Corporate agents—highly structured collectives like firms, governments, and branches of the military—are already among the most powerful actors on earth. As scientific and social technologies enable ever larger and more effectively integrated corporate agents, it becomes increasingly important to address the question of whether they are bound by moral obligations. All and only moral agents are bound to act in accordance with moral standards, so the question then is: ‘Are corporate agents moral agents?’ Among the many grounds for doubting the moral agency of corporate agents is the fact that corporate agents are not phenomenally conscious. Phenomenal experience plays such a central role in our own experience of moral agency that it is deeply counterintuitive to imagine moral agency operating in the absence of such experience. My goal in this article is to explore (and ultimately reject) that concern, at least in the specific context of Kantian moral agency.
For my purposes here, an entity is a ‘Kantian moral agent’ if it can—and therefore should—meet Kantian standards for moral action, and it ‘acts morally’ if it acts on the basis of morally relevant information because the latter is morally relevant information. Within the Kantian system this will involve (inter alia) acting on the basis of a universalizable maxim because it is a universalizable maxim or acting in a certain way because acting in that way treats humanity as an end. The test for the ‘because’ is not ultimately grounded in intention but in the counterfactual possibilities. If Acme would not have φed if φing failed to treat humanity as an end, then Acme φed because φing treated humanity as an end (see Hess Reference Hess and Crane2015 for a more detailed discussion). In this article I will restrict myself to the example of a business entity (Acme) for the sake of simplicity, but the claims made apply to corporate agents wherever they may be found. I do not address the possibility that less structured collectives, like book clubs and climbing groups, might qualify as Kantian moral agents, nor will I explicitly consider whether other accounts of corporate agencyFootnote 1 can meet the Kantian challenge addressed here. Finally, the treatment here is specifically concerned with the role of phenomenal consciousness, which involves the felt, qualitative aspects of intentionality and agency—the ‘what is it like’ to believe, want, intend, etc. While Schwitzgebel (Reference Schwitzgebel2015) and Silver (Reference Silver2018) have argued that corporate agents may in fact be phenomenally conscious (but see List Reference List2018), for the purpose of this article I accept the general consensus that they are not. This leaves open the possibility that they may be ‘access conscious’. Block (Reference Block1995) introduced the concept of access consciousness, which involves the practical availability of content with no regard to qualitative aspects; David Copp (Reference Copp2006) and Deborah Tollefsen (Reference Tollefsen2015) make excellent use of this concept to argue that corporate agents are ‘conscious’, or at least as conscious as they need to be for moral agency and responsibility. I do not address that possibility here.
The thesis of this article, then, is that corporate agents can and therefore should act in ways that fulfill Kantian moral obligations and that the lack of phenomenal consciousness, in particular, is no bar to this. I begin by presenting my own, metaphysically robust account of corporate agents and outlining three mechanisms by which they exercise their agency, two of which are distinctive of this account (section 1). I then consider three increasingly demanding accounts of Kantian moral agency, showing in each case that the corporate agent can meet the requirements in its own right in the absence of phenomenal consciousness (section 2). In section 3 I explore and reject the possibility that any necessary role remains for phenomenal consciousness to play, and I conclude by noting some of the issues raised by these claims.
To be clear, I do not argue for the possibility of corporate agency in this paper. The claim here is conditional: if firms (and other corporate agents) can be sophisticated agents in the manner described here—if the phenomenon described qualifies as an agent—then such firms are Kantian moral agents. The lack of phenomenal consciousness presents no additional difficulties. The arguments for the antecedent are long, technical, and have been presented elsewhere, so the focus here is on the consequent.
1. Corporate Agents
A corporate agent, as I use the term, is a group of agents bound together by a structure that embodies a rational point of view. The ‘structure’ of a corporate agent is a network of social relationships internal to the corporate agent that imposes constraints on the actions of the members. It provides templates of interaction that favor certain forms of action and coordination over others, and it makes certain kinds of things available to the members (concepts, resources, social relationships, power) under certain conditions, essentially ‘structuring the possibility space for [their] agency’ while they are acting as members (Haslanger Reference Haslanger2016: 127). The elements of the structure will include things like French's (Reference French1984) CID structures, Werhane's (Reference Werhane1985) systems, and List and Pettit's (Reference List and Pettit2011) voting procedures (formal and informal)—all as enacted—as well as things like Gilbert's (Reference Gilbert1992) joint and Bratman's (Reference Bratman2013) shared intentions and Miller's (Reference Miller2001) collective ends. It will also include things like the explicit incentive schemes and demands documented in formal statements (as enacted) as well as unintentionally developed, often tacit elements like the real power structure (vs. the one on the organization chart), the incentive schemes bound up in the culture and peer expectations, and demands embodied in office politics and unpublished actual practices.
The structure that unifies a corporate agent is neither neutral nor fixed. It is not neutral in that it embodies what Rovane (Reference Rovane1998) described as a rational point of view (‘RPV’): a logically coherent set of commitments about fact and value—about how the world is and what matters in it—that reliably guides action. The term ‘commitments’ here is intentionally broad and noncommittal, a placeholder for a broad range of agential states either identical or analogous to human agential states from intentions and beliefs to things like mere habits of mind and behavior, perhaps even subpersonal ones. What all of these things have in common is that they pick out a reliable pattern in behavior, one that resists change but—once identified—can be overridden at the agent's choice. They arise from different mechanisms and exhibit different levels of rationality and robustness, but nonetheless play an important role in shaping and accounting for the ways we behave. In alluding to corporate commitments, I mean to allude to a similarly broad array of corporate states without making any strong commitments (here) about precisely what is included in that class. (My thanks to a referee for raising this issue.)
