What is it like to be a bat in pain?

Kinds of Animal Minds and the Moral Comparison Principle

The day may come when the rest of the animal creation may acquire those rights which never could have been witholden from them but by the hand of tyranny. The French have already discovered that the blackness of the skin is no reason why a human being should be abandoned without redress to the caprice of a tormentor. It may one day come to be recognized that the number of legs, the villosity of the skin, or the termination of the os sacrum are reasons equally insufficient for abandoning a sensitive being to the same fate. What else is it that should trace the insuperable line? Is it the faculty of reason, or perhaps the faculty of discourse? But a full-grown horse or dog is beyond comparison a more rational, as well as a more conversable animal, than an infant of a day or a week or even a month old. But suppose it were otherwise, what would it avail? The question is not, Can they reason? nor Can they talk? but, Can they suffer?

--Jeremy Bentham

If a being suffers, there can be no moral justification for refusing to take that suffering into accountÖ If a being is not capable of suffering, Ö, there is nothing to be taken into account.

--Peter Singer

We describe bat sonar as a form of three-dimensional forward perception; we believe that bats feel some versions of pain, fear, hunger, and lust, and that they have other, more familiar types of perception besides sonar. But we believe that these experiences also have in each case a specific subjective character, which it is beyond our ability to conceive.

--Thomas Nagel

I. Introduction

The topic of this paper lies within the densely structured but largely unexplored region of overlap between the philosophy of mind and normative ethics. For convenience, I call this region the ethics of mind. What I want to discuss in this paper, in particular, is the moral significance of the phenomenal consciousness or subjective experience of different types of animals, and in particular the comparative moral significance of the subjective experience of pain across different types of animals.

II. The Mental-Mental Gap and its Consequences

Everyone knows that Thomas Nagelís famous essay "What is it like to be a bat?" is all about gaps in the philosophy of mind. It is now commonplace to distinguish carefully between an explanatory gap, and an ontological gap. An explanatory gap obtains just in case one set of concepts cannot be reduced to another set of concepts, whether by analytical definition, or by some weaker kind of reduction such as necessary coextension. By contrast, an ontological gap obtains just in case one set of properties or facts cannot be reduced to another set of properties or facts, whether by identity, or by some weaker kind of reduction such as necessary equivalence or logical (strong) supervenience. An explanatory gap is consistent with, but does not automatically entail, the existence of an ontological gap. That is, one can defend nonreduction for some pair of (sets of) concepts, but still assert reduction for the corresponding pair of (sets of) properties picked out by those concepts. So it is possible to defend the existence of an explanatory gap in the philosophy of mind while denying the existence of an ontological gap. It is also possible to assert the existence of an explanatory gap, and remain agnostic about the existence of an ontological gap. And although this option was not so obvious when "What is it like to be a bat?" first appeared, it is now clear that this is Nagelís thesis. What very few seem to have noticed, however, is that he actually discloses two explanatory gaps in the philosophy of mind, and not just one.

First and foremost, there is Nagelís well-known explanatory gap between mentalistic concepts and physicalistic concepts. Mentalistic concepts are concepts whose content and ascription imply phenomenal consciousness, subjective experience, or the first-person point of view: in Nagelís terms, what it is like to be for an organism. Physicalistic concepts by contrast are concepts whose content and ascription imply only first- or second-order physical properties or facts, the objective character of the natural world, or the third-person/impersonal point of view: what it is for something to be fundamentally physical. Nagelís claim here is that physicalistic or third-person/impersonal concepts can never adequately explain the subjective character of conscious experience. Let us call this "the mental-physical gap."

But second, for Nagel there is also a seemingly equally intractable explanatory gap between the mentalistic concepts that we apply to animals such as ourselves, and the mentalistic concepts that we apply to the conscious states of other kinds of animals. Nagelís claim here is that we are incapable of understanding the specifically subjective character of the conscious experience of other species. As he puts it in another essay:

We ascribe experiences to animals on the basis of their behavior, structure, and circumstances, but we are not just ascribing to them behavior, structure, and circumstances. So what are we saying? The same kind of thing we say of people when we say they have experiences, of course. But here the special relation between first- and third-person ascription is not available as an indication of the subjectivity of the mental. We are left with concepts that are anchored in their application to humans, and that apply to other creatures by a natural extension from the behavioral and contextual criteria that operate in ordinary human cases. This seems definitely unsatisfactory, because the experiences of other creatures are certainly independent of the reach of an analogy with the human case. They have their own reality and their own subjectivity. Let us call this "the mental-mental gap."

