2. The case for the moral animal

Often it can seem as if an animal’s actions are guided by moral concerns. Some anecdotal examples of animals who appear to be acting morally are so old as to have their own literary traditions: reports of dolphins rescuing sailors from shipwrecks, for example, occur as early as Plutarch, and the trope of the loyal and heroic dog has been a cultural touchstone throughout much of human history, persisting from the Welsh fable of Gelert to Lassie of the silver screen. But the ubiquity of moral animals in popular culture is hardly a reason to take them to exist, otherwise we might find ourselves believing in witches, ghosts, and all other manner of folk superstitions. Furthermore, there are few rules considered more fundamental to the study of animal minds than Morgan’s Canon, which warns that “in no case is an animal activity to be interpreted in terms of higher psychological processes if it can be fairly interpreted in terms of processes which stand lower in the scale of psychological evolution and development” (Reference Morgan1894).

On the face of it, Morgan’s Canon seems to tell against the hypothesis that there are moral animals in that whenever an animal appears to act morally, there are ways to explain that animal’s behavior in terms of less sophisticated mental processes. To illustrate, imagine that you are visibly upset, and your beloved spaniel Brownie rests his head on your knee in a way that gives you comfort. One explanation of Brownie’s behavior is that Brownie, sensitive creature that he is, recognizes your distress and intends to alleviate it. On this explanation, Brownie behaves morally. However, consider, as Morgan’s Canon would have us do, the spectrum of alternate, lower-level explanations: perhaps Brownie, mindless brute that he is, acts purely on instinct (perhaps such behavior was adaptive at some point in his species’ developmental history). Or perhaps he has been praised for similar actions in the past, and has thus learned to respond in this manner through operant conditioning. Or perhaps he more mindfully—but still amorally—takes you to be the leader of his “pack” and recognizes that you are an ineffective guardian in your current state; alleviating your distress, whether or not Brownie recognizes it as distress, is merely a way for him to restore you to the condition of a reliable leader and protector. What is it that has led some researchers to think that the moral explanation for animal behavior like Brownie’s is often the best explanation, despite the psychological complexity of the processes involved?

3. Animals and moral judgment

So far, I have established that the existence of the moral animal is a hypothesis that the scientific community considers a live option and has investigated accordingly, as well as one that many philosophers have been compelled to accept. I have also noted that some philosophers have gone so far as to defend animal moral agency or similar views that have as a consequence that animals are in some substantive sense responsible for their moral (or immoral) actions. Now I shall argue that if moral animals do exist, whether they are agents or no, they are likely to be severely limited in their ability to morally evaluate one another—if they are even capable of doing so at all. My further claim is that the severity of these deficits is such that impartial human observers would generally be more competent to see to matters of animal desert than the moral animals themselves. This will prepare the way for my arguments in the following sections that if animal moral agents exist, then such human observers, due to their greater competence, have a strong reason and perhaps even an obligation to play this unusual role.

As human beings, we enjoy not only the capacity to act for moral reasons, but also the capacity to judge the actions of actions as moral or immoral. Although they are closely related, these are competencies that can come apart. In some respects, the latter capacity is more cognitively demanding. Judging others arguably requires a theory of mind that recognizes not only that another person is an entity that takes in the world from its own unique perspective—and to whom we might incur a moral obligation—but also that they are the kind of entity that can incur moral obligations of their own. In other words, returning to our example from section 1, even if we can make a good case for Brownie being a moral agent, we cannot assume without argument that he is also capable of judging the actions of his compatriots at the dog park. It may well be that Brownie and creatures like him are wholly incapable of moral evaluation. And if moral animals cannot morally evaluate, then they cannot hold one another to account.

This is a little too quick. Although we cannot assume all moral animals, or even animal moral agents, are moral evaluators as well, that is not to say that no argument can be made for that further achievement. Recall the enforcement of social norms among animals discussed in 2.d. As I discussed previously, I do not think such behaviors are incontrovertibly expressions of moral normativity, but let us set that objection aside for the time being. If we grant that animals who enforce social norms are holding each other to a standard of moral conduct, do we then preempt any argument that humans ought to also be in the business of punishing immoral animals? I think not. For given their cognitive limitations, even if we took animals enforcement of social norms as evidence that some animals are moral evaluators, such animals are nevertheless unlikely to be effective moral evaluators.