The commitments of the RPV provide the corporate agent with a specific, unique orientation to the world, qualifying as beliefs and desires on the standard accounts that were developed to explore human intentionality. Tollefsen (Reference Tollefsen2002, Reference Tollefsen2015) draws on interpretationism to support this claim while List and Pettit (Reference List and Pettit2011) draw on what is essentially dispositionalism; I have acknowledged both interpretationist and dispositionalist approaches, but argued (Hess Reference Hess2014a) that corporate commitments also qualify as beliefs on representationalism—a much more demanding and more widely accepted theory. Briefly, corporate commitments qualify as beliefs and desires on interpretationist accounts because attributing such commitments to corporate agents is the most effective way to ‘‘interpret’’ their behavior—to account for past actions and reliably predict future actions. Corporate commitments qualify on dispositionalist accounts because corporate agents that possess such commitments are, in fact, disposed to act in rationally predictable ways because they possess those commitments. Representationalist accounts are less easily summarized, but more revealing. On representationalist accounts (drawing on Dretske Reference Dretske, Heil and Mele1993, Reference Dretske1988), to believe that x or desire that y is to have a representation that reliably guides the entity's actions—effectively, for the corporate agent to have a kind of ‘picture’ of the world gathered from the world that leads the corporate entity to act in one way rather than another. Note that this ‘picture’ need not be an image (Dretske Reference Dretske1988: 71); computers represent the world in bits and bytes, and our bodies represent the world in chemical surges. (It is interesting to note that the phenomenal aspect of human intentionality is so utterly absent from the original literature, aimed as it was at accounting for human intentionality. Phenomenal consciousness is rarely mentioned at all, and I have yet to identify a major account that gives it a central role in intentionality or agency.)
With that, we can recognize these familiar corporate commitments as the beliefs and desires of the corporate agent, which guide its actions. The many elements of the structure provide multiple, overlapping reinforcements to ensure that the collective actions of the members effectively pursue the things the corporate agent desires (those things toward which the structure reliably guides members’ actions) in accordance with the ‘picture’ of the world that forms the content of corporate beliefs.
The structure—with its RPV and its various mechanisms—shapes the agency of the corporate agent. That is the aspect that is relevant here, but that is not the structure's sole function. Unlike the single mechanisms or ‘decision procedures’ that define accounts from French, Werhane, List and Pettit, and others, the structure also unifies the members into an integrated social whole that qualifies as a material object capable of its own action in the material world (not just ‘secondary’ or ‘vicarious’ action) and defines its identity through change (see Hess [Reference Hess2014b, Reference Hess, Hess, Isaacs and Igneski2018b, and ‘Re-Bunking Corporate Agency’, an unpublished manuscript] for further development).
Importantly, neither the elements of the structure nor the commitments implicit in it are irrevocably fixed. Wringe rejects this possibility in his brief consideration of the possibility of Kantian corporate moral agency, suggesting that certain corporate commitments are irrevocably fixed (Reference Wringe2014: 275). This is a fair application of French's formalist account of corporate agency, in which the agency of a corporate agent is established and delimited by its legal status and formal documentation in a way that human agency is not. On my account this is not the case. (This raises questions about what does anchor corporate identity through time and change; I address this in [Hess Reference Hess2014b].). On my account, any element of the structure can be altered, with new commitments added or old ones removed, either by strategic member action (per French, Werhane, List and Pettit) or by simple erosion or evolution over time. The possibility of Kantian corporate moral agency will turn on the question of what kinds of commitments can be incorporated into the corporate RPV and the processes by which this can occur. The latter concern can be recast as the question of whether the corporate agent is truly ‘autonomous’—whether its commitments are truly its own rather than those of (or intended by) its members. The possibility of legitimate changes to the corporate RPV in the absence of strategic member action—a possibility not acknowledged by the dominant accounts from French or List and Pettit, but evident from empirical observation—is thus especially important when it comes to questions of the autonomy of the corporate agent.
In the remainder of this section I sketch three mechanisms by which Acme could come to adopt commitments that none of its members held or desired. I will return to these examples later, in the discussion of Kantian corporate moral agency. First, a cautionary note: In the following scenarios I repeatedly describe situations in which members unintentionally—and often unknowingly—instantiate new corporate commitments through their actions. I do so to demonstrate that things can happen this way, that there is no necessary linkage, conceptual or causal, between member intent and the corporate commitments and structures that arise from member activity. This shines the starkest possible light on the discontinuity between member intent and commitments, on the one hand, and corporate commitments and action on the other; it emphasizes the crucial independence of corporate intentionality and action. It is not, however, necessary that the members act unintentionally or unknowingly or that the corporate commitments established by their actions be inconsistent with their own.