Those closely familiar with the structure of "What is it like to be a bat?" will notice that I have reversed Nagelís order of argumentation. He initially discusses the mental-mental gap, and then consequently discusses the mental-physical gap. Officially then, the mental-mental gap is supposed to be the basic premise in the argument for the mental-physical gap:

This [mental-mental gap] bears directly on the mind-body problem. For if the facts about experience--facts about what it is like for the experiencing organism--are accessible from only one point of view, then it is a mystery how the true character of experiences could be revealed in the physical operation of that organism. In other words, for Nagel the mental-mental gap is supposed to entail the mental-physical gap. Now I do not know whether I am the only one who has ever felt a little queasy about this step in Nagelís argument. But in any case it has always seemed to me that although the mental-mental gap is perfectly consistent with the mental-physical gap, nevertheless the two gaps are in fact logically independent of one another.

On the one hand, it is possible to hold that we cannot understand the specifically subjective character of the conscious experience of other animal species and also that reductive materialism is true, so that there is no mental-physical gap. It is true that at least one version of reductive materialism--a hyper-strong version of the type-type mind-brain identity theory that implies both the analytically or logically necessary identity of mentalistic concepts with physicalistic concepts, and also the metaphysically necessary identity of mental properties with physical properties--is sufficient for closing the mental-mental gap: therefore closing the mental-mental gap is necessary for this version of reductive materialism. But since not every version of reductive materialism is of this hyper-strong sort, and since most reductive materialists, including the defenders of classical "type-physicalism," have in fact also explicitly rejected the hyper-strong type-type mind-brain identity theory, it seems obvious that closing the mental-mental gap is not generally necessary for the truth of reductive materialism. Nor, indeed, is closing the mental-mental gap generally necessary for materialism of any sort.

On the other hand and conversely, even if there is a mental-physical gap, it does not necessarily follow that there is a mental-mental gap. Assume the existence of the mental-physical gap. Then assume the truth of some non-Cartesian version of dualism to the effect that some or even most nonhuman animals have immaterial souls that are fundamentally the same as ours. Then further suppose that one fine day, like Dr Dolittle, we somehow learn to "talk to the animals" and are thereby able to read the conscious minds of other species. Or far less implausibly, again assuming the existence of an explanatory gap between mentalistic concepts and physicalistic concepts but this time rejecting dualism, suppose that by some completely non-magical means--say, via the imaginative simulation of non-human animal embodiment, guided by empathy and cognitive ethology (i.e., the scientific study of animal minds and especially non-human animal minds in the context of macro-biology, cognitive psychology, and behavioral psychology)--we come to understand the specifically subjective character of the conscious mental states of some or even most non-humans with the same reliability and conceptual anchorage as our understanding of the subjective character of the conscious mental states of other humans. So there can be a mental-physical gap but no mental-mental gap.

For these reasons, Nagelís mental-mental gap is logically independent of his mental-physical gap. From this point forward, I am going to assume that the mental-mental gap is logically independent of the mental-physical gap, and also that Nagelís argument for the existence of a mental-mental gap is fundamentally sound. The main purpose of this paper is to argue that if there is a mental-mental gap, then we have good reason to believe not only (i) that the pain of human or non-human animals that are persons is more morally significant than the pain of any animals of any species that are non-persons, but also (ii) that the pain of any animals of any species that will become persons in the natural course of their genetic development is more morally significant than the pain of any animals of any species that will never become persons in the natural course of their genetic development. This two-part thesis is what I call the moral comparison principle or the MCP.

If true, the MCP entails we have good reason to believe that the pain of normal humans at any point in the natural course of their genetic development past the achievement of sentience, the pain of other suitably emotionally complex but abnormal humans (including many victims of Alzheimerís disease, autism, cerebral palsy, Downís syndrome, and spina bifida), and the pain of normal apes at any point in the natural course of their genetic development past the achievement of sentience, are all more morally significant than the pain of any bats, cats, dogs, horses, cows, etc., and also more morally significant than the pain of any seriously brain-damaged and merely sentient humans. The overall view of moral status implied by the MCP is what I call "the person-oriented conception."