It is easy to see why this is the case. As I said above, the capacity to morally judge is more sophisticated than the capacity to morally act, probably even requiring at least a rudimentary theory of mind. Even if you are sympathetic to the notion of animal theory of mind, the further informational demands of moral evaluation are great. Often, for example, internal factors, such as an agent’s intentions or beliefs about the situation, are relevant to whether we find them worthy of blame or praise. In our Brownie example, for instance, it made a difference whether Brownie comforted his upset owner because he simply wanted to soothe hem or whether he did so because he, more self-interestedly, wanted to make hem a more alert and capable leader. Many believe that only in the former case is Brownie genuinely praiseworthy. Accordingly, for an animal to judge the actions of another, they may well need to infer facts about that other’s motives or epistemic situation. For creatures of limited intelligence, this will be a difficult task.

Indeed, there are some relevant factors that even very intelligent animals will have difficulty tracking. Consider, for instance, that false beliefs sometimes excuse one from being fully blameworthy for a given harm: if I had every reason to believe I was giving you medicine rather than poison, I am not culpable for poisoning you. Yet, to single out an animal thought to be among the most sophisticated social cognizers—chimpanzees—by all experimental indications, they are incapable of attributing false beliefs to others notwithstanding whatever other theory of mind abilities they might have (Call and Tomasello Reference Call and Tomasello1999; Call and Tomasello Reference Call and Tomasello2008; Hare, Call, and Tomasello Reference Hare, Call and Tomasello2001; Krachun et al. Reference Krachun, Carpenter, Call and Tomasello2010). So, even if a chimpanzee could attribute blame, he would likely not be able to take false beliefs into account as a mitigating or exculpatory factor.

There are a range of other cognitive thresholds a competent moral assessor must meet. To name but a few, judgment often involves attending to the actions of others without distraction, remembering past events and histories of behavior, differentiating unprovoked outbursts from those that are given cause. All of these demands present challenges to nonhuman animals, who typically fall short along one or other of these dimensions. Even if there were animals that are capable of some form of moral judgment, then they would almost certainly be deeply flawed evaluators. Indeed, they could arguably do no better than a human observer with even a rudimentary grasp of their species’ behavioral repertoire. Even very young humans, for example, are adept at attributing false beliefs. Furthermore, generations of successful fieldwork in comparative cognition prove that human observers are well able to observe animal groups, differentiate their members, and systematize and ultimately interpret many of their behaviors. The notion that such researchers might be able to apply moral as well as psychological criteria to animal behaviors as well is, at least, not prima facie absurd.

These epistemic and cognitive limitations are far from the only reasons that humans might fare better than moral animals in identifying instances of animal wrongdoing, exculpatory or aggravating circumstances, and so forth. There are also significant practical constraints on the ability of animals to punish one another. Because punishers incur risk, we should expect the circumstances in which animals are willing to punish instances of wrongdoing to be limited, and empirical work confirms this prediction. Recall that the chimpanzee was cited above as an animal one might argue does mete out punishments within the group, at least where thievery is concerned. It seems, however, that chimpanzees do not mete out third-party punishments of this kind: although they will sometimes aggress against those whom they catch stealing their own food, Riedl et al. (Reference Riedl, Jensen, Call and Tomasello2012) have shown that chimpanzees are unwilling to act when the stolen food belongs not to themselves but to another chimpanzee. Such punishments, then, are personal rather than impartial.

Furthermore, to continue on the subject of chimpanzees, even when their own food has been stolen, chimpanzees are significantly less likely to retaliate when the thief is a dominant individual, presumably because of the risk they would incur by doing so; whereas, when dominant individuals are the victims of food theft, they tend to retaliate swiftly (Riedl et al. Reference Riedl, Jensen, Call and Tomasello2012). This is not the only way animal punishment is likely to go awry where dominance hierarchies are concerned. Some species, such as mandrills (Schino and Sciarretta Reference Schino and Sciarretta2015) and Japanese macaques (Aureli et al. Reference Aureli, Cozzolino, Cordischi and Scucchi1992), practice a form of redirected aggression whereby instead of punishing a more dominant individual for a perceived slight, the wronged subordinate attacks the wrongdoer’s more vulnerable kin.