1.1 Explicit decision making
In the first mechanism, the decision to modify is quite intentional and explicit: a group of ranking executives votes, the majority wins, and the firm adopts a new position. This is the kind of mechanism that French, Werhane, and List and Pettit have in mind, and it often involves the development of collective intentions similar to those proposed by Gilbert, Miller, and Bratman. Even in these cases, as these and other scholars have insisted, the resulting position adopted by the firm with regard to a particular issue does not necessarily reflect or derive from the members’ own commitments regarding that issue. For example, Acme's executives know that Acme is committed to seeking profit, producing industrial chemicals, and being environmentally responsible. Reasoning from these commitments, Acme's executives could easily conclude (and vote) that Acme should produce steel additives. They could also conclude that this should be done in ways that are protective of the environment, but it is clear that Acme's commitments to profit and production are more central to Acme and take priority. And so Acme develops a new commitment to producing steel additives and the work goes forward, preferably in an environmentally responsible way—not because any executive cares about steel additives (or the environment) but because the new commitment is a logical extension of the existing commitments. Should the commitment to environmental responsibility prove too cumbersome, the executives may well vote to abandon it—or lead Acme to abandon it in practice without even realizing they have done so—despite their own personal commitments about the environment. Going forward, Acme's actions will nonetheless express this altered commitment rather than the commitments of its members.
1.2 Distributed decision making
In the second mechanism, Acme may abandon its commitment to environmental responsibility through a distributed process—one which is much less explicit and often effectively opaque to the members involved. While the first mechanism is more common at executive levels, this one typically occurs in the middle tiers, and the participating members may well be unaware that they are establishing a particular commitment for the corporate agent—one that will nonetheless shape future corporate action. Distributed decision making is a bit like assembling a car, with each member contributing one fact or decision and then passing the project on to the next without any member being in a position to see the ongoing collective actions that the corporate agent will reliably enact as a result of their individual choices.
To sketch an unrealistically simplistic example, assume that as a result of Acme's new desire to produce steel additives:
Member A requests proposals from Departments α, β, and γ.
Member B picks the one from α and modifies it slightly to reduce costs.
Member C modifies the proposal to improve materials handling.
D modifies the proposal to improve efficiency,
E modifies it to improve health and safety compliance (less worker exposure),
F modifies it to use different (domestically available) chemicals, and
G modifies it to reduce costs again.
(To make this more realistic (1) assume that each ‘member’ in the example is a department; (2) allow all of this to happen more or less simultaneously rather than in a clearly ordered progression; and (3) add additional steps to address further concerns about tax implications, public relations, supplier relationships, etc.) As a result of these piecemeal modifications and others during implementation, each innocuous and rational within its own limited sphere, the new production line results in a continuing discharge that—as it continues unabated over time—pollutes a local river. This new carelessness around environmental impacts is a side effect rather than an intended result, but the effect is the same: Acme has begun to act, consistently, in an environmentally irresponsible manner. As the discharge continues and members occasionally, slowly, become aware of it, it comes to be seen as an accepted practice when in fact it is simply hidden. Later, it thus becomes an accepted practice, and (consciously or not) the members adjust their own expectations and behavior on the job. And Acme takes another step away from its prior commitment to environmental responsibility.
1.3 Cultural shift
Continuing this process brings us to the third mechanism by which Acme can adopt new commitments—one that is even more broadly distributed (and thus even more opaque to the members) and can originate with members at the lowest levels. When processes like the one just described happen repeatedly throughout the firm, Acme becomes a firm that has either lost its prior commitment to environmental responsibility or (more strongly) gained a positive commitment to not being environmental responsible. With that, both old members and (especially) new members coming in will incorporate this new commitment—consciously or not—into their understanding of their jobs. Behavior that conforms to this new commitment will be rewarded or at least ignored, and behavior that contradicts it will be greeted with surprise, skepticism, or outright hostility. This is a cultural shift.
In each of these three cases Acme has adopted a new commitment that does not derive from its members’ commitments or even runs counter to its members’ commitments, and in the latter cases it has done so without the members even being aware that it has done so. The new commitment will nonetheless shape the individual actions of the members in ways that yield reliable corporate actions consistent with that new commitment. Pace Miller and Mäkelä (Reference Miller and Mäkelä2005: 639, 642), there is nothing spooky or mysterious about any of this. The members of the corporate agent, acting idiosyncratically within the parameters established by the structure, can modify that structure and the commitments it embodies in ways that none of them necessarily intended or even noticed.
Because of the way this account incorporates nonstrategic changes, it treats the corporate agent as a less docile beast than that depicted in more formalistic accounts. The intentionality of the corporate agent is far more independent of member commitments, far more resistant to executive control, and far more sensitive to the actions of lower-level members than is traditionally acknowledged. In all of these ways it is more true to actual practice in ways that have important practical and normative implications (see Hess Reference Hess, Tollefsen and Bazargan2019).
2. Kantian Corporate Moral Agency
With that introduction to the mechanics we can turn to the question of whether and how such an agent qualifies as a Kantian moral agent—again, as an agent with the capacities to fulfill the demands of Kantian morality.Footnote 2 The core requirements for moral agency are relatively consistent and uncontroversial across theories (at least in Western accounts): all require rationality, autonomy, and some kind of normative competence—some kind of sensitivity to morally relevant considerations paired with the ability to respond appropriately to them. The first two terms have been extensively debated, and I will not engage those debates here. For our purposes, ‘rationality’ involves a capacity for a certain level of sophisticated intellectual processing and ‘autonomy’ involves a capacity for a certain level of effective action on the basis of intellectual processing. These are minimal threshold capacities of the kind typically taken to distinguish human persons from nonhuman animals. The literature—including the Kantian literature—is full of richer, more demanding interpretations of the terms, but my aim here is simply to begin with the bare minimum and see how far that can take us.