It should be already obvious that for me, the crucial moral difference between different kinds of animal minds is not the difference between human minds and non-human minds. Rather for me the crucial moral difference is between (1) animals that have merely a capacity for phenomenal consciousness, and (2) animals that in addition to possessing a capacity for phenomenal consciousness, are also persons in the moral sense. Now it is obvious, I think, that every actual or possible animal capable of fully understanding--even if disagreeing with!--this essay is a person in the moral sense, by virtue of possessing psychological capacities for logical reasoning, practical decision-making, recognizing and attributing moral obligations, and reflective self-consciousness. But for reasons I will discuss in section IV, I want to hold that some pre-linguistic humans, some partially linguistic humans, and some partially linguistic or non-linguistic non-human animals are also persons, by virtue of possessing a psychological capacity for the momentary (but not necessarily long term) and self-conscious (but not necessarily reflective) evaluation of their own (but not necessarily othersí) affective states, together with a capacity for intentional animal agency. So I reject speciesism. Speciesism says that some or another species-specific difference between animals is the sole or primary determinant of the moral significance of those animals or moral discriminations between those animals. Speciesism is false and wrong, just as racism and sexism are false and wrong. But just because speciesism is false and wrong, it does not follow that every species-specific difference between animals is wholly irrelevant to the determination of moral significance or moral discriminations. On the contrary, Nagelís mental-mental gap entails that at least some species-specific differences between animals play a necessary role in the determination of the moral significance of animals and moral discriminations between animals.

As we will see in section V, perhaps surprisingly, two of the basic reasons for taking the MCP to be true, along with Nagelís mental-mental gap, are (1) Hilary Putnamís "multiple realization thesis," and (2) Jaegwon Kimís "structure-restricted correlation thesis." But perhaps even more surprisingly, as I will argue in section III, the fourth basic reason for accepting the MCP is a sharp distinction between pain and suffering, which carries along with it the notion of personhood, or the concept of a person in the moral sense. So while Bentham and Singer are correct that an animalís capacity for suffering is of fundamental moral significance, they are nevertheless mistaken that every animal capable of experiencing pain is thereby capable of suffering. All and only persons in the moral sense can suffer. In particular then, I think that we have good reason to believe that bats are capable of experiencing pain, but no good reason to believe that bats are capable of suffering, precisely because bats are not persons in the moral sense.

I anticipate that not only speciesists but also many animal ethicists are not going to like my conclusions. Nevertheless in order to reject my conclusions, dissenters are going to have to reject some very solid and widely-accepted theses in mainstream contemporary philosophy of mind: the mental-mental gap, the multiple realization thesis, and the structure-restricted correlation thesis. As I mentioned, the other basic premise in my argument is the sharp distinction between pain and suffering, and along with it, a corresponding concept of a person in the moral sense. These are the topics of the next two sections.

III. Pain and Suffering

I accept Nagelís canonical description of consciousness as what-it-is-like-to-be-for-an-organism. This is the same as what contemporary philosophers of mind call Ďphenomenal consciousnessí. Phenomenal consciousness, in turn, is the subjective experience of an animal. Phenomenal consciousness is "subjective" because it necessarily includes an ego or first person, and it is "experience" because it necessarily includes embodied sensations, feelings, and affects, particularly pleasure or pain, and more generally various mental states of animals to the extent that they have specific intrinsic qualitative or structural characters of sentience.

