1. Introduction: Accountability without Equality

It is a regrettable, if understandable, feature of discourse about animal rights, in both ethics and politics, that it frequently makes use of concepts better restricted to the domain of interpersonal relationships. Examples include the pioneering work of Peter Singer and Tom Regan. The former asserts equality of moral status among all sentient beings (^{Singer 1975}). The latter asserts equal status at least among humans and a wide range of animals, including most or all mammals (^{Regan 1983}). The basic moral entitlements of human persons include, in normal circumstances, both consensuality and social equality. As common sense tells us-and as is elaborated below-animals incapable of reflective self-criticism are incapable of consent. It is also likely that we are not capable of social equality with them. It is not a defect of our relationships with our pets, for example, that we rather than they are empowered to decide where to go on vacation, when to move into a new apartment, or whether to visit the doctor. It strains the idea of equality to assert equal status in the face of these facts, and it strains credibility to deny these are facts of our moral landscape.

The same point holds of more recent discussions of animals’ political status. Here the most prominent work is that of Sue Donaldson and Will Kymlicka, who use the concept democratic citizenship to organize proposals about animal rights (^{Donaldson and Kymlicka 2011}, ²⁰¹⁴). Though there is much of value in their writings that can be recovered in an alternative framework, one aim of this essay is to explain why these overarching frames distort their subject matter.

My primary purpose, however, is not a negative one. I do not discuss at length which views espoused by the above authors should be discarded, and which preserved. Nor do I even argue directly that their frames are misguided. It is an input to the present investigation, not a result of it, that (i) there are inequalities of moral status among animals and (ii) no animal who is not a person should be classified as a democratic citizen.¹ By an “input” I mean that I take these views to have default status due to their grounding in common understanding. Accordingly I think it worthwhile to explore these views, and to attempt to explain them, on the working supposition that they are true. That said, those explorations should be open to the possibility that common explanations of these claims may be in error. We should furthermore acknowledge the fallibility of these claims, remaining open to the possibility that careful reflection might undermine them. Meanwhile the burden of argument, however, lies with those who deny these claims.

Below I articulate positively both what I take to be a state-of-the-art understanding of animal psychological capacities and a morally sensitive understanding, grounded in part in actual social practice, of what those capacities entitle animals to. It will emerge that many phenomena theorists discuss almost exclusively in interpersonal contexts also occur in relationships between human persons and non-human animals; these include responsibility, punishment, instruction, assent, and decency. This does not entail attributing personhood to any non-human animal. Attention to the status of non-persons illuminates relatively neglected topics, moreover, such as non-moral responsibility, non-legal punishment, non-educational instruction, non-consensual willingness, and decency within unequal social hierarchies. I do not treat all these neglected phenomena at length here. I mention them to underscore the significance of this project if it is on the right track.

I begin discussion of these topics in a perhaps surprising area of the philosophical literature: the theory of perception.

2. Reason without Reflection

Tyler Burge’s Origins of Objectivity may be the most important philosophical work published thus far this century. This is in part because Burge has articulated the most authoritative account of perception yet produced, but the text’s significance extends beyond that.² It exemplifies a new philosophical method, provides a synopsis of the past century of theorizing in the philosophy of mind, and enunciates fertile approaches to a range of topics in the philosophy of mind distinct from perception.

I will not attempt even to summarize Burge’s book here. But I will note that Burge articulates a devastating criticism of the accounts of perception propounded by Peter Strawson and Gareth Evans (^{Strawson 1959}; ^{Evans 1982}; ^{Burge 2010, pp. 154-210}). As Burge explains in detail, these theorists overintellectualize perception, failing to give sufficient attention to the perceptions of animals and infants. This leaves them vulnerable to refutation by the commonsense position that these individuals perceive even as they lack abilities to think about perceptions. It also leaves them vulnerable to more conclusive refutation by empirical work confirming this commonsense understanding, rendering it more precise and explanatorily rich.

Burge notes, but does not discuss at length, the fact that an analogous critique looms for any account of beliefs (or other propositional attitudes) that requires the presence of thoughts about beliefs (or other propositional attitudes). Here too common sense and scientific practices permit the possibility. It is extraordinarily likely that animals such as apes, whales, and elephants have beliefs, even though it is not known-indeed it appears unlikely-that these animals think about reasons or reflect on their own beliefs (^{Burge 1999}, ²⁰¹⁸). Similarly, though this is an ongoing matter of empirical and philosophical investigation, the range of animals who possess beliefs may be quite wide, even if the range of animals able to assess or even to think about their thoughts is exceedingly narrow (^{Varner 2012}).

Relatedly, empirical study of which animals have beliefs or other thoughts does not proceed by investigating whether these animals reflect or criticize. Rather, it proceeds-and appropriately so-by investigating whether they infer (^{Allen 2006}; ^{Hurley 2006}). This too suggests a capacity for belief or thought does not presuppose any capacity to evaluate, or even to think about, beliefs. As I briefly explain below, moreover, this view is independently suggested by features of many actual social practices we have with animals.