I take the term ‘normative competence’ from Wallace (Reference Wallace1994), and Hindriks (Reference Hindriks2014) emphasizes the significance of this requirement for questions of corporate agency. Some scholars have argued that rational autonomy is not only necessary but sufficient for moral agency, either in its own right or because rational autonomy entails any necessary ‘normative competence’. Kant is often read this way, in fact, and Peter French (Reference French1984) makes this claim. I take the existence of rationally autonomous (in this sense) sociopaths to cast doubt on the adequacy of pure rational autonomy, but the treatment here does not take a position on the issue.
The complexity arises primarily in that third requirement, as each normative theory requires its own distinctive kind of normative competence. A utilitarian moral agent, for example, must be rationally autonomous and able to recognize, quantify, and predict the occurrence of pleasure and pain, while an Aristotelian candidate must be rationally autonomous and capable of forming and intentionally shaping its own character traits (in each case, inter alia). In what follows I consider three increasingly sophisticated and demanding possibilities about what Kant requires in the way of normative competence.
2.1 Bare bones
At the most basic level, two central tasks confront any candidate: it must be able to ‘act only according to that maxim whereby [it] can at the same time will that it should become a universal law’ and it must be able to ‘treat humanity, whether in [its self] or in that of another, always at the same time as an end and never simply as a means’ (Kant Reference Kant and Ellington1993: 30, 36). All of the individual members can presumably do this for themselves, and if a single member could somehow effectively control all of Acme's relevant actions, then that member could force Acme to mimic Kantian moral behavior as a kind of puppet. That is not the question. The question is whether Acme can do these things—whether the necessary commitments could arise and the necessary actions take place as a result of one of the three mechanisms outlined above.
First, the universal law formulation. The process of conforming to Kant's universal law (acting on universalizable maxims) is not essentially different from the process of conforming to positive law—something most firms attend to as a matter of course. If anything, it is probably easier to conform to universal law; at least the universal law is not changing all the time. Obviously Acme could commit itself to acting on universalizable principles, most easily via ‘explicit decision making’. The most plausible possibility—speaking philosophically rather than empirically—is that ranking members could conclude that such a commitment was appropriate for the firm; if they act on this conclusion and successfully implement it, it becomes Acme's commitment. Alternatively, it seems possible that such a commitment could arise via the distributed decision making of lower ranking members, though this seems less likely. In either case the subsequent process of identifying and conforming to relevant principles could be distributed across one or more departments in exactly the same way that conforming to legal requirements is, such that it would be Acme that ‘conformed to universal law’ rather than any individuals within it.
Acme could adopt the humanity formulation in much the same way though here it seems that any of the three methods could be effective. First, ranking members could bring Acme explicitly to adopt a commitment to treating people as ends (respecting rights, fulfilling duties, etc.). Second, lower-ranking members could institute piecemeal processes that, taken together, would have the effect of taking Kantian ‘humanity’ into account in selecting corporate actions. For example, imagine that a midlevel member concerns herself with being especially sensitive to the situations of racial minorities outside the firm, working to establish internal practices that ensure that Acme treats them respectfully and in ways that support their Kantian ‘humanity’. Another member does the same regarding women, another the same regarding LGBTQ issues, while yet another does the same regarding religious minorities. As these specific concerns and sensitivities begin to spread through corporate practices, we can imagine that they would lose focus and begin to foster a more generalized concern with respectful treatment and support that is consistently expressed in Acme's behavior: a corporate adoption of the humanity formulation. Third, if individuals at the lowest levels, with no concern for corporate (collective) policy or practice, consistently chose more respectful options—knowingly or unknowingly and only within their own very limited spheres—then it seems possible that such behavior could likewise become codified in the culture and enforced against members whose behavior would lead corporate actions to deviate from this principle. (Note that the question is not whether corporate culture can lead the members to be more respectful in their individual actions. It is well established that it can, but that is a separate matter. The question is whether the culture could lead multiple individuals to shape their behaviors so that the collective, corporate actions to which they contribute become more respectful. My claim here is that it can. To the skeptic, I would point out that there is little question that this kind of scattered behavior can lead to corporate behavior that is reliably disrespectful.)
Thus, on the account described above, qualifying firms are capable of acting as Kantian moral agents by acting on universalizable principles and treating ‘humanity’ as an end, and the commitments to do so can arise from either intentional, strategic behavior of members or more organically from the daily workings of the agent as a whole. This is a valuable conclusion in its own right, regardless of further claims: to the extent that Kantian morality relies on an objective account of value (of rational autonomy, say, or respect or dignity), then the very fact that a firm can fulfill Kantian obligations in this manner suggests that it ought to do so.
2.2 A more sophisticated reading …
As acknowledged, the foregoing account undoubtedly oversimplifies the role of the Kantian agent. It thus remains possible that corporate agents can fulfill the caricatured role of a Kantian agent outlined above while being unable to meet the real requirements. So what is missing from that quick sketch of Kantian moral practice that corporate agents might still be unable to do? Three possibilities immediately present themselves.
First, perhaps it is not enough to act on universalizable principles; one must be able to give those principles to oneself.Footnote 3 The above discussion of the mechanisms by which corporate agents adopt new commitments, however, reveals that this is exactly what happens. All of the scenarios sketched above describe situations in which Acme itself adopted new commitments. It ‘gave those commitments to itself’ via the work of its executives (identifying Acme's existing commitments, incorporating information about the world, reasoning from those commitments and acting from within their positions at Acme) or via the daily piecemeal processing of its lower-level members. Just so, executives reasoning from their positions within Acme and giving weight to Acme's existing commitments to adopt a new commitment—which is a universalizable principle—would be a process by which Acme gave that principle to itself, and the same can be said for the lower-level mechanisms. (In contrast, note that if Acme's executives simply imposed an alien commitment, giving no weight to Acme's existing commitments—perhaps because of government or shareholder demand—that would not be an instance of Acme's giving a principle to itself. Rather, that would be an alien intrusion into Acme's RPV, a co-option or corruption of the corporate agent's autonomy by an outside force, akin to the brainwashing, hypnosis, and control by evil demons or scientists of the free will literature; see Hess [Reference Hess2014b] for further discussion.)