For the purposes of this paper, animals are sentient living organisms that become and remain individual members of their species just insofar as they possess a species-specific neurobiological causal basis of phenomenal consciousness or subjective experience (see section IV). But it is important to note that the notion of sentience is not precisely the same as the notion of subjective experience. Sentience is the organismís capacity for embodied sensation, feeling, and affect, particularly pleasure or pain. More broadly speaking, one can think of sentience as "the feeling of life" or "vital experience." All phenomenal consciousness necessarily includes sentience as a fundamental component. But the available psychological, neurobiological, and ethological evidence strongly suggests that Ďsentientí is a vague predicate, and that sentience is to some extent a matter of degree. Thus it seems very likely that not every species of living organism which arguably has some appreciable degree of sentience--say, bees, sharks, or snakes--actually has a subjectively centered or ego-based type of sentience, that is, a subjective "point of view." What this means is that the notion of a sentient living organism or animal, insofar as it is studied by cognitive ethologists and philosophers of mind, generally presupposes that the animal has passed some basic threshold of degrees of overall psychological, neurobiological, and behavioral organization into a definitely centered or ego-based form of sentience, that is, phenomenally conscious sentience. But it must also be admitted that there are extended uses of Ďsentient organismí that apply to borderline cases falling just below this basic threshold (e.g., bees, sharks, or snakes), and thus that the concept of sentience is not precisely the same concept as the concept of phenomenal consciousness or subjective experience. Another more linguistic way of putting this is to say that for me while Ďsentientí is a vague predicate, Ďconsciousí is not a vague predicate because it applied to all and only those sentient organisms that have passed a definite basic psychological, neurobiological, and behavioral threshold into subjectivity or egocentricity. But it is up to cognitive ethologists, not philosophers, to determine precisely where that definite basic threshold lies.

As we have just seen, sentience is a living organismís capacity for experience, that is, embodied sensation, feeling, and affect, particularly pleasure or pain, whether this sort of experience is subjectively centered (egocentric) or not. But in any case necessarily all phenomenally conscious animals are sentient. So obviously phenomenal consciousness is closely bound up with the capacity for experiencing pain.

Pain is the subjective experience of an animal in direct response to tissue damage or neurobiological systemic disruption caused by various intrusive exogenous stimuli such as burns, cuts, and collisions, or by various noxious endogenous stimuli including relatively enduring conditions such as disease or neurosis, and more temporary conditions such as migraine or emotional distress. It is empirically known that in humans at least, both the degree and also the specific character of pain are not wholly determined by the amount of tissue damage or neurobiological systemic disruption but in fact are partially determined by other factors such as anxiety-level, attention, prior experience, and suggestion. This is what I call the subject-dependency of pain. Moreover it is widely held by contemporary philosophers of mind that the causal-functional characterization of pain--that is, pain as characterized abstractly and relationally in terms of the overall pattern of causal transitions from sensory and behavioral stimulus inputs to the animal, through the specific neurobiological constitution of the animal, to behavioral outputs from the animal--can be held fixed, while systematically varying the specific neurobiological constitution of the animal. This is what I call the multiple realizability of pain. And it is also widely held by contemporary philosophers of mind that the specific character of the subjective experience of pain can be held fixed, while systematically varying the causal-functional characterization of pain. This is what I call the multiple functionality of pain.

Pain needs to be distinguished from pain-behavior. Behavior in general is how an animal moves or orients its body in response to exogenous or endogenous stimuli. Pain-behavior in particular is the characteristic set of unlearned or uncultivated species-specific behavioral responses to tissue damage or neurobiological systemic disruption. It seems correct to say that necessarily, other things being equal, if an animal is in pain, then it will also exhibit the pain-behavior of its species. This in turn implies that pain-behavior is, under some conditions, a reliable indicator of pain. But pain is not strictly speaking identical to pain-behavior, because when other things are not equal, it is possible for an animal or indeed even an entire species of animals to be in pain but fail to exhibit the pain-behavior of its species; and conversely, even when other things are equal, it remains possible for all the members of that species to fake pain by exhibiting the relevant pain-behavior without actually being in pain.

This brings me to a crucial distinction between (1) bodily pain and (2) suffering.

Bodily pain is pain that is phenomenally spatially localized in some part or parts of the animalís body for a certain definite duration of time, and also has a bodily cause. The actual bodily cause of bodily pain may not be spatially localized in the same area that pain is phenomenally spatially localized: this is vividly evident, for example, when pain is phenomenally spatially localized in a phantom limb. Nevertheless, bodily pain is necessarily always phenomenally spatially localized somewhere or another in the animalís body or its body-image. All bodily pain has a bodily cause, in the sense that some event inside or at the surface of the animalís body is a sufficient condition, under some psychological-cum-neurobiological law (i.e., a law that is "hedged" or ceteris paribus ), of a pain-event that is not earlier than the first event. But given the subject-dependency of pain, the degree of bodily pain may be altogether out of proportion to the actual extent of tissue damage or neurobiological systemic disruption. Bodily pain is also multiply realizable and multiply functional. It is possible for both humans and bats to have the same causally-functionally characterized type of bodily pain (e.g., by being burned or cut); and it is also possible for the same subjective experience of bodily pain to have different causal-functional characterizations and thereby play different causal-functional roles (e.g., the ordinary subjective experience of being burned or cut vs. masochism).