None of these observations undermines, however, the common view that beliefs and other propositional thoughts are reasons-responsive psychological states. Accounts of belief typically presuppose, and rightly so, that beliefs of their nature can be formed by rational inferences and can serve as a basis for rational inferences.³ It is thus highly plausible, and an appropriate default for discussion, to think that a capacity for belief entails a capacity for reasons-response. Similar remarks apply for all other propositional attitudes.

Belief and other propositional attitudes are thus importantly distinct from perception in that they constitutively involve a response to logical or rational relations, as contrasted with response merely to causal and functional relations (^{Burge 2010, pp. 532-551}). There is no good reason to think individuals must think about reasons in order to have thoughts, however, any more than there is reason to think representing objects perceptually requires that individuals also represent perceptions or the success conditions of perceptual states. The burden of argument is on those who maintain that belief requires metarepresentation, and it is extremely doubtful that burden will be met.⁴

This discussion indicates there is an important ambiguity in the idea of a rational being or a rational agent, a distinction too infrequently noted. This distinction is between those beings or agents who have a capacity for thought and so respond to reasons, and beings or agents who have a capacity to think about reasons and to explicitly evaluate their thoughts (as well as thoughts of others). The latter category entails capacities to reflect on and to criticize thoughts such as beliefs and intentions. This brings with it the ability to think explicitly about justifications, in deliberation among other contexts. The former capacity does not involve these abilities. It involves only the ability to functionally form psychological states in response to relations of rational support.

While we do not know with certainty the extensions of either of these categories, we do know that on Earth the former extends much more broadly than the latter within the animal kingdom. Burge refers to those in the former category as “higher animals” and to those in the latter category as beings with “critical reason” (^{Burge 2013}). I follow these uses, though I often refer to animals as “thinking” or as “non-critically rational” rather than as “higher”.

Considerable work, philosophical and empirical, remains to be done on the nature of this non-critical reason. Nevertheless, I think it is important to begin to develop an account of the moral and social statuses that flow from this capacity. I think we are in a position to know thinking animals have non-moral responsibilities, are appropriate targets of non-moral reactive attitudes, and are entitled to certain forms of recognition.⁵ (Though not a focus of discussion here, I also think non-thinking animals, including some conscious animals entitled to our concern, lack responsibilities and lack entitlements to recognition.) As I explain below, however, non-critically rational beings nevertheless lack all moral and legal responsibilities. These individuals are not entitled to social equality with human persons. Nor are they entitled to mutually consensual relationships.

Previous authors, notably including Mark Rowlands and Susana Monsó, have noted both the existence of reasons-response among non-human animals and the significance of this fact for how we should relate to animals (^{Rowlands 2012}; ^{Monsó 2017}). As I discuss in what follows, I do not think these authors provide satisfactory accounts of what reasons-response is or of how reasons-response is related to morality. But before indicating what differentiates the present investigation from their work, I think it important first to acknowledge the value of their contributions. Monsó and Rowlands make progress toward correcting the over-intellectualized accounts of acting for reasons developed in the writings of influential authors such as Donald Davidson and Christine Korsgaard (^{Davidson 1982}; ^{Korsgaard 2005}).

Another strength of work by Rowlands and Monsó lies in their recognition of the significance of non-reflective reasons-response without claiming there are no important ethical distinctions between reflective and merely thinking animals. Rowlands distinguishes “moral subjecthood” from “moral agency”, where the former is the ability to act for “moral reasons” but the latter is being responsible for how well one responds to these reasons (^{Rowlands 2012}, ²⁰¹⁷; ^{Delon 2024}). Along similar lines Monsó distinguishes “minimal moral emotions” from the more robust capacities she holds to be necessary for moral judgment and moral responsibility (^{Monsó 2017, pp. 350-353}).

I believe these authors fail to identify the most important representational capacity between perception and reflective critical reasoning, however, and accordingly that this important work can be improved upon. Rowlands and Monsó emphasize emotional evaluations, but both emotion and evaluation occur below the level of thought or reasons-response. The most basic evaluations made by animals are representations of success and failure. Many non-thinking animals represent the difference, for example, between another animal tripping and its being blown over by a wind. This involves distinguishing failures of agency from external impediments to agency. If an animal possesses emotions, these too may represent evaluatively. All of this occurs in animals who do not think and who do not act in response to reasons (^{Burge 2018, pp. 412-414}).

What primitive evaluations do involve is representation of agency, attribution of action. This may sound sophisticated, as though it involves or borders on metarepresentation, since agency itself is commonly over-intellectualized. As Burge has convincingly explained, however, the most basic type of agency is not constitutively psychological; and consequently the representation of agency need not involve attribution of psychological states to oneself or to others (^{Burge 2009b}, ²⁰¹⁸). The most basic type of agency is functional whole-entity movement. This capacity is what intuitively distinguishes animals from plants. It occurs in some macroscopic organisms, such as jellyfish, who lack psychologies. This primitive type of agency also occurs in some, although not all, microorganisms (^{van Houten 2000}). Representing action involves representing purposes or goals non-mentalistically, and it is crucial in the lives of many animals. But this representation of action can occur without any of the marks of thought or reasons-response. In particular it occurs in animals without comprehension of the world, without propositional attitudes, and without the ability to infer.

