deficits can leave no one to hold responsible, to blame or punish, for an act that causes human loss. More importantly, perhaps, is that deficits can leave no one in a clear position to take responsibility. Without the motivations responsible individuals have of alleviating guilt, seeking forgiveness, and avoiding and dispelling blame, no one may be effectively motivated to change the structural features of a group that have in the past led to harmful consequences. To leave tragedies caused by organized groups without such responsible parties may be particularly troubling, once we remind ourselves that groups have extraordinary power–especially with today’s technologies–to do both good and evil. Groups can manufacture thousands of cars or operate fleets of airplanes or invest billions of dollars. As a result, when something goes awry in a group, even negligently, it can cause harms of staggering magnitude.
Collective responsibility theories
Collective responsibility theories offer a solution to the deficit: even if we can’t pinpoint individuals who are responsible in deficit cases, we can locate responsibility in the group ‘as a whole, understood as an entity that transcends the aggregation of its individual members and that is capable of having intentions of its own’ (French, 1984). If groups can be responsible, then, in the examples above, the air force squadron (or the whole air force or the state) and the car manufacturer would be independently responsible. Even if no individual acted recklessly to cause the deaths of the bombed town’s residents or the car drivers, the air force and the corporation, respectively, acted recklessly, and therefore can be called to account, blamed, required to compensate victims, and perhaps even punished.
I should clarify one point before proceeding. While collective responsibility theorists believe in group agents that are independent of their members, there need not be anything mystical or dualistic about such independence; a group can be independent of its members just because it is not reducible to them. Such a group might encompass not just a collection of members but the relations among them and the procedures that allow them to jointly decide on goals and act to achieve them (Pettit, 2007). From here on, I will call such a group-that-is-more-than-its-members a ‘collective’ and its responsibility ‘collective responsibility’. I am aware that ‘collective responsibility’ is often used more thinly to mean the individual responsibility of each member of a group. But here the term is evocative of my thicker meaning: the responsibility of a collective itself.
In this section, I first describe the basics of collective responsibility. I then argue that, even assuming that a collective can be an agent, it cannot be the sort of agent that can effectively take moral responsibility for its actions. Moreover, collective responsibility suffers from an unacceptable theoretical disadvantage: it must be distributed among individuals, but any such distribution will be unfair.
The collective and its moral responsibility
An act most clearly reflects an agent’s moral care if she intended it (even if she didn’t intend all of its effects). Advocates of collective responsibility have thus generally argued that collectives can have ‘intentions’ that are not reducible to those of their individual members and that they can – through their members – act on. Collective intentions, however, are not like the intentional states of human agents (Ohlin, 2008; Mäkelä, 2007). Rather, they are constructed from the intentions of individual collective members, often in complex ways. There is much sophisticated philosophical work on the phenomenon of ‘joint intentions’, or intentions shared among persons, that explain how this is possible (Bratman, 1999; Tuomela and Miller, 1988). Collective intentions often do guide the actions of members of a group, much like individual intentional states guide individual actions.
Assuming that a collective can intend, however, it is a further step to say that such a collective is a unified agent, and a further one still to say that it can be morally responsible. Most advocates of collective responsibility agree that only some groups that can form joint intentions are also morally responsible, though they disagree about how to define the responsible subset. Few suggest that what might be thought of as a ‘random collection of individuals’ can hold joint intentions, much less be morally responsible. For instance, the first 10 people who walk into crosswalks at 50th Street and Fifth Avenue in Manhattan tomorrow at noon will, on almost no one’s interpretation, jointly intend anything. Even if each intends to exit the crosswalk as quickly as possible, such an intention is merely coincidentally shared, not joint. But even if a group has joint intentions, this is not sufficient for agency. Bonnie Parker and Clyde Barrow probably jointly intended their robberies. Still, few theorists would deem ‘Bonnie-and-Clyde’ a third party responsible for their crimes.
Most collective responsibility theorists require that, to be morally responsible, a group agent have some method of declaring beliefs and intentions as its own. According to the well-known view of Peter French, a collective becomes responsible in its own right if it has an ‘internal decision structure’ specifying the hierarchy of collective decision-makers and the rules and procedures through which those decision-makers decide on behalf of the collective (French, 1979: 212). Some proponents of collective responsibility, prominent among them Christian List and Philip Pettit, advocate even more stringent requirements for collective responsibility. List and Pettit require that a responsible group have not just an internal structure for declaring group intentions but the capacity to ensure the consistency of those intentions over time (List and Pettit, 2011: 158–161). The group can become responsible only when, according to List and Pettit, it is no longer possible to describe the group’s beliefs/intentions as a strict function of its members’ beliefs/intentions.