Suffering by contrast is the emotional pain of a person, which may or may not also involve any bodily pain, hence need not necessarily be spatially phenomenally localized (although suffering is always experienced during a certain definite duration of time), and need not necessarily have a bodily cause that is also the cause of bodily pain. Like bodily pain, suffering is subject-dependent (the degree or specific character of suffering is partially dependent on anxiety-level, attention, prior experience, and suggestion), multiply realizable (martians can suffer too), and multiply functional (the same subjective experience of suffering can play different causal-functional roles--there are suffering masochists, just as there are bodily pain masochists). Both bodily pain and suffering can be alleviated by drugs: sometimes by the same drug (e.g., alcohol), although usually by different ones (e.g., ibuprofen vs. anti-depressants). But suffering cannot always be alleviated by drugs, whereas bodily pain always can be. In persons, bodily pain can indeed constitute suffering, which is to say that in persons there can be a spatiotemporal coincidence between phenomenally localized pain and the cause of suffering. For example, I can suffer when my left leg hurts and just because my left leg has been damaged. But in principle, that token experience of suffering could have been spatiotemporally coincident with another different phenomenally localized bodily pain, and similarly with a different token of that type of suffering: I might have identically suffered, whether it was my left leg or my right leg that was hurting (token suffering can be preserved under phenomenal enantiomorphism in the body); and I might have suffered in just the same way, whether it was my leg or arm or head that was hurting (type suffering can be preserved under change of phenomenal spatial localization in the body).

This points up the crucial fact that bodily pain is not equivalent with suffering, despite the obvious fact that bodily pain and suffering often go together. It is possible to be in considerable bodily pain but not suffer at all (e.g., high performance athletes, professional dancers, sadomasochistic sex, Stoics, etc.), and it is also possible to suffer intensely but not be in any sort of bodily pain (e.g., extreme embarrassment, extreme shyness, guilt, jealousy, extreme disappointment, anxiety, fear, depression, and as the result of certain forms of emotional trauma such as rejection, betrayal, loss of a loved one, etc.). More generally, suffering essentially expresses a self-conscious subjectís frustrated or despairing sense of the inner or outer limits of his or her own intentional agency. I suffer because I am not what I want to be or the way I want to be, or because the world is not what I want it to be or the way I want it to be, or because other people are similarly recalcitrant in relation to my desires and my will. Our world is a vale of tears. But agent-centered frustration or despair can occur without any bodily pain, and bodily pain can occur without agent-centered frustration or despair.

It follows that pain does not always or necessarily involve suffering. This fact in turn directly implies that Benthamís and Singerís famous implicit inference from animal pain to animal suffering is fallacious: so for later reference, I will call this "the Bentham-Singer Fallacy."

A crucial feature of suffering, as opposed to bodily pain, is that all and only persons can suffer. This is because suffering requires an emotional complexity that is characteristic of all and only persons. Briefly put, all and only persons are capable of higher-order or self-conscious volitional pain, or suffering--pain that essentially expresses a self-conscious animalís frustrated or despairing sense of the inner or outer limits of its own intentional agency--and a capacity for having higher-order or self-conscious volitional states is at least a necessary condition of personhood and perhaps also a sufficient condition of personhood. Now of course I need to say more about the nature of a person in the moral sense. But before I do that I must also say something about animals and individuals.

IV. Animals, Individuals, and Persons

On the view I wish to spell out and defend, necessarily all persons are animals, but not all animals are persons. Furthermore, necessarily every person is also an individual animal belonging to some species or another, but not every particular member of that species is even an individual, much less a person.

What is an animal? The Oxford English Dictionary tells us that the word Ďanimalí means "a living organism which feeds on organic matter, usually one with specialized sense organs and nervous system, and able to respond rapidly to stimuli." In biology on the other hand, Ďanimalí has a more technical meaning, in that animals constitute one of the five kingdoms of living things: Monera (bacteria), Protoctists, Fungi, Plants, and Animals. The class of animals in this biological sense includes both vertebrates and invertebrates. My usage of Ďanimalí in this paper however is a precisification of the ordinary language term and is intended to coincide with its use in cognitive ethology.