So, even if animals do sometimes have a disposition to punish, it remains the case that they won’t consistently punish the appropriate offender, or will punish in an otherwise inappropriate manner. In such cases, an impartial human observer who is not at the mercy of an animal group’s dominance hierarchies might well be better suited to punish its wrongdoers than are the animals themselves. Impartiality, to the extent that it can be achieved, has long been taken to be one of the most fundamental desiderata of adjudicative processes. Indeed, the putative impartiality of the human criminal justice system is considered one of the strongest justifications for its existence.

If animal moral agents do exist and, as I have suggested, humans are likely to often be in the best position to morally evaluate them, then would humans thus be obligated, or at least given strong reason, to punish animal wrongdoers? One might think we surely cannot be so obligated for, in many respects, the prospect of discharging such an obligation seems absurd. Ought we hold trials? The prospect calls to mind the darkly comic 1993 film The Advocate, in which Colin Firth plays a medieval public defender whose “client” is a pig accused of murder.Footnote ⁶

Could we possibly have reason to fastidiously monitor the social world of animals, occasionally stepping in to, say, dethrone a particularly sadistic chimpanzee alpha or to save from retaliation a low-ranking member of his troop whom he mistakenly believes to have slighted him in some way? Let us call this position human punishers:

4. Retributivism and animal punishment

In this section, I will show that if you hold that animals are moral agents and if you are a retributivist about punishment, then you have good (though not insurmountable) reason to accept human punishers.Footnote ⁷

Retributivism is the view that punishing wrongdoers is intrinsically good regardless of any extrinsic benefit it confers, such as deterrence. On strong versions of the view, we are invested not only with the right but also the duty to punish wrongdoers. Depending on the extent to which animals are thought to attain standing as moral agents, then, we might expect strong retributivists to hold that humans are obligated to play the role of the world’s policeman in the animal kingdom if they are (within reason) able to do so. Weaker forms of retributivism, which acknowledge a reason but not an obligation to punish the guilty, may sometimes escape incurring the obligation to punish animals, but can still be expected to concede that it is often permissible to do so depending on the circumstances and what other reasons are in play.

Just how many retributivists fall within the strong as opposed to the weak camp is a matter of some debate, but it is likely that a great deal of them do, perhaps even the majority. For example, if Michael Moore is right, there is a sense in which all forms of retributivism entail not merely a right to punish but also an obligation to do so. Moore writes:

Moore is certainly painting with a broad brush, and not all retributivists will assent to his characterization of their position. Still, many retributivists do explicitly endorse the form of reasoning Moore sets out and situate themselves in the strong retributivist camp. Take Immanuel Kant, for instance, who argues that

Retributivism goes hand in hand with the view that there is a sense in which all instances of moral wrongdoing, both public and private, merit criminalization, though the question of whether we should in fact criminalize a given type of wrongdoing is informed by further considerations, both principled and pragmatic (Duff Reference Duff2014). One might expect these further considerations, especially those of principle, to adjudicate against animal punishment and thus let the retributivist off the hook—but let us see how far we can go.

Perhaps the most obvious candidate reason that the retributivist might be able to refrain from dispensing animal punishments on principled grounds concerns jurisdiction. Even if animal punishment is a good, the retributivist might argue, surely humans do not have the jurisdiction to mete it out. This sounds very plausible on its face. With the possible exception of certain domesticated animals, human life proceeds largely at a remove from the lives of the animals who are candidates for punishment. Historically, we have not participated in animal systems of punishment and reward, nor have animals participated in ours. Thus, there is a sense in which alleged wrongdoing in the animal kingdom seems to be “none of our business”—it is regrettable, perhaps, but not something we have the authority to address, given that we are in no sense members of their communities. For a human interloper to hold an animal to account would be to overstep or interfere somehow, the intuition goes—similar to if extraterrestrials were to suddenly alight on Earth and attempt to try us for our own misdeeds.