Describing a collective as responsible seems especially useful when an act is largely attributable to structural features of a collective, like those discussed above. Take the 2015 crash of Germanwings Flight 9525 by a suicidal pilot. While outsiders still struggle to learn all of the inside facts, one could reasonably surmise that the pilot’s action was partly attributable to lax policies of security, testing, or even hiring, or to poor mental health benefits. Some collectives will consistently act in a harmful way due to systematically unethical policies and/or culture. Businesses driven solely by profits, for instance, might regularly cut corners to reduce production costs, even at the expense of employee or customer safety. The rash of oil spills and oil well explosions in the past couple of decades might be evidence of such policies among oil corporations. Such collectives might, at least metaphorically, be described as having defective moral ‘character’.
Collective responsibility also offers simple descriptions of many complex deficit cases. Say that a pharmacy customer requests a drug that the only pharmacist on duty refuses to dispense for reasons of personal conscience. Even if this pharmacist isn’t responsible for the dispensing failure, the pharmacy could be, according to collective responsibility theorists. These theorists might invoke an abstract collective obligation that supervenes on shifting configurations of individual obligations: even if this pharmacist wasn’t responsible for dispensing the drug, some pharmacist was – and therefore the pharmacy was. The collective's responsibility would hold even if no specific pharmacist was responsible (e.g., if all other pharmacists had resigned right before the customer arrived). Such supervening collective responsibility also neatly explains responsibility for classical collective action problems. It can also make identifying responsibility easier in cases in which we can’t quite figure out which policy led to a certain bad result, or how much each policy causally contributed; at least, the collective is responsible.
The collective, as an agent, can also, in at sense, take responsibility and be held responsible. It can, sometimes quite effectively because of its central decision structure, fulfill backward-looking obligations of responsibility, by issuing formal apologies (when the collective has a designated spokesperson), compensating victims and repairing damage (especially when it has its own treasury), and bearing punishment (such as dismantlement or fines). It can also take forward-looking responsibility, reforming defective aspects of its moral ‘character’. Such actions often require coordinated effort among multiple members.
The limitations of collective responsibility
Collective responsibility, however, has more weaknesses as a solution to responsibility deficits than at first appear, as I highlight in this subsection. In the next section, I offer my alternative individualistic account, arguing that it better fills apparent deficits.
Even if we accept that collectives can be agents and that collective responsibility theories offer the advantages canvassed above, these theories face the problem of pinpointing who will be motivated to fulfill the obligations that come with moral responsibility. While a collective agent might – according to philosophical arguments in vogue – ‘reason’ and ‘intend’ about moral issues independent of its members (Pettit, 2007), it cannot feel moral pressure independently of its members. Someone, after all, has to take it upon themselves to initiate a vote to reform offending policies and systems of organization. There may also be a danger of moral scapegoating – that if ‘the collective’ in the abstract is supposed to take responsibility, then every particular agent may think that someone else will do it (den Hartogh, 2009).
A common reply is that the collective’s responsibility lands derivatively upon the shoulders of individual members of the collective (Pettit, 2009: 168–171). 17 Admittedly, there is some disagreement in the literature about how this occurs, but on most accounts derivative responsibility is distributed across all or nearly all members of the collective according to some stable criterion, such as membership or hierarchical status, that has little to do with the members’ own acts or attitudes with respect to the bad result in question. 18 The idea, I take it, is that each member deserves some blame for the acts of collectives of which he is a part. In addition, as most collective responsibility theorists acknowledge, some members may also bear individual (non-derivative) responsibility for their individual contributions to the collective act (Pettit, 2009: 170).
This move to distribute collective responsibility, however, comes at a moral cost. Holding a person morally responsible for a bad result imposes substantial burdens. He may be exposed to demands that he explain or apologize; to blame, resentment, and other negative attitudes directed toward him; to feelings of guilt; and potentially even to punishment. Imposing these burdens on a person who has committed no act reflecting inadequate moral care raises worries about moral fairness.