It is not entirely clear how broad the ambit of moral subjecthood is in the views of Rowlands and Monsó, since-rather than attributing these traditional markers of thought-they speak of “reliable mechanisms” for “tracking” facts of putative moral importance (^{Rowlands 2012, p. 57}; ^{Monsó 2017, p. 346}). If they mean to restrict moral subjecthood to thinking animals, then this essay provides resources for formulating their views more perspicuously, even as it dissents from their accounts of morality. Such a restriction is suggested by the prominence, in their discussions, of the most psychologically sophisticated non-human animals: apes, elephants, and cetaceans (^{Monsó 2017, p. 339}).

If they do not mean to restrict moral subjecthood to thinking animals, this essay exposes their approaches as even more radical than they initially appear. They are cognizant of their departure from certain traditional views, including in particular the view that metarepresentation is required for liability to moral norms; emotional response to distress behavior suffices, on to their views, there is no need to represent others as distressed psychologically (^{Monsó 2017, pp. 347-353}). But relative to thought, representation of distress behavior-including in evaluative representational contents, such as emotions-occurs much more widely. If we do not restrict moral subjecthood to thinking animals, we extend the domain of morality far more broadly than is suggested by their usual discussion examples. This in turn suggests something has gone awry in the account, that the capacities on which they focus are too primitive to be plausible candidates even for a revisionary account of moral subjecthood.

Again, my main purpose is not to argue against these views; that would require a much fuller exposition of the views and a more extensive comparison of them with the approach articulated here. My purpose is positive, to sketch that new approach. The approach draws heavily on what I see as the most sophisticated available philosophical interpretation of empirical work on animal psychology, that of Burge. Burge does not say everything I say here, of course, since his topic is not ethics. Nor do I claim he would endorse everything I say. But this essay seeks to remedy a major shortcoming of the animal ethics literature, namely, underuse of Burge’s unparalleled work in the philosophy of mind. In particular I attend to Burge’s careful defense and development of his views that agency is possible without psychology, that perception is possible without thought, and that thought is possible without critical reflection. This may be of service even to those who remain unconvinced this essay’s approach is superior to that of Rowlands or Monsó. Philosophers attracted to their views, or to other views articulated in the wake of their writings, may wish to incorporate lessons from Burge into their own approaches.

3. Responsibility without Morality

In the opening section I announced my view that many non-human animals have non-moral responsibilities; in this section, I develop and motivate that view. I begin by observing that we speak non-metaphorically of punishing animals who are strong candidates to be thinking animals, but not other types of animals (^{Garthoff 2020b}). Although we might not outright reject talk of “punishing” a pet fish or spider by withholding food or sequestering it from its conspecifics, such talk is likely extended or partly metaphorical. Talk of punishing a dog for jumping on people or punishing a horse for its recalcitrant behavior contrasts with this. If a dog is denied access to the house because it defecated indoors, this is a punishment in a straightforward sense. Analogous points hold for one- and two-year-old children. All of these individuals may be punished fittingly even though they lack developed capacities to evaluate, and indeed even to reflect upon, their intentions and other thoughts.

Note next that accounts of punishment typically hold-in my view correctly-that an act of punishment is appropriate only when the individual punished is responsible for their behavior (^{Greenawalt 1983}). This highly plausible claim has grounds in common sense, in reflection on actual social practices, and in reflection on the morality of punishment. It is only infrequently observed, however, that paradigm cases of punishment involve young children and non-human animals. Much more typical as a context for discussion are the institutions of criminal justice, which do not apply to very young children or to animals.⁶

This feature of philosophical discussions of punishment extends, moreover, to the literature on responsibility. This literature commonly discusses adults with severe mental impairments, but until recently it rarely considered animals or the extremely young. Indeed, even when non-moral responsibility is discussed-which is itself rare-the context typically includes only adults (^{Wolf 2015}). A focus on thinking animals better distills the phenomenon of non-moral responsibility.

The distinction I focus on here, it bears emphasizing, is not the distinction between moral and non-moral responsibility in human persons. Some aspects of non-moral responsibility in adults rely on the fact that adults have developed capacities for critical reasoning. If I am part of a team of engineers demolishing a bridge, for example, then I am responsible for using critical reason in certain ways. Suppose my role is to research non-visible constituents of the bridge, since the type of explosives used during demolition turns on whether the bridge contains hazardous materials. Suppose further that with competence and exercising due diligence I come to the conclusion that the bridge contains no asbestos, when in fact it does contain asbestos; and that as a consequence the team uses an explosive that spreads asbestos through the region in a dangerous and damaging way. Suppose also that this is because those who built the bridge later falsified records to avoid liability, and that this espionage was conducted so expertly that I am not culpable for the mistake in my analysis.