This turns out to be a tricky response for the retributivist to retreat toward. This is because retributivists tend to take a straightforwardly instrumentalist approach to the question of who should punish wrongdoers, holding that punishment should be meted out however is most effective. Indeed, similar approaches are more or less conventional in the philosophy of punishment in general, and not just within the retributivist literature. As it manifests in the retributivist framework, however, the instrumentalist argument runs as follows. Given that the punishment of wrongdoers is a prima facie good, its value is not contingent on who realizes that good. Wrongdoers should therefore be punished by whoever is best positioned to punish them, i.e., whoever can deliver that punishment most effectively and efficiently. In civil society, this will often be the state, but it needn’t be.

John Locke is perhaps the earliest philosopher to explicitly subscribe to this instrumentalist line of reasoning. When it comes to punishing wrongdoers, Locke (Reference Locke and Cox1689) held that “every man hath a right to punish the offender, and be executioner of the law of nature.” For Locke, whatever special authority and jurisdiction a state has to punish its subjects is grounded only in its ability to “restrain the violence and partiality of men,” i.e., to dispense proportionate punishment more reliably and with less bias than can individuals who independently hold each other accountable (Reference Locke and Cox1689). Andrew von Hirsch expresses a similar sentiment in Doing Justice (Reference Von Hirsch1976): “[T]here will be less social disruption,” he writes, “if offenders are punished by the state rather than left to private retaliation.”

Not only does this argument tend to favor state arbiters of punishment over their vigilante counterparts, it also tends to support the restriction of a state’s jurisdiction to wrongdoings that occur within its own borders, again purely on instrumental grounds. For the instrumentalist, there is no principled reason why the government of the United States is entitled to mete out punishment to its own residents but not to those of the United Kingdom, Austria, or Morocco. Rather, there is only the pragmatic reason that, in our world of sovereign states, wrongdoers are typically punished more effectively and efficiently when countries “police their own.”

Of course, by their very nature, instrumentalist restrictions on jurisdiction are sound only insofar as things do go better when justice is dispensed officially by a state rather than informally by its citizens, or when it is dispensed autonomously within states rather than internationally. If a country were to fail to dispense justice to its people, for example, this might constitute a reason for some other state or group of states to take up that burden.Footnote ⁸

Similarly, if a group of people existed outside the jurisdictional authority of any state and relied on a system of vigilante justice inferior to that which a state could provide, neighboring states might have sufficient moral reason to extend their jurisdictions to encompass members of that group. This, I argue, is the situation in the case of animals should they turn out to be moral agents. As I have discussed, even if some animals are moral agents, they are likely to be extremely limited in their ability to reliably call wrongdoers to account. Furthermore, animals are often brought low in their attempts to sanction one another by what Locke would call their innate “violence and partiality” (Locke Reference Locke and Cox1689): the chimpanzees discussed earlier who punished only subordinate food thieves are but one example of this phenomenon. Humans have the advantage of not being embedded within the animal dominance hierarchies that make retribution difficult to exact against high-status individuals. Indeed, one can take as a proof of concept the ways in which humans benefit from official systems of law and order that are designed to curtail our own baser instincts.

Thus, if animals are moral agents deserving of punishment, those retributivists who are instrumentalists about jurisdiction (as most of them are) ought to recognize a duty to step in to punish animal wrongdoers when doing so is more likely than nonintervention to implement a principle of retributive justice. This is of course only a prima facie duty—there are a number of possible pragmatic reasons to leave animals to their own devices. However, if the only reasons not to punish moral animals are pragmatic, that is still an extremely interesting result since, presumably, those pragmatic reasons won’t always be decisive.

Are there other principled grounds besides that of jurisdiction on which the retributivist might resist incurring the obligation to punish animals? I will start by going over one tempting response which I do not think does the trick, namely what I call the objection from relativism, before turning to what I take to be better responses available to the retributivist who hopers to accept animal moral agency without the baggage of human punishers.

5. Can the relativist avoid animal punishment?

The objection to human punishers from relativism begins with the thought that the moral norms of animals may be very different from our own, and concludes with the claim that such differences would release us from any obligation to punish animal wrongdoers, and perhaps even prohibit us from doing so.