Yet such individuals will often be found within a collective that has committed a bad act. Some may have been reasonably unaware of the act or any faulty rules/procedures that led to it, and others may have been aware but powerless to change anything. Indeed, remember that it is the absence of individual moral responsibility, traditionally understood, that leads to concerns about responsibility deficits. Consider another case. The Resistant Insider: Return to the case of the car manufacturer, and to the engineer who suspected the ignition switch was unsafe. Call her “Heroic.” But now imagine that Heroic, concerned for the company’s customers, went beyond voting against the switch’s safety: from her first suspicions until the first deaths from the switch were reported, she repeatedly raised the issue with her superiors, implored her fellow engineers to reconsider their opinions, sent letters to board members (whose mail was, unbeknownst to Heroic, automatically screened), and published an online blog post about the switch. Because of her persistence, she was demoted for “disruptive behavior.”
The case is designed to demonstrate that Heroic showed adequate moral care because she resisted as much as could reasonably be expected, persistently and consistently advocating against use of the switches and then attempting to mitigate their damage, even at considerable personal cost and even though the exact danger was uncertain. But perhaps the reader still thinks Heroic resisted less than she was obligated to, and therefore showed inadequate moral care. If so, substitute for Heroic in the following discussion a ‘Martyr’ who quits her job and sacrifices her life in a grand public protest brilliantly designed to halt production of the new car line (though production ultimately continues). Martyr, at least, must have shown adequate moral care. 19
Of course, not every member’s situation will be as compelling as Heroic’s. Perhaps if Heroic had not resisted, or even if she had resisted much-less-than-heroically, she would have been responsible. Perhaps dissenters who had even more power and knowledge than Heroic within the corporate decision-making structure – say, the board members – would have born some responsibility no matter what. My thesis is only that the appropriateness of blaming any particular member of a group for a group act intuitively maps onto a spectrum – and, crucially, a spectrum that begins at absolutely inappropriate. Some employees who did not even resist the corporate decision will be blameless. Specifically, employees should not be blamed if they were either: (a) wholly uninformed about the decision and the procedure by which it was made or (b) in no position to influence the decision or the procedure. A mailroom employee would likely meet both conditions. A custodian who overheard many board discussions while cleaning the boardroom would likely meet only the second.
One might object that Heroic – and other employees of the manufacturer – showed inadequate moral care in a way not so far considered: by working for the corporation in the first place. 20 Perhaps Heroic’s causal contribution to the corporate act came through participating in the general operations of the corporation that enabled it to manufacture any ignition switches at all. Or perhaps her coming to work for the corporation bound her fate to its own contractually. Either way, we might think that joining a group the activities of which pose a high risk to others – say, by manufacturing even potentially deadly cars – is itself morally reprehensible. Law draws on analogous judgments about dangerous activities in assigning legal liability. If a pet tiger bites someone, the person who decided to keep the animal as a pet is liable, no matter how carefully he managed the risks. Yet to the extent that we judge activities like owning tigers reprehensible in themselves, it is because they pose risks of grave harm without any redeeming social value. By contrast, we don’t generally blame people simply for driving cars, though it imposes risks even on non-drivers. Nor would we want to blame those involved in the manufacture of cars, unless the particular manufacturer had a dubious track record.
There is one final defense of collective responsibility to consider. One might argue that it is fair to distribute collective moral responsibility among collective members because this derivative responsibility is a special, lesser category of responsibility. Perhaps it entails unusually light burdens: it cannot license hostile attitudes; or can justify only punishments targeting the group, and not members in their personal capacity (punishments from which members are to be allowed to suffer only incidentally). One explanation for the leniency might be that derivative moral responsibility is mediated through the collective: one is blameworthy, but only qua member of the group; one is morally evaluated negatively, but only qua member of the group. Of course, this is highly inconsistent with our ordinary practices of blaming and holding responsible. More importantly, it generates a dilemma. This re-definition of moral responsibility is meant to remove an unfair burden upon the individual. But to the extent that it succeeds, it will also remove the moral pressure to actually take responsibility that accompanies non-derivative responsibility. In other words, it will leave the responsibility deficit unfilled.
A better solution: Heightened individual responsibility
I have argued that collective responsibility fails to motivate the taking of responsibility and unfairly burdens even morally pure members of collectives. Fortunately, collective responsibility isn’t needed to close the responsibility deficit. All moral responsibility for group acts – including the intuitively ‘missing’ responsibility in deficit cases – can be explained in terms of individuals’ responsibility for their own acts. One might be skeptical, given the fact that deficit cases are precisely those in which we struggle to pinpoint responsible individuals. However, the individualist account becomes far more plausible once we recognize that individuals have moral obligations to exercise more effort and diligence when acting as part of a group than when acting solo.