In this case I am not-by stipulation-morally responsible for what happened. I was cautious and careful in my research. Nevertheless I am non-morally responsible for what happened, since the source of the team’s error lay in the part of the job that was my responsibility. This non-moral responsibility is one that only a critically reasoning individual-a person-could have. Thinking animals who cannot reflect on their thoughts critically cannot conduct research. These individuals accordingly have no responsibilities of this kind.

This is not the phenomenon I am interested in. I am interested in non-moral responsibilities an individual may have even if they lack capacities for critical reasoning altogether. Suppose I am part of a security patrol at an airport searching for explosives. Suppose further that this security team includes non-human animals, namely dogs trained to smell chemical traces characteristic of explosives. Suppose next that, during an exercise, the team fails to detect an explosive because the dog fails to communicate the presence of these chemicals even though it smelled them. In this case the dog fails in its responsibilities as a member of the team. As in the previous case, this is not a moral failure. But here the individual’s non-moral responsibility does not require that they possess critical reason.

Importantly, the dog’s responsibility in this case is not merely causal or functional. The claim is not merely that the cause of the failure lay in the dog’s behavior rather than in mine. Nor is the phenomenon of interest merely that it is in the dog’s role that malfunctions occurred, rather than in mine. These are not the only senses in which the dog is responsible. Nor even is the claim merely that a failure occurred in the dog’s psychological functioning. These three claims are all true, as I intend the scenario to be understood. But none captures the phenomenon of interest. The failure of the dog need not merely be a deviation from a standard or failed behavioral conditioning. The failure need not consist merely in the fact that the dog’s inclinations proved unsuitable given its role as a member of the security team.

The pertinent failure, as I intend the case to be filled out, is the dog’s failure to respond to a reason. Given that the dog detected the chemicals, it had a reason to communicate this to me, if we stipulate the dog is a thinking animal properly trained to be sensitive to such considerations in its practical inferences, its thought-guided decisions about what to do. Under those stipulations, this is a failure of reasons-response. Failures of this kind are always psychological and functional failures. But it is due to the dog’s more specific failure to respond well to practical reasons that it is responsible in a way such that rebuke or punishment may be an appropriate response. This is what makes rebuking or punishing the dog’s behavior a way to hold the animal accountable.

These ideas of accountability and punishment go beyond ideas of deviation, inclination, and conditioning; but they are not specifically moral notions, like thinking the dog acted wrongly or attributing to it a guilty conscience. We should not respond to the dog with moralized attitudes, as though it were responsible for acting justifiably. The dog’s training enables it to learn to respond to certain reasons when deciding what to do. This training does not enable the dog to learn that it failed to respond well to reasons. Nothing could enable the dog to learn that, since the dog lacks abilities to attribute thoughts to itself and to represent reasons as such.

The case reveals a domain of practical concern intermediate in psychological sophistication between that of deviation, inclination, and conditioning and that of wrongdoing, conscience, and education.⁷ Concepts which refer to this domain include transgression, (dis)satisfaction, and instruction (^{Garthoff 2020b}, ²⁰²⁶). Transgression involves more than mere deviation from a standard, for it involves failure to respond to a reason. Transgression does not, however, entail specifically moral wrongdoing. The mental states of satisfaction and dissatisfaction are taken in states of affairs, as rational responses to states of affairs. They are not merely causal or functional responses to states of affairs. These attitudes do not, however, entail a conscience. Instruction involves more than mere conditioning; it involves learning to respond well to reasons. It develops functional sensitivities in thought-mediated action, not only in action mediated by emotions or other representations that can occur below the level of comprehension and inference. But rational learning of this type does not suffice for education; instruction can be achieved without using the reflective and critical capacities characteristic of persons.

Since responsibility is keyed to failure in reasons-response and thought is the most primitive rational capacity, only thinking animals have non-moral responsibilities of this type. Thus if frogs lack beliefs and other reasons-responsive states, then they have no genuine responsibilities.⁸ It follows that frogs are not dissatisfied by anything, cannot transgress, and cannot be instructed. By contrast, elephants are dissatisfied by things, can transgress, and can be instructed.⁹ That we have practices of instruction for numerous domesticated animals, such as dogs and horses, is evidence that those animals too have beliefs and other reasons-responsive mental states. Careful reflection on these concepts reveals they apply to dogs and horses only if these animals have responsibilities and are capable of responding to reasons. This evidence is not on its own conclusive; empirical investigation is required. But insights from reflection on these practices ought to set the defaults and working hypotheses that frame empirical work on the issue.¹⁰

Regardless of which animals think, this motivation for claiming that non-human animals have non-moral responsibilities is not anthropomorphic. The account offered here is scrupulous in its insistence that no specifically human traits-including especially status as a person or as a critical reasoner-ground the type of responsibility under discussion. It is because thinking animals are non-critically rational agents that they respond to reasons and so are in that way accountable. To be accountable, to have (non-moral and non-legal) responsibilities, they need not ask or answer questions about justification; nor need they provide explanations of how to live.