Second, perhaps it is not enough to treat humanity as an end; one must treat one's own humanity as an end, acting to protect one's capacity for moral action. Given the equivalency Kant draws between ‘humanity’ and rational autonomy, this requires that moral agents be rationally autonomous and that they treat rational autonomy—in themselves and in others—as an end. Human agents fulfill this obligation through respectful treatment that preserves and engages with the rational autonomy (‘humanity’) we encounter in the world and through careful preservation of our own capacities for moral action. Kant gives a detailed discussion of the latter practice in the Metaphysics of Morals, beginning with the general discussion of ‘duties to oneself’ (Reference Kant and Gregor1996: 543ff.). The corporate agents described here certainly are rationally autonomous (in the sense outlined above), and they certainly are capable of engaging respectfully with others—of refusing to exploit or coerce, respecting rights, keeping promises, and so on. Like human agents, they are equally capable of engaging disrespectfully with others, but again it is the ability to act morally that matters, and that they do have. They are likewise capable of developing and preserving (or destroying) their own capacity for moral action: the policies they implement, the operations they regularize, the factors they take into account, the training they provide, and the member behaviors they reward or punish—all have lasting effects on the agent's fundamental capacity to act morally.Footnote 4
Third, perhaps it is not enough merely to act on universalizable principles or treat humanity as an end; one must do so out of respect for the law. This is the most interesting of the three possibilities. In the first section of the Grounding for the Metaphysics of Morals (Reference Kant and Ellington1993), Kant is quite insistent that ‘respect’ plays a crucial role in moral action. On the most extreme interpretation, Kant suggests that we fulfill our duties only when we act out of respect for the law, and we act out of respect for the law only when we ‘exclude the influence of inclination and therewith every object of the will. Hence there is nothing left which can determine the will except objectively the law and subjectively pure respect for this practical law’ (Kant Reference Kant and Ellington1993: 13). This is an extremely demanding standard for moral action, but it seems plausible to suggest that a moral agent should at least be able to act in this way—that is, on purely moral grounds.
Kant himself has given us the standard: a moral agent acts out of respect for the law when it ignores other desires and ends and acts wholly without regard to the agent's own interest, acting as the moral law dictates because the moral law requires it (rather than entirely for other reasons). The mechanisms and examples already outlined describe several ways corporate moral actions could come about, but two points remain to be addressed: first, whether corporate agents can act morally—respecting persons, for example—because the moral law requires it rather than exclusively for some other reason and, second, whether the corporate agent's inability to feel respectful while doing so is significant.
Regarding the first, of course they can. Once we have established the basic rational autonomy of corporate agents—described in the first section of this paper and assumed for the purposes of this discussion—this is not even a hard question. With corporate agents (as with human agents) the truth of the matter is best captured by counterfactual statements: an agent (human or corporate) φs because the moral law requires it if and only if the agent would not have φed in the absence of that law (or had the law required otherwise). Once the agent has adopted the commitment to act on universalizable principles or respect humanity—via any of the three mechanisms described above—it can act on those commitments as reliably and effectively as it acts on other commitments related to profit or the provision of goods and services (or at least as reliably and effectively as human agents do).
The more interesting question is how we would know that a corporate agent had acted morally, out of respect for the law rather than entirely for other reasons. This is an epistemic question, irrelevant to the question of whether corporate agents are Kantian agents, but it is a fair concern. It will be difficult to determine whether a corporate agent has truly acted out of respect for the law—rather than, say, out of enlightened self-interest—for all of the reasons it is difficult to determine whether a human agent has done so. Both are equally capable of lying, equally capable of a kind of self-delusion, and equally likely to be unaware of the true underlying drivers of their actions. In response I suggest that with corporate agents, as with human agents, the truth is best revealed through consistent actions. The best possible evidence of a moral motivation occurs when an agent acts against its own interests, of course, but this cannot be the true standard for moral action because moral action is so often in the best interests of the agent—at least in the long term. The simple fact that an agent benefitted from a moral action does not prove that the motivation was selfish or render the action itself immoral or morally unworthy. In looking for evidence of action out of respect for the law, with corporate agents as with human agents, we are thus reduced to watching the agents’ behavior over extended periods—does the agent consistently act morally?—or, of course, we can just ask them. (The process for asking a corporate agent involves asking one or more members to report, from a privileged ‘inside’ position, on the commitments of the corporate agent rather than on their own.) In each case, the subsequent report is subject to suspicion but can carry some weight.