To see why, recall the earlier discussion of moral care. Exercising adequate moral care demands not only restraint against intentionally causing harm, but also less obvious efforts in the form of attention, observation, reflection, taking precautions, etc. To reiterate, if we care about someone, then we will take pains to avoid harming them even inadvertently. However, there is a limit beyond which further efforts are possible but no longer required. We tend to think, for instance, that even very harmful consequences that are very unlikely to occur can be ignored. When the wind whisks a hat from a person’s head and into the face of a cyclist, who then crashes, we do not blame the hat’s owner for failing to anticipate this outcome when choosing to wear the hat. We also tend to think that certain harmful consequences that would take very costly measures to prevent, or that could be prevented at immensely lower cost by potential victims themselves, can be ignored. For instance, we do not expect car drivers to prepare for the possibility that a pedestrian will jump in front of their car on a highway (absent a crosswalk). Doing so would demand so much attention and preparedness as to preclude highway driving altogether. Hosts are expected to warn temporary house guests of a spill on the entryway floor, but not of a stone statue under a bush that might injure anyone jumping from the second-floor window. In other words, adequate moral care requires effort only up to a point.
How much effort adequate moral care demands can increase above average levels when we pose an unusually high danger to others (however unintentionally) or possess unusual capacities to prevent harm to others. Under these circumstances, demonstrating the same moral care as a person who is less dangerous or less capable may require doing more. On the heightened-danger side, a victim of an infectious and deadly disease may have to take unusual, burdensome precautions to avoid infecting others, or an abnormally strong person may need to take unusual, burdensome precautions when shaking hands or embracing others. 21 On the heightened preventive-capacitiesa side, these capacities can include personal abilities, knowledge, position, or resources. A dermatologist who notices a deadly skin lesion on a fellow subway passenger may have a duty to warn him, even if a medically untrained passenger who surmised the same would not. Likewise, if I alone encounter a person drowning while out in a boat, I may uniquely have a duty to rescue them.
Acting as part of a group can heighten our capacities both to do and prevent harm. If a decision made by a group member is broad in scope, any consequences of it may be augmented in magnitude and likelihood. For instance, one managerial decision within a popular fast-food franchise can affect the diet and health of a nationwide population. Or, as mentioned earlier, one decision about replacement parts at a major car company can affect thousands of households and roads. Yet the very efficiencies and complexities of group action that make such broad-scope decisions possible – including general adherence to policies and procedures and compartmentalization – can also increase the risk of erring in the first place. With so many steps involved in making and executing a group decision, it will be easier for participants at every stage to have inaccurate or incomplete information about the consequences of their own contribution and its interaction with others’ contributions. The result is that, in order to assess the consequences of her action for others as demanded by adequate moral care, an individual operating within a group may be obligated to investigate and reflect more than she would if acting similarly outside the group. She may also be obligated to encourage good moral reasoning in her fellow members, raising moral questions whenever possible, advocating moral concerns as far as reasonable, and urging caution whenever morally relevant factors are implicated.
A special case of heightened moral obligations occurs when we have the capacity to control something that is unusually dangerous to others. The source of danger might be a substance or object, like a radioactive chemical in our laboratory. It might also be another agent, one who poses a high risk to others when acting because it cannot independently exercise moral care. Children and pets are examples. By undertaking to raise children, parents are expected to morally train them and, until they are trained, keep their conduct within moral bounds. By choosing to keep a dangerous dog breed, a dog owner is expected to appropriately confine and constrain the dog and post warnings. Similar observations might apply to our use of technology and, in particular, artificially intelligent agents. While the burgeoning literature on AI and moral responsibility worries about a possible responsibility ‘gap’ caused by AI (Matthias, 2004), the gap is at least partially closed by producers’ obligations in designing their artificial agents, and even in deciding whether to unleash these agents at all. I call these obligations of design and constraint over other agents ‘caretaker’ obligations.
Groups need caretakers because, somewhat like children, they lack the full capacity to exercise adequate moral care by themselves. The institutional structures that prevent harm to others – the policies, procedures, and hierarchy resembling a collective moral character – must be enacted and maintained by individuals. Whether or not these structures exist and are adequate is attributable to individual agency, as are any harmful consequences that predictably result from them. Even the absence of a person with appropriate oversight abilities may be attributable to someone’s failure to recognize the need for oversight.