The above example may be difficult to interpret, since even professional philosophers are not accustomed to thinking in terms of types of psychological representation intermediate between perception and reflective thought. Thus it is worth connecting interpretation of that example to a topic that has been fruitfully taken up in the literature on animal ethics: the fact that many-but certainly not all-non-human animals are appropriate targets of reactive attitudes like praise and blame. This important observation is sometimes presented, in my view mistakenly, as evidence that many non-human animals are morally responsible for their actions (^{Cova 2013}; ^{Ferrin 2019}; ^{Borchert and Rumana 2023}).

It is surely correct that many non-human animals are appropriate targets for praise and blame. A strong default in favor of the position is provided by ordinary practice-including practices of training pets-and many such practices survive reflective scrutiny (^{Garthoff 2020b}, ²⁰²⁶). Those authors who have called attention to this fact in the literature have accordingly done an important service. It is also insightful that they connect an animal’s liability to being targeted by reactive attitudes to its capacity to respond to reasons when deciding what to do (^{Cova 2013}; ^{Ferrin 2019}; ^{Borchert and Rumana 2023}).

In so doing, we must take care to distinguish thought or reasons-response as such both from less sophisticated and from more sophisticated representational capacities. Part of the point of the reactive attitudes (also punishment) is to enable those targeted by them to learn. But the learning these attitudes enable is crucially distinct from operant conditioning. Conditioning is a mode of learning that does not involve responsibility in its target as a success condition (^{Garthoff 2020b}). The reactive attitudes do not function, accordingly, to condition; they function to instruct. If the training of dogs and horses (such as in the above example) were exhausted by conditioning, that training would support neither thinking these animals are appropriately targeted by reactive attitudes nor thinking they are capable of reasons-response. A frog can be conditioned; but a frog cannot be instructed. This is because frogs cannot respond to reasons, and it is why we are correct in lacking practices that direct reactive attitudes at frogs.

On the other hand, we need to be careful not to deflate modes of learning that extend beyond instruction to involve education, where this constitutively involves appreciation of explanations and justifications (^{Garthoff 2020b}). Excuse-like or apology-like interactions among animals are not excuses and apologies, even if these animals respond to reasons. These interactions do not function to signal mutual cognizance of rational norms, for animals lacking metarepresentation cannot cognize those norms as such. This is also why non-reflective animals wholly lack legal responsibilities, even when they possess responsibilities of another kind (^{Garthoff 2019}).

This distinction explains, furthermore, why the generally non-reflective character of human action is compatible with the traditional view that moral norms apply to human persons in virtue of their capacity for reflection.¹¹ Those who attribute thoughts to themselves and evaluate those thoughts-those whose reason is reflective and critical-are capable of a distinctive type of self-regulation. The tradition claims this self-regulation partly constitutes morality. That is why the tradition focuses moral assessment on explicitly represented maxims of action or on reflectively trained traits of character. Less traditional views allege instead that moral assessment is wholly grounded in non-reflective states such as intentions, emotions, behaviors, and outcomes. Disputes about whether to characterize the psychology of thinking animals as moral recapitulate these different approaches to ethical theory and moral psychology. These disputes cannot be litigated fully in this essay. The point is to show how wisdom contained in views that are deflationary relative to the tradition can be captured without deflation; and that fact may be of interest to all parties involved in the conversation.

Thus I think it best not to characterize the type of responsibility possessed by non-reflective animals as “moral” responsibility. I do not think the ordinary term “moral” includes this type in its reference; where not mistaken, it is an extended or metaphorical use. Nor do I think we should expand what the term refers to, since we would then need a different term to refer to the type of responsibility characteristic of reflective beings.¹² But terminology is not paramount. The concept moral is vague; including borderline or extended cases is perhaps not unreasonable. The more important point is that if we use “morality” and “moral responsibility” in this way, we must take care not to deflate the distinctive domain of moral and social relations among reflective beings.

It should be noted that neither the distinction between reflection and thought nor that between thought and more primitive capacities corresponds to the distinction between accountability and attributability (^{Watson 1996}; ^{Delon 2024}). Virtues can be attributed at every level, relative to that level’s characteristic function: action, perception, emotion, thought, reflection. The difference between reflective and non-reflective thought marks a distinction, moreover, between two types of accountability: mediated by mutual metarepresentational awareness of norms of justification and not mediated by any such awareness of rational norms.

Once we mark these distinctions among tiers of representational sophistication with care, we can see the astute observations that animals respond to reasons and that animals are appropriate targets of reactive attitudes are both compatible with the traditional claim that moral evaluability is the exclusive domain of critically reflective thinking beings.