The first thing to note is that the objection from relativism is not an objection from all forms of relativism since some forms of the view do not pose a problem for human punishers.Footnote ⁹ To illustrate: on one version of moral relativism, which John Tilley calls Appraiser Relativism (2000), the truth of a moral judgment depends upon the norms of the community to which the one making the judgment belongs.Footnote ¹⁰ According to Appraiser Relativism, the sentence “Slavery is wrong” is true when uttered by someone in twenty-first-century America, but false when uttered in ancient Greece. If anything, Appraiser Relativism actually vindicates the notion that humans should punish animals they take to be immoral for the ones making such judgments are human and, according to human standards, we should indeed punish those who act as the animals in question have acted.

The objection from relativism is grounded not in Appraiser Relativism then, but rather in its contrast, Agent Relativism (Tilley Reference Tilley2000). According to Agent Relativism, the truth of moral judgments is grounded in the cultural norms not of the judge of an action but those of the agent who committed it (Reference Tilley2000). If Agent Relativism is correct, then it is wrong for a person to impose the moral norms of her own community on those living in communities with differing norms for the two communities have equally valid yet irreconcilable systems. So, for example, perhaps we should not attempt to stop some society from torturing their criminals even if such punishments are unjust by our own society’s standards. Or, in the animal case, even if animals do turn out to be moral agents, then we humans ought not to punish them, for the animals’ moral communities adhere to different moral norms than our own. This is the objection from relativism.

I will not attempt to argue against Agent Relativism on the truth of which the objection from relativism depends. For one, there is already work in the literature to this effect (Tilley Reference Tilley2000). But more importantly, I contend that the objection from relativism is flawed even if Agent Relativism is true.

There are two basic problems with the objection from relativism. First, the fact that we cannot punish immoral animals according to our own norms does not entail that we cannot punish them at all, for we may be able to identify some of the basic moral principles of their societies and then be fully licensed in bringing their wrongdoers to justice even if their notions of wrongdoing or justice don’t quite line up with our own. In section 3, I briefly discussed some observable patterns of negative reciprocity in the animal kingdom—wolves who won’t play with unfair players, chimpanzees who punish theft, etc. Even if we can’t ascertain an animal community’s full set of moral norms, it seems plausible that observable behaviors like these could help us piece together some proper subset of those norms. If this is right, then the arguments of my paper would still apply for breaches of norms belonging to that subset. Ergo, the believer in animal moral agency who embraces Agent Relativism cannot fully shirk the arguments of this paper unless the moral norms of animals are truly so alien as to be fully beyond our discernment. Yet if the moral norms of animals are truly so alien as to be fully beyond our discernment, then the objection from relativism will still fail, as the second problem I raise will make clear.

The second problem with the objection from relativism is that it is self-undermining. Since the objection is intended to save the believer in animal morality from also incurring reason to punish bad animal actors, it takes as a premise the claim that some animals are moral. Section 2 of this paper outlined some reasons one might want to endorse this premise, yet all of those reasons are fundamentally grounded in the connection between animal behavior and human morality. For example, animals seem to punish behavior that is recognizably immoral—i.e., immoral by our own lights. Or, animals seem to endure pain in order to avoid causing harm to their peers, something that we take to be good. The assumption that these behaviors are evidence for animal morality is underwritten by the deeper assumption that if animals were moral, then their sense of morality would observably resemble our own. If the relativist accepts this deeper assumption, they can justify their belief in the existence of moral animals by referencing the literature reviewed in section 2, but if they want to go so far as to accept animal moral agency on the basis of the same, they must then contend with the fact that there is animal conduct that, by their own lights, is worthy of punishment. If the relativist rejects this deeper assumption, they can avoid my arguments above, but at the cost of now having little clear evidence to justify their belief in the existence of moral animals. At best, they now defend only an exotic counterfactual: if there were moral animal agents (whose morality would fail to line up with our own in any observable sense), then humans ought not punish them. Since we have no reason to believe the antecedent of that conditional, the objection from relativism turns out to be of limited interest.

There are, however, more promising grounds on which the retributivist might mount a principled objection against incurring the obligation to punish animal moral agents exacted by human punishers.