Only some members occupy positions within the group that afford them the knowledge and authority necessary to shape these structures. 22 I submit that these members have caretaker obligations over their group. Such a caretaker will be obligated to do as much as can reasonably be expected of her given her position to ensure that the group’s actions are adequately sensitive to moral concerns. 23 Caretaker obligations are distinct because they have a bearing on decisions about how the group is organized, while ordinary heightened obligations in groups have a bearing on decisions given how the group is organized.
To be clear, then, any heightened moral obligations within groups are based not on mere membership but on one’s position within the group – that is, the knowledge and authority one has or can have in the group’s processes of decision-making and acting. Ordinarily, the authority that entails serious caretaker obligations is adopted voluntarily, just as a parent chooses to raise their child rather than having it adopted, and so fairness concerns fade. The higher the rung one occupies in a collective hierarchy, the likelier one is to bear responsibility for a group act. 24
Heightened duties of care for individuals acting in groups mean that moral responsibility in some of the paradigm deficit cases may be attributable to individuals after all. Heightened duties also help to explain why some initially intuited a deficit in the first place. The intuition of a deficit comes from the difference between what we have to do to demonstrate adequate moral care when acting on our own versus acting in a group. In certain cases, it also comes from the fact that caretaker obligations often have to be filled long before the cared-for agent acts, and so in at least some cases it will be true that no single group member on the ground at the point of action did anything wrong. When a bus crashes due to a brake failure, the driver may lack any responsibility (given the ordinary duty of care), but other bus company employees may be responsible: such as one who should have performed more thorough checks given how well-used the bus would be (given a heightened duty) or one who should have mandated driving record checks in the first place (given a heightened, caretaker duty).
The individualist account also closes the deficit in a more satisfying way than collective responsibility does: heightened individual responsibility falls on those who (a) had the power to influence the bad group act, and (b) likely still have the power to instigate corrective action and need motivation to exercise it. In the original car company case, heightened individual moral responsibility can be located in the senior engineer who didn’t revisit his conclusions on a high-moral-stakes question in spite of considerable doubts in others, and in the non-heroic engineers who didn’t try to register their doubt outside formal channels. Caretaker responsibility can be found in those who established the safety approval process, which was egregiously deficient in requiring only three engineers’ review, majority voting, and a joint report with no dissenting opinions; no means for appeal of an initial safety finding; and compartmentalized decisions about safety and production volume. It can likely also be found in those who knowingly perpetuated this system, such as the design executive and possibly the board. In addition, the individualist account properly spares powerless group members. 25 So, Heroic gets off the hook in Resistant Insider, as may the bombers in the introductory example.
Heightened moral demands within a group, especially caretaker ones, may be steep and even require training to meet. The training ought to be better within a group that poses especially high moral risks, like an energy company or transportation agency. If one is going to make decisions that will affect thousands of people, or set policies governing how others make decisions that will affect thousands, then one had better be exceptionally prepared to do so in a morally sound manner. Depending on the group’s size, training may require extensive experience with or study of collective decision-making and action. It will require an ongoing effort to keep abreast of moral failures in groups, including other groups. If an airplane crashes for the first time in history for a preventable but heretofore unexpected confluence of factors, it may be excusable; but if another airplane, at the same company or elsewhere, crashes for the same reason, someone has probably shown inadequate moral care. In other words, the efforts morally demanded of group members will rise over time as our technology and knowledge of social organization improves.
I count it as a strike in favor of this individualist account that it applies even to groups that lack centralized decision-making structures. Recall that many collective responsibility theorists stress that collectives cannot be responsible without such structures. However, I submit that ‘deficit’ intuitions arise, and can be answered, even in other cases. Consider the classic case of sunbathers on a beach who witness a swimmer drowning but fail to come together to form the long human chain that would have saved him. 26 It seems clear that, even if each sunbather stood ready to form a chain, they have violated a further obligation to try to coordinate their efforts – simply in virtue of each's position to possibly influence others.
Objections
Here I briefly address two counterarguments often raised against my proposal. Doing so also gives me an opportunity to clarify an important aspect of my view of moral responsibility.