4. Cooperation without Personhood

In section 2, I elucidated the idea of non-critical reason and motivated the position that many non-human animals are rational agents in that they respond to reasons but not in the sense that they are able to think about reasons. In section 3, I related these psychological claims to practices of punishment and responsibility. In this and the next section, I extend the ambit of discussion to include the social status of thinking animals.

The first thing to register is that we do not now have well-developed social or legal structures that properly reflect and constitute good relations with thinking animals. This is partly due to the sorry treatment of animals by industry, which necessitates attention to the most basic protections rather than to a richer understanding of appropriate social relations. Current legal practice at best captures that many animals are worthy of concern, that complete indifference to the interests and status of these animals is to be condemned.

Efforts to expand legal protection for animals sometimes take the form of advocating for the legal personhood of animals (^{Francione 2008}). I am not sufficiently steeped in law or activism to know whether such a tactic is pragmatically warranted. As I have elsewhere explained, however, I do not think that this approach most perspicuously accounts for the moral status of animals or the social relationships we should realize with them (^{Garthoff 2010}, ²⁰¹⁹, ^2020a, ²⁰²³, ²⁰²⁶). I will not attempt here to rehearse the rationale for that position, but the material presented above already puts us in a position to appreciate its crux: in both ordinary and technical contexts, critical reason is the capacity that most crucially grounds personhood. Merely thinking animals lack this capacity, so we either speak inaccurately or risk confusion when we refer to their moral, social, or legal status as personhood.¹³ The central notion of personhood is liability to obligation and more generally to norms of justifiability, including moral and legal justifiability (^{Garthoff 2019}, ²⁰²³, ²⁰²⁶). We sacrifice too much explanatory power if we use “person” to refer to all animals who respond to reasons, are targets of reactive attitudes, or have responsibilities of any kind.

Against this, I think innovation is needed, and not only due to the inadequacy of current legal forms to capture the status of animals. Fetuses are not well treated by current legal categories either.¹⁴ Ethical theory increasingly appreciates this point; it is common now to attribute moral status to all conscious beings. This status is at times indicated by saying such beings possess “inherent value” or that these individuals “merit concern” (^{Regan 1983}; ^{Anderson 1993}; ^{Darwall 2002}; ^{Garthoff 2010}). This is often distinguished from the “dignity” of human persons. Legal practice has long included mechanisms (trusteeship, notably) for representing interests of those incapable of representing themselves, and these legal practices have expanded alongside the (painfully slowly) growing concern for the status of non-human animals in law.

What is far less frequently observed, however, is the need for a further distinction beyond this.¹⁵ Conscious lower animals merit concern and possess an entitlement against our exhibiting complete indifference to them. Thinking animals have a distinct moral status, one grounded not only in their consciousness or sentience but also in their capacity for non-critical reason (^{Garthoff 2023}, ²⁰²⁶). We need social and legal forms that reflect this.

As the discussion of the previous section indicates, we already have fragments of the needed social forms. Our practices of instructing higher animals, and under appropriate circumstances punishing them or directing praise or blame toward them, indicate features these broader social forms must include. We should build out from these practices in a morally sensitive way as we work to better understand social relations with these animals.

Even though the greater danger by far in actual practice is social forms underestimating the status of thinking animals, it is also possible to overstate their status. They are not entitled to social equality with humans. Social subordination of non-reflective animals to human persons need not constitute a defect in how we relate to them. Equality is the hallmark of dignity, the moral status of human persons or reflective, critically rational agents. It is not the mark of moral status in general, nor of the status of non-critically rational agents in particular.

Why not? We can begin to explain this by pointing to what has already been noted: thinking animals have no moral or legal responsibilities, since these presuppose a capacity to understand justifications. Critical thinkers vote, hold offices, participate in lotteries, and contribute directly to deliberations about what to do. Non-critical thinkers have vastly more limited abilities to do these things. Indeed to the extent they do them at all, this is parasitic upon the activity of persons. It is likely not even metaphysically possible to realize social forms where human persons are equal to non-critically rational animals. In any case this is neither required nor desirable.

Consider further that non-critically rational agents are incapable of consent, since consent presupposes permission and permission is intelligible only when thinking about justifications. Mutual consent is often constitutive of appropriate social relations among persons (^{O’Neill 1990}; ^{Pallikkathayil 2011}). This is not the case for thinking animals, nor between human persons and thinking animals. If they are of sound mind, we cannot take our human friends to the hospital against their will. We can justifiably take our animal friends to the hospital against their will.¹⁶