The second and more significant concern is the fact that, however much corporate agents may respect the moral law, they cannot feel respect for the moral law. With this, we arrive at the first specific demand for phenomenal consciousness in Kantian moral agency. As noted in the introduction, for the purposes of this article I assume that corporate agents are not phenomenally conscious. There is no phenomenal point of view to accompany the rational point of view described above, and so no phenomenal component to the stance of respect for the law. Interestingly, Kant himself explicitly states that respect is not an ‘effect’ (a feeling) but an act of the will.Footnote 5 He acknowledges in that infamous footnote that respect is a feeling (or is perhaps accompanied by a feeling) but insists that it differs from other emotions in that it is ‘self-wrought by a rational concept’. It is an act of the will consisting in the ‘subordination of my will to a law, without mediation of other influences on my sense. … Respect is actually the representation of a worth that infringes on my self-love’ (Kant Reference Kant and Ellington1993: 14). Without digging too deeply into Kantian metaethics, in short: there is a feeling of respect, but it is not the feeling that is doing the work. An agent has respect for the law when it binds itself to the law, subordinating its own desires to the constraints established by that law. On a Kantian reading, one acts out of respect for the law simply by obeying it for non-self-interested reasons, regardless of the accompanying phenomenology or lack thereof. Thus, a nonconscious agent could have respect for the law and act out of respect for the law. With that, it seems that firms and other corporate agents can fulfill all of the requirements that Kant explicitly identifies.
2.3 … and more sophisticated yet
Recent scholarly work suggests that there is still more to Kantian agency, specifically that it requires empathy and emotion in ways rarely acknowledged. For example, Sherman (Reference Sherman and Cohen2014) has suggested that empathy and other moral emotions play an epistemic role in Kantian moral agency by providing access to information that the agent could not otherwise have (see McKenna [Reference McKenna2006] and Russell [Reference Russell2010] for similar claims about moral agency more broadly). The general idea is that our ability to ‘feel with’ another enables us to notice essential details that would otherwise be hidden or deemphasized—information relevant to whether moral action is required and to the precise shape that action should take. This is thus the second point at which Kantian moral agency might seem to require phenomenal consciousness. The concern is that an entity lacking the felt experience of empathy and other moral emotions would miss these cues and either fail to act at all or act inadequately; Sherman mentions offering support, comfort, assistance, and apology in tactful, respectful ways as morally obligatory actions that require these kinds of insights (21–23). Since firms and other corporate agents cannot ‘feel with others’ (or at all), the reasoning goes, they are unable to act morally in these ways and are thus excused from doing so. They are not Kantian moral agents.
While I agree that empathically generated insights can be valuable in our efforts to act morally and that an agent lacking in empathy and other moral emotions might miss some nuances, to suggest that moral action is impossible in the absence of such information seems to overstate the case radically. This is especially true in a Kantian system: Kant is hardly one to rely on felt experience for basic moral function, and Sherman herself acknowledges that empathy plays only a supporting role. Further, the role that empathy is supposed to play here is epistemic. The focus is on gathering information to guide moral action rather than feeling one way or another; the qualitative aspects of empathy (if any) play no role. With that, the members can use their own empathy to gather the relevant information (about the likely feelings and experiences of others) and then offer that information for corporate uptake and processing the same way they offer all of the other information that the corporate agent draws on to form its beliefs and desires. The members are an integral part of the corporate agent; they are not external to it, and reliance on their capacities for corporate function is not illicit ‘outsourcing’ of essential moral capacities.
Not every corporate agent will be able to do this. Drawing on the empathy of the members in this way would require a corporate agent with structures to elicit this information and incorporate it into the corporate decision-making processes outlined above (again, see Goodpaster [2007] and Anderson [2011] for real-world examples). A corporate agent that lacked those structures would be blind to this information if it were not otherwise available. But any corporate agent could have such structures, and it is therefore not true that corporate agents as a class cannot access this information in the absence of their own empathic experiences.
There is thus a twofold response to the claim that Kantian moral agency might require empathy and emotion (and thus phenomenal consciousness): First, these play at most a supporting role in the exercise of Kantian moral agency; they are not essential to it, and therefore the corporate agent's lack of these capacities is not problematic. Second, to the extent that this information is essential, any corporate agent could develop structures enabling it to draw on the empathically generated insights of its members in its efforts to act on universalizable principles, to treat humanity in itself and in others as an end, and to act out of respect for the law.
3. Phenomenal Consciousness?
The considerations presented thus far support the claim that the rationally autonomous corporate agents described in section 1 are fully capable of acting as Kantian moral agents, even to the extent that this requires potentially phenomenal capacities such as respect, empathy, and other moral emotions. Given these sophisticated capacities, is there still something missing? Is there something more to being a Kantian agent than simply being able to fulfill the role of a Kantian agent?
The most obvious ‘something’ that is missing is phenomenal consciousness and all the rich experiences that come with it. As outlined above, corporate agents can believe that humanity is an end in itself, want to behave morally, and even respect the moral law, but there is no feeling or experience behind these states—nothing that it is ‘like’ for them to believe, desire, or respect. As noted, standard theories of intentionality do not require this, and Kant himself was quite dismissive of the role of emotion and other phenomenal experience in his discussion of moral agency. It is nonetheless possible that there is a necessary role for phenomenal experience to play, and this is a deeply appealing idea. Phenomenal experience plays such an enormous, central role in our own experience of moral agency that it is deeply counterintuitive to imagine moral agency operating in absence of such experience. But that is exactly what I propose.
When human agents practice Kantian moral agency, there is ‘something it is like’ for them to do so—the intellectual effort of testing principles, the empathetic tug of another's humanity, perhaps the feeling of respect for the law. I acknowledge that such things are often present during the human exercise of moral agency and that, when present, they can have an impact on both the choices that are made and the way they are made. This is a significant difference between human agents and corporate agents, one that marks a significant difference in the ways they go about the practice of moral agency. But the question is not whether there are differences between human and corporate moral agents. It is obvious that there are. The question is whether those differences matter—whether the missing capacities make corporate agents not only different from human moral agents but therefore incapable of being moral agents.