6. Can cognitive shortcomings excuse animal moral agents from punishment?

One common proposal about moral responsibility is that an agent must possess certain capacities in order to be morally responsible (or at least in order to be liable for punishment should liability and moral responsibility come apart). (See, e.g., Wallace [Reference Wallace1994] and Fischer [Reference Fischer1999].) Which capacities are named as the relevant ones varies across accounts, but the capacity that is most often singled out is the capacity to respond to moral reasons in a way that is consistent (Wallace Reference Wallace1994; Fischer Reference Fischer1999). Someone who believes in the existence of animal moral agents but wants to avoid punishing them might argue, then, that while some animals are genuine moral actors, they are too hit-or-miss when it comes to responding to moral reasons to lay claim to the consistency that is required to be fully possessed of the relevant capacity—whatever agency they possess, on this understanding, falls short of what is required for them to be liable for their actions.

I will grant the assumption that any capacity to respond to moral reasons that an animal might possess is likely to be a fairly local, contingent capacity. An animal that is motivated to prevent another animal from being hurt may not be motivated to prevent the same from being stolen from, or even painlessly killed; alternately, an animal that is motivated to prevent all of these things normally may fail to be so motivated when they are hungry or distracted. Furthermore, as discussed previously, animals might be more consistent at moderating their own behavior than sanctioning the behavior of others for various epistemic and pragmatic reasons.

But while I grant that animals’ capacities for responding to moral reasons are plausibly local, I reject the premise that a global capacity to respond to moral reasons is necessary for punishment to be justified. When circumstances do permit the exercise of an animal’s (local) moral capacity, it follows that animal should be held liable for failing to act as they ought (commensurate with their ability to have acted otherwise). This is in line with how we think about human agency and responsibility. A man with an impulse control disorder, for example, might be unable to consistently respond to reasons not to steal. Nonetheless, he may be able to consistently respond to reasons generated by the welfare of others, and, for that reason, he can be held responsible for striking a mall cop who catches him stealing even if he is in some sense not responsible for the theft itself. To the extent that we can single out a range of circumstances within which an animal moral agent is responsive to moral reasons (which is no small feat), then within that range one is licensed to punish the animal’s performance accordingly. Thus, if we can meet the largely empirical burden of establishing these local spheres of moral competence in animal species, then we will still face situations where we have reason to punish.

So although an objection to animal moral responsibility on the basis of an animal’s global moral inconsistencies cannot succeed on its own, it can make some traction in that as the range of cases where animal moral agency obtains narrows, so too does the range of cases in which human punishers is potentially in effect.

What of animals’ other shortcomings? Might we turn to them to argue that even animal moral agents operating fully within their local spheres of moral competence are not liable for their wrongdoing such that we ought to punish them? A retributivist might have more luck avoiding human punishers by invoking certain strategies that have been used to shield young children and the criminally insane from retributivist punishment. This strain of response is in keeping with the standard retributivist literature, which typically endorses a relatively high punishment liability threshold, one that children are often thought to be incapable of meeting because of low intelligence, irrationality, or poor impulse control. Yet insofar as an animal moral agent is a moral agent as regards some particular act of wrongdoing, then low intelligence, irrationality, or poor impulse control won’t suffice to fully excuse their behavior. However, these factors can certainly be mitigating, perhaps even substantially so, and perhaps even enough to bring them below the liability threshold for punishment. This position is in parallel with that of those courts and legal scholars who have argued that diminished rationality only mitigates or partially excuses criminal culpability in the case of human offenders (e.g., Atkins v. Virginia Reference Virginia2002; Morse Reference Morse2003; Steinberg and Scott Reference Steinberg and Scott2003). But see also Cynthia Ward’s provocative arguments that even children as young as six can sometimes meet the standard of mens rea when they act deliberately and with knowledge of the inflicted harm (Reference Ward2006). Depending on where the retributivist stands on these matters, this is a promising tactic for the retributivist looking to secure moral animal agents a more complete reprieve from punishment.

It should be stressed, however, that a retributivist who takes this approach walks a fine line in holding both that some animals are moral agents and these animals’ diminished mental capacities excuse them from punishment.

The comparison to the status of children is illuminating here. Consider, for example, the active “children’s rights movement” of the 1970s and 1980s, which sought and won various legal victories on behalf of minors, such as the right to procure an abortion without parental consent. A major part of these activists’ strategy was to draw on work in psychology and the social sciences that showed adolescents were less impulsive and more rational than previously assumed. The successes these activists achieved by this strategy, however, had spiraling consequences, as the very same research was soon used to argue on behalf of stricter punishments for children, and even for the abolishment of a separate juvenile justice system.