The claim that higher animals are appropriately subordinated to human persons is of course liable to misunderstanding and misapplication. It does not imply that human interests always take precedence over animal interests whenever the two seriously conflict. Nor does it imply there is no such thing as wrongly exploiting or dominating animals. The fact of justifiable subordination is not mysterious, and it is not a discovery. Families and elementary schools, among many other institutions, have and should have moral constraints on their operation grounded in the status of their participants even though the participants are not social equals.¹⁷ There are interesting and difficult questions about the boundaries of this, especially in the case of adults. While their actual instantiations tend to be oppressive, I take no position here about whether military hierarchies or boss-worker relations are as such incompatible with broader social equality. The present point is only that at least some analogs to these sorts of structures, ones that place human persons in the superior position and thinking animals in the inferior position, are justifiable. Neither inequality nor hierarchy is, as such, inimical to fully cooperative relationships. In interpersonal relations hierarchies and other social inequalities must be situated within a broader environment of equality and consensuality, if they are to be fully cooperative. But in social relations with non-reflective thinking animals, no such backdrop is needed for cooperative relations.¹⁸

That reflective animals are often entitled to make judgments that override the judgments of non-reflective thinking animals does not imply that persons have a default entitlement to impose their will on them. Though non-reflective thinkers cannot consent, there is a difference between what they do willingly and what they do unwillingly. Their type of willingness has been aptly labeled “assent” (^{Healey and Pepper 2021}). Human persons should often, when exercising the superior social status they appropriately possess, attend carefully to what animals do and do not assent to. This often includes accepting their judgment about what to do, especially regarding their own treatment, while retaining authority to override them non-consensually when warranted.

Thus incorporating limitations to the claim that thinking animals are fittingly subordinated to human persons does not prevent that claim from having substance. Humans often substitute their judgment for that of animals, as in the hospital case above, and are correct to do so. Humans often appropriately control major decisions of social groups they share with animals. In circumstances of triage human persons are often fittingly given preferential treatment vis-à-vis the non-reflective animals, moreover, including the thinking animals.¹⁹

That the core notion of personhood refers to the capacity of reflective criticism is compatible with extended uses of the notion to refer also to those lacking such capacities. A being who now lacks reflective capacities might nevertheless be a person due to their past capacities, their future capacities, the capacities characteristic of their kind, or their ability to occupy social positions normally occupied by reflective beings. There are limits to this; the caretaker-child relation is not identical to the caretaker-puppy relation. Even the ability of a human to occupy a social position may be unclear, partial, or limited. It is extraordinarily difficult to render the distinction between persons and non-persons precise in all borderline or extended cases.

But this does not undermine the main claim of this section. Hard cases make bad law, and the difficulty of specifying a distinction precisely does not undermine its significance. Nor does this difficulty undermine default understandings, supported both by ordinary practices and by careful reflection, about what personhood’s core consists in (^{Burge 1999}, ²⁰¹³; ^{Kitcher 2005}). In fact, careful reflection about hard cases deploys this same understanding: cases are hard because they are borderline or extended cases of those who possess the relevant reflective capacities.²⁰

5. Recognition without Dignity

In the previous section, I emphasized that thought is not personhood, and accordingly that the marks of appropriate interpersonal relations-social equality, consensuality-do not characterize appropriate relations between human persons and thinking animals. Can we say anything more positively about appropriate relationships with thinking animals? Unlike mere pain and pleasure, or mere aversion and attraction, reasons-response involves understanding. This is not theoretical understanding, which requires appreciating explanations, and so in turn a capacity for critical and reflective reason. It is a non-theoretical, non-explanatory mode of understanding. One example is the comprehension involved in having a thought, which any reasoning animal deploys even when they cannot represent their own reasoning. Another is what I will call “primitive interpretation”. A non-rational animal may see an object as food or a creature as dangerous, but such mere classification or attribution is not interpretation. A non-critically rational animal can not only see-as but also think-as. A stick may be thought of as a tool, or an instruction as to-be-followed, so that the stick or instruction is poised for use in practical reasoning. This is not merely attributing features to objects or categorizing objects by needs. It is a rudimentary understanding of the world, for it is mediated by response to the logical implications of mental states, even when these relations are not represented as such.

The fact that some animals have this mode of understanding is reflected in some actual social practices. Even if one attends to a dog’s other needs with devotion, ensuring the animal’s comfort and health and stimulation, it would be ghastly to live with that dog yet never address oneself to it. We speak to thinking animals in ordinary life, and we concern ourselves with their interpretations of what we do. This is especially prominent when training thinking animals. Few things are more natural than saying “good dog” or “bad dog” to a dog (^{Garthoff 2020b}).

This contrasts with our dealings with lower animals, even in the context of conditioning and training. We might speak to a frog, even saying “good frog” or “bad frog” at times analogous to when we would utter these phrases to a dog. But in such cases we are speaking to ourselves. We do not expect the frog to understand these utterances at all. By contrast, a dog, even though the animal is non-linguistic, possesses some understanding of the significance of the locution.

I think careful reflection on this difference reveals that a major constituent of the moral status of thinking animals is an entitlement to be treated intelligibly (^{Garthoff 2023}). We must concern ourselves not only with their interests-that is true of lower animals, if they are conscious-but also with how the animal understands its treatment. As we innovate social forms with thinking animals, accordingly, we must understand mutual intelligibility and responsibility as among their most central features.