I am aware of two positive claims along these lines—claims that the lack of phenomenal consciousness makes corporate agents incapable of moral agency (and therefore not bound by moral obligations; such claims would likewise disqualify other nonconscious agents, like artificial intelligences, with the result that they too would be exempt from moral obligations). The first is a conceptual claim perhaps best voiced by David Rönnegard in his The Fallacy of Corporate Moral Agency (Reference Rönnegard2015). Unlike many scholars involved in the debate, Rönnegard explicitly argues for a phenomenalized account of moral agency (24–27).Footnote 6 He does so by describing the familiar (human) process of consciously forming a morally significant intention and acting on it, then arguing that this is simply what moral agency is (see Hess [Reference Hess, Hess, Isaacs and Igneski2018b] regarding difficulties with this methodology). According to Rönnegard, the very concept of moral agency is necessarily drawn from the familiar, phenomenal, human practice of it, and it is illegitimate to use the phrase if we mean something different when we apply it in a different—i.e., corporate—context (Reference Rönnegard2015: 9–14, 24–27). Regardless of any other competencies and capacities that corporate agents might have, they cannot be ‘moral agents’.
The difficulty with this objection is that it claims too much or too little. The argument rests on the claim that moral language and concepts are in their essence human-specific. Taken seriously, that commitment closes off any possibility of a nonhuman moral agent, however similar to human beings. That seems implausible and claims too much. It invites familiar slippery-slope strategies of argumentation in which the proponent begins with a human agent (or perhaps a Martian indistinguishable from a human agent) and changes one tiny thing at a time, challenging the critic to identify the point at which it is no longer a moral agent (as some human beings are not).
Here the critic has two choices: Her first option is to engage in line-drawing. To do so, she will have to say something about why the particular feature or capacity in question—the point at which she draws the line—is essential to moral agency. Regarding phenomenal consciousness that would bring us back to exactly this point in the dialectic, trying to determine why phenomenal consciousness might be necessary for moral agency. That move thus does not advance the discussion or respond to the considerations presented here. The critic's second option is to bite the bullet and affirm this limitation despite any alleged implausibility: only human beings (or things indistinguishable from human beings) can be ‘moral agents’, and ‘moral’ language is restricted to the activities of such entities.
At this point the proponent can simply grant the critic's conceptual claim about the essential humanness of moral language and reply as follows: ‘The central, motivating concern here is whether firms (and other corporate agents) may act in certain ways—whether it is appropriate for them to treat other entities in certain ways, to bring about certain goods or bads, to move both their internal and external worlds in certain directions. These are coherent, urgent questions that need to be explored and answered. They obviously parallel familiar moral concerns (to the extent that they do not simply embody them), and we have thus used moral language to pursue that project. If the traditional moral terms are to be reserved for human practices, then we will simply coin new, parallel terms (‘schmoral agents’, ‘schmoral obligations’) that lack the human essentialism claimed for the traditional moral terms and continue the project unaffected by the above claims’.
Here the critic has thus claimed too little, claiming rights to the words but failing to touch the central question; surely her objection ran deeper than that, but it is all she has earned with her conceptual claim. The onus is now on the critic to explain why this move is unavailable.
The remaining positive claim that Kantian moral agency might require phenomenal consciousness turns on motivation. The idea here is that an agent who lacked phenomenal consciousness and thus could not feel the sting of shame or remorse at moral failure, the glow of pride at moral success, or the urgency of another's plight could not be motivated to act rightly. This claim seems (ironically) unmotivated. A human agent might feel no particular emotion in connection with the thought of Syrian refugees adrift and vulnerable yet still donate to relief efforts simply because she believes it is the right thing to do; in Kantian terms, she might act out of respect for the law. Further, even if it were (somehow) impossible for human agents who lack empathy (or other phenomenally rich states) to act morally, this tells us nothing about what is necessary for Kantian moral agency per se. It simply tells us what is necessary for human beings to act morally, at which point this becomes a psychological claim rather than a philosophical one.
Kant himself was careful to separate the two, dealing with normative moral claims in the Grounding and the Metaphysics of Morals and human psychology in his Anthropology. As Kleingeld notes, ‘within Kantian ethics empirical psychology does not play a role at the level of the justification of the basic moral principle. As Kant put it, such empirical knowledge “must not precede a metaphysics of morals or be mixed with it”’ (Reference Kleingeld and Cohen2014:160, quoting the Metaphysics). Sherman puts the point beautifully when she describes Kant's efforts to reconcile the stringent theoretical demands of duty in his moral writings with the need to realize duties in ‘human material’ in his Anthropology (Sherman Reference Sherman1995: 375). If we fail to heed Kant's warning, and thus fail to follow his own practice in separating these questions, then our understandings of agency and moral agency will always be limited to the human versions of these things. It is reasonable and coherent—and increasingly urgent—to answer questions about how these corporate agents should act. Drawing on unduly humanized accounts to answer these questions defeats such an inquiry before it even begins.