As Elizabeth Cauffman and colleagues write (Reference Cauffman, Woolard and Reppucci1999), the children’s rights movement’s appeal to the rational agency of the young proved a “double-edged sword,” for “[h]ow could adolescents be mature enough to make their own decisions about abortion, but not mature enough to face the consequences of committing armed robbery or using marijuana?”Footnote ¹¹ Similarly, those who want to argue that animals can respond to moral reasons as agents but are not morally responsible may not be able to easily have their cake and eat it, too. Research suggesting animals are capable of genuine altruism or malice, especially when coupled with recent, more general work on animal rationality and decision-making, might just as easily be repurposed to argue that those animals are liable for the consequences of their actions.

Furthermore, although there are various conceptions of moral agency at large in the literature, a great deal of them are inextricably linked with moral responsibility. Take Mark Rowlands’ definition, for example, according to which “X is a moral agent if and only if X is (a) morally responsible for, and so can be (b) morally evaluated (praised or blamed, broadly understood) for, his or her motives and actions” (Reference Rowlands2013). On conceptions such as Rowlands’s, agency and moral responsibility really cannot come apart, and so the retributivist who holds (as Rowlands does not) that some animals are moral agents will still need to contend with the question of punishment.

For these reasons, although the considerations raised above may be of some help for retributivists who want to accept animal moral agency but reject some of its less-savory consequences, I think it’s equally as likely that they will lead retributivists to rethink their commitment to animal moral agency in the first place.

7. Punishment as communication

I shall now discuss one final way the retributivist might block human punishers on principled grounds while preserving a commitment to animal moral agency. The retributivist can hold that the expressive or communicative character of punishment is what justifies its exercise. For example, according to one advocate of such a view, R. A. Duff, the hard treatment of punishment is “a material and forceful expression of the apology that [the wrongdoer] owes to those whom she wronged—to the direct victim of her crime, if there was one, and to the wider community whose values she flouted” (Reference Duff2003).Footnote ¹² Punishment is a “reciprocal and rational” engagement with the offender, who “is expected (but not compelled) to understand that he committed a wrong for which the community now properly censures him” (Duff Reference Duff2001).

In the case of animal wrongdoers, the difficulty with human punishers on this kind of communicative account is straightforward: we cannot reasonably expect an animal wrongdoer to understand that whatever hard treatment we might mete out is meant as censure. On a communicative retributivist view, an animal may well be capable of acting wrongly or rightly, and may well be responsible for their actions accordingly, but what matters for punishment is not just whether one is a responsible agent. Even if a moral animal could understand some part of this, their comprehension would certainly fall short of what is required for their punishment to constitute a genuinely reciprocal and rational engagement. Thus, for the communicative retributivist, animal punishment cannot be justified, and thus human punishers need not follow from animal moral agency.

This is a promising way forward for the retributivist, but, depending on one’s views, may be bittersweet. As Joel Feinberg eloquently puts it, for many retributivists “the ultimate justifying purpose of punishment is to match off moral gravity and pain, to give each offender exactly that amount of pain the evil of his offense calls for, on the principle of justice that the wicked should suffer pain in exact proportion to their turpitude” (Reference Feinberg1965). A retributivist who concurs with this sentiment may find it uncomfortable or at least unfortunate that there should exist a great many wrongdoers whose turpitude must be left unanswered in virtue of their failures to satisfy the cognitive preconditions for communicative engagement.

8. Summing up

If the literature concerning the moral animal is to develop, we must increasingly discuss not only whether some animals are moral but also whether they are moral agents, and what it would mean for human beings if they were.

I hope to have shown that a commitment to the existence of animal moral agents comes at a cost for the retributivist about punishment. The retributivist who accepts that some animals are moral agents but prefers not to take up the mantle of punishing animal wrongdoers must, if they have not already, adopt a version of retributivism that allows them to avoid it, such as communicative retributivism. Alternately, perhaps the retributivist will take up the mantle, in which case they might suffer some minor derision, but will at least not need to reconcile themselves to the existence of blameworthy agents whom they cannot in principle punish—only ones whom they cannot (as yet) practically punish.