The first counterargument is that it is important for corporations and other legal entities to be legally liable for their actions. I agree. However, nothing that I say above precludes victims of corporate action from seeking legal recourse. Whether an agent is in some sense responsible (or liable) for paying the costs of a harmful act – and especially legally liable for such – is a separate question from whether they are morally responsible for that act. We often assign insurance companies responsibility for the costs of motor collisions for which none of their employees could be morally responsible. Similarly, agents sometimes ‘take’ responsibility for costs they are not morally responsible for, such as a deceased parent’s debt.
Unsurprisingly, moral responsibility and responsibility for costs are related. As mentioned above, a person who is morally responsible for the consequences of an act is often also responsible for bearing the costs of those consequences, at least unless someone else steps in. And moral responsibility should be a factor to consider in distributing costs under law.
However, an agent can be what David Miller calls ‘remedially responsible’ for rectifying harm even when they are not morally responsible for it (Miller, 2007: 81). Numerous factors might justify assigning remedial responsibility in such a case, among them the agent’s relative ability to pay or otherwise provide a remedy, contractual or promissory obligations, causal responsibility, and the undertaking of risky activities (e.g., Smiley, 2014). 27 For instance, the babysitter may be remedially responsible for cleaning up his young charge’s mess even if he wasn’t a cause of the mess. Law itself often assigns remedial responsibility for pure reasons of social policy. This is why, for instance, dog owners often must compensate victims whom their dog bites and parents often must pay for damage caused by their child – even if these owners or parents were model owners or parents. The case for holding corporations responsible as a matter of law for the consequences of their employees’ acts, unpredictable or not, is often especially compelling because of corporations’ deep pockets.
The second frequent counterargument to my view is that we often experience guilt, or at least guilt-like feelings, for acts by groups of which we are members, even when our own moral care is not reflected in the act. Perhaps this is proof of distributed collective responsibility. A commonly raised example is a nation-state that acts ‘on our behalf’. I believe that the feeling in these cases, however commendable and indicative of high moral standards, does not track moral responsibility like genuine moral guilt tends to. 28 It would be inappropriate for anyone else to exhibit reactive attitudes toward the guilt-experiencing party or otherwise hold them responsible, absent actual personal wrongdoing.
That these guilt-like feelings detach from moral responsibility should be clear because they often arise even in cases in which only the action of another individual is involved. For instance, some people feel guilt-by-association for wrongdoing committed by family members or comrades in social organizations. These feelings may be explained by erroneous judgments that are often false, such as that one could and thus should have prevented the other’s wrongdoing or that the act proves that all members of the group share a common immoral tendency.
In certain cases, these guilt-like feelings may stem from actual responsibility, but of the remedial sort only. Remedial obligations can be created by association with others who are culpable. If an agent does not satisfy these obligations, then she would be morally responsible. Two facts about association with a culpable agent might lead to these obligations. First, the non-culpable agent’s association may give her an advantage in correcting a harm done. For example, a company that has caused a natural calamity or a nation that has invaded another will often be best placed to remedy these harms. New board members who join the company, or citizens who protested against the invasion, will still have an obligation to vote for these remedial actions – rather than leave the victims without recourse. Second, the association may benefit the non-culpable party so as to make her complicit unless she eschews the benefit or else works to rectify the wrongdoing. For example, white persons in states that have oppressed people of color have obligations to help in remedying those past injustices in part because they currently benefit from white privilege.
Conclusion
I have argued that we need not accept the idea of collective responsibility in order to identify responsibility in cases in which groups do harm. When an individual joins a group and assumes a position of authority and knowledge within it, she also gains an increased capacity for and likelihood of doing harm. With this power and dangerousness comes a heightened standard of care that requires her to account, in her own actions, for the complexity of collective action and the proper design of the decision-making rules and procedures of her group. Once we recognize these heightened duties, we can see how failures to live up to them cause many collective disasters – and thus how the apparent blameworthiness of the collective can be reduced to the blameworthiness of individuals.
Footnotes
Acknowledgments
This article owes much to the generous feedback of Charles Beitz, Chloé Bakalar, Johann Frick, Harrison Frye, Philip Pettit, and Anna Stilz, as well as to the input of Emad Atiq, Shuk Ying Chan, Ari Koslow, Josh Simons, Co-Editor Andrew Williams, the two anonymous reviewers of this article, and the participants in the 2017 iterations of the Harvard Graduate Conference in Political Theory and the Princeton Political Theory Graduate Research Seminar.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.