Conclusion
I have argued that the corporate agents described in section 1 possess the normative competence required for Kantian moral agency—that they can use the mechanisms that enable their rational agency to act morally by giving principles to themselves, treating their own ‘humanity’ as an end in itself, and acting out of respect for the law, even to the extent that any of this requires insights from empathy and other moral emotions. The lack of phenomenal consciousness is not dispositive. Note that the argument here expresses no commitments regarding the relationship between rational and moral agency more generally. My only claim is that the corporate agents described here are capable of both and that the kinds of mechanisms that enable corporate agency can also enable Kantian moral agency. I make no claims about instantiations in human or other ‘material’. Corporate agents can act morally, in Kantian terms, in their own right, and it follows that they are Kantian moral agents that should do so. They will not feel bad if they do not (though the rest of us will), but this is irrelevant. A human agent able to comply with moral imperatives should do so regardless of how she feels about it (or whether she feels at all), and a corporate agent is the same. To the extent that Kant is right about the demands of morality, firms and other corporate agents are bound to abide by it.
This raises two significant issues I can address only briefly here. The most pressing is the question of whether Kantian moral agents are necessarily Kantian moral subjects (or patients) and thus morally considerable. If so, the claim that firms and other corporate agents are Kantian moral agents will entail that they are morally considerable:
(1) According to Kant, all moral agents are morally considerable.
(2) Corporate agents are moral agents.
(3) ∴ Corporate agents are morally considerable.
Ethically, P3 will suggest that corporate agents must be treated as ends in themselves; politically, P3 will suggest that they are entitled to a significant array of rights and protections, perhaps even the same rights and protections as human persons. Critics of corporate moral agency can affirm P1 (with or without the antecedent limiting the claim to Kantian systems) and can then argue that the remainder of the argument presents a reductio for corporate moral agency—that the conclusion is absurd and that we should reject P2. In response, scholars more sympathetic to corporate moral agency have pursued two general strategies.
The first general strategy available to proponents of corporate moral agency is to avoid the absurdity by rejecting P1 rather than P2. Manning takes this tack, arguing that ‘corporations’ can be moral agents without being ‘moral persons’—by which she means ‘rightsholders’ entitled to ‘fair treatment’ (Reference Manning1984: 79)—because the criteria for the two are different. Moral agency requires only rational autonomy (essentially), while ‘personhood’ requires the ability to feel pain (80–81). Wringe acknowledges the bite of the above argument in the context of Kantian ethics and the critic's option above, but he describes it as ‘uncomfortably ad hoc’ and an ‘unmotivated kind of “individualist chauvinism”’ particularly at odds with Kantian commitments to the significance of rational nature wherever it may be. He concludes by suggesting instead that we abandon—or at least revisit—our Kantian commitments (including P1 above; Wringe Reference Wringe2014: 280). Hindriks acknowledges this option without necessarily endorsing it, noting with Manning the possibility that moral considerability might require sentience (Reference Hindriks2014: 1579–80).
The second general strategy is to accept the argument in its entirety but deny that P3 is absurd. Hindriks acknowledges this option as well, though again without necessarily endorsing it. He takes ‘agency’ as the primary marker for moral status but suggests that we can avoid absurd outcomes by considering the possibility that agency (or autonomy) itself comes in degrees. If we accept this possibility then the potentially lesser autonomy of corporate agents may entail that they are entitled to weaker moral consideration or perhaps to none at all in their own rights (Hindriks Reference Hindriks2014: 1580–82). Silver (Reference Silver2018) pursues this strategy with considerably more enthusiasm. He argues at length that proponents of corporate moral agency are bound to accept the entailment at P1 (on Kantian and other grounds) and then argues that P3 is neither absurd nor even particularly problematic. Corporate agents are sufficiently different from human agents that there is no reason to think that the ‘consideration’ to which they are entitled will look anything like the ‘consideration’ to which human agents are entitled. In fact, he concludes that current practices may even be sufficient or require only slight modification; his primary demand is that we give the matter the attention it deserves.
It is well beyond the scope of this article for me to address the matter in any detail here. Briefly, however, I am inclined to follow Manning and Wringe and suggest that we reexamine our assumptions about the Kantian ethical system or about the criteria for moral statuses more generally. One of the major benefits of work in collective and corporate moral agency is the way that it exposes unquestioned assumptions in our work on human moral agency, and this may be just such a case. Having argued at length that phenomenal consciousness is less important for Kantian moral agency than might be assumed, it seems entirely plausible that it might be far more important for Kantian moral considerability than any of us have yet considered (Kant included). I have argued elsewhere (Hess Reference Hess2013) that corporate agents are not entitled to the same political rights as human persons, but have remained silent regarding whether they are morally considerable. I am neither convinced nor irrevocably opposed to the claim that they are, and I find Hindriks's and Silver's suggestions intriguing and worthy of development.
There is a second significant issue left open by this treatment arising from the fact that what I have argued for here is moral obligation, a prospective, forward-looking form of moral responsibility. In contrast, much of the debate about whether collectives of various kinds can be ‘morally responsible’ has focused almost exclusively on the retrospective, backward-looking questions of whether collectives are apt targets for praise and blame. To the extent the two statuses are mutually entailed and the two classes of entities (those that are morally obligated and those that are apt targets for praise and blame) are thus coextensive, this does not matter. For those scholars who believe that the two statuses come apart, however, this leaves the possibility that corporate agents have moral obligation without moral responsibility—perhaps because the corporate agents lack reactive attitudes (but see Björnsson and Hess Reference Björnsson and Hess2017) or because, again, they lack phenomenal consciousness. I address this briefly in my ‘Who's Responsible? (It's Complicated.): Assigning Blame in the Wake of the Financial Crisis’ (Hess Reference Hess2018a) but leave a full treatment of that question for future work.