I suggest we label this broader status, one encompassing both a capacity to act responsibly and a capacity for understanding, as recognitionworthiness. Recognition is typically theorized alongside human dignity in moral and political theory (^{Honneth 1996}; ^{Scanlon 1998}; ^{Darwall 2006}). But in fact it has broader application. We need communities of recognition with thinking animals that involve more than a responsiveness to their interests. These communities should both attribute a respect-like moral status of recognitionworthiness to these animals and constitute a parallel social status (^{Garthoff 2023}).

This status is neither mere concernworthiness nor the dignity of persons. It does not demand thinking animals author the trajectories of their lives, nor that they consent to their treatment. It does not call for them to be social equals with human beings. That said, the types of subordination to which thinking animals are legitimately subjected are far less dominating than those they are subjected to by societies today. Thinking animals can be involved in decision-making even when they are non-linguistic and so cannot contribute directly to deliberations about what to do. They can be involved because the people participating in deliberations can be sensitive to how thinking animals understand their treatment, including especially how they understand their position in a social system that is hierarchical but mutually beneficial. Potential for mistreatment of thinking animals is accordingly not fully captured by policies of anti-cruelty or non-indifference. At the same time, the mistreatment of thinking animals-god-awful though in fact it often is-is neither well captured by analogizing animal work to slavery nor well criticized by asserting all animals are equal.²¹

There is a tripartite distinction among virtues which corresponds to this tripartite distinction among moral statuses. These virtues characterize both individual persons who respond well to the status in question and social systems that do this (^{Garthoff 2026}). In each application, comparison and contrast among these virtues illuminates animals’ social status-especially the intermediate status of thinking but non-reflective animals.²²

A humane individual or social system gathers information from non-thinking animals that is relevant to satisfying their needs, and the individual or social system communicates information to them to meet those needs. The aim is both to meet those needs and to secure their willingness to go along with their treatment rather than to struggle against it. But there is no need for them to understand their needs as having been met.

To be decent, an individual or society must furthermore attend to the understandings of the thinking animals and must communicate with them so that they may understand this attention has been paid. The aim is to secure their assent to the social system, their willingness to do their part in that system out of an understanding that participating in that system is what is to be done. This is not consent, for we do not require their permission to treat them as we do. But it is more than securing the animal’s mere willingness; it is securing a type of willingness which is mediated by thought or understanding.²³

A just individual or society must furthermore place people in position to appreciate their treatment by the social system as justified. The aim is to secure their consent to that system, their willingness to go along with that system from an understanding that participating in the system is justified. This involves commitment to a fundamental social equality among all citizens, but does not involve fundamental equality among all members of society if animals are counted among the societal membership.

Even when they are not aspirationally democratic, the internal social and political ideals of all human societies call for using critical reason in the process through which higher classes consult lower classes. In view of this, and even though non-democratic societies are a crucial resource to draw on when theorizing animal social status, I doubt the status of non-reflective animals is well characterized as any type of citizenship, democratic or otherwise. It unhelpfully muddles internal understandings of aspirationally democratic societies to include unequal citizenship; better is to limit citizenship to natural persons. Thinking animals living in human societies-domestic, wild, or in between-are better understood as non-citizen residents. As we develop fuller accounts of their social status, furthermore, we must attend to how this status differs both from that of citizens and from that of non-citizen resident persons. A complete conception of these differences must deploy the distinctions in representational sophistication that ground differences in moral status among conscious but non-thinking animals, thinking but non-reflective animals, and animals who reflect critically on themselves in light of norms of justification.

6. Conclusion: Social Relations with Thinking Animals

In this essay I have attempted to make progress in characterizing appropriate social relations with animals, especially thinking animals who lack capacities of reflective critical reasoning. I have grounded the framework sketched here in part in Tyler Burge’s interpretation of empirical work on animal psychology, since it is the most sophisticated understanding available and it is underappreciated in the animal ethics literature.

In so doing I have attempted to lay bare connections among several topics, notably including reasons-response, responsibility, and moral and social status. There are of course limitations to any discussion of these connections within a single article; much more remains to be said. Since these topics all bear on the social status of animals, however, my hope is that addressing them within a unified discussion brings out the appeal of the view that non-critical thought grounds a tier of interconnected social and ethical phenomena.

One way to summarize these reflections is to say they elaborate a conception of accountable relations among unequal beings. These relations are accountable because they involve reasons-response, responsibility, instruction, cooperation, and recognition. Such accountability can occur even in animals lacking language and critical reflection who cannot literally account for-in the sense of explain-themselves. These robust ways that people relate to thinking animals thus need not come with the surprising implication that human persons are equal in status to animals with no reflective capacities. That is a good thing, since it is not socially possible to instantiate equality without including social mechanisms-important examples include laws, offices, elections, and lotteries-that are metaphysically impossible for non-reflective beings to participate in on equal terms with human persons.²⁴