Cognitive ethologists, by hypothesis, take the animals they are studying to be organisms with various mental capacities, including sentience, phenomenal consciousness, sense perception, imagination, desire, and volition. That includes all or at least most vertebrates, and possibly some but probably not all invertebrates. This may seem distressingly vague. But in fact the boundaries of the class of animals as conceived in cognitive ethology cannot now be precisely fixed. For example, are insects, fish, and reptiles animals in this sense? No one really knows. No doubt, sooner or later the boundaries will be more precisely fixed. Nevertheless, as I mentioned in the context of my earlier discussion of the sentience vs. phenomenal consciousness distinction, I regard such boundary-fixing as an empirical matter that is up to cognitive ethologists, not philosophers of mind. So no matter how cognitive ethology ultimately determines the boundary between animals and mere living organisms, I can assent to it.

What I am primarily interested in is how cognitive ethologists characterize those living organisms that now definitely fall into the class of animals in their sense of Ďanimalí: for example, humans, monkeys, bats, cats, horses, wolves, dogs, birds, bears, cows, mice, and so on. It is of course not possible to summarize the empirical results of cognitive ethology in a few words. But even at the risk of over-simplification, I can nevertheless say with some confidence that animals in the cognitive ethologistís sense are living organisms that (i) are token-reflexive foci of a variety of causal and semantic (informational) transactions with their local environments; (ii) are causal sources of a variety of self-initiated and informationally-sensitive movements in space and time; (iii) are capable of sense-perception, feeling, and affect, particularly pleasure and pain, and therefore have a sentient life of some sort; and finally (iv) have a "point of view," that is, have an egocentric, somatic (proprioceptive and kinesthetic) perspective, which may, as in the case of primates (and birds) be dominated by vision, but need not be, as in the case of echolocating bats. Speaking generally and philosophically, then, I will say that animals, as characterized by cognitive ethology, are sentient phenomenally conscious active living organisms.

Animals are individual members of their species. But not every particular living organism that belongs to a given species is an individual member of that species. For example, a human zygote or embryo (the sperm-fertilized ovum) is a particular living organism that belongs to the human species in the purely biological sense of sharing our species-specific genetic code: but a human embryo is not necessarily a human individual. This is for three reasons. First, early human embryos up to about the 14th day of their existence are "totipotential." This means, among other things, that one embryo can split and become two distinct human individuals (twins), and also that two embryos can fuse and become a single human individual (chimeras). Second, conjoined twins (i.e., so-called "Siamese twins," or humans with two distinct brains but only a single set of vital organs) constitute a particular living human organism, but they are not the same human individual. Third, anencephalic human babies (i.e., babies born with a functioning brain stem but no cerebral hemispheres or higher brain functions) are particular living human organisms, but not human individuals.

What then is an individual member of some species? My proposal is that something X is an individual belonging to some species S if and only if X is a particular living S-type organism and X also possesses a neurobiological causal structure or organization sufficient for the manifestation of a unique phenomenal consciousness or subjective experience under determinate real-world conditions. It should be noted that the uniqueness of subjective experience does not necessarily imply a unified subjective experience, where a capacity for the subjective unification of experience (i.e., "unity of consciousness" in one sense of that phrase) is the subjectís capacity simultaneously to combine all currently experienced qualitative characters or representational contents within a single conscious phenomenal field. For example, I am capable of divided conscious attention but nevertheless remain a unique consciousness even if for some reason (say, blindsight, commissurotomy, or masking) it is impossible for me to bring together all my currently experienced qualitative characters or representational contents with a single conscious phenomenal field. So according to my criterion of individuality, not only is each conjoined twin a distinct individual, but also each commissurotomy patient is a distinct individual. Or quite generally speaking: no matter what species S you belong to, you are precisely or numerically the very same S-type individual from the very moment you acquire the neurobiological causal basis of your unique S-type consciousness, and you continue to be precisely or numerically the very same S-type individual as long as this neurobiological causal structure or organization continues to exist, until the very moment this neurobiological causal basis is destroyed or permanently disrupted, and you are never precisely or numerically the same S-type individual otherwise. Or on other words, an individual is not merely alive: it is the subject of a life. This implies, for example, that the unfortunate Karen Quinlan was not the same human individual before and after her catastrophic car accident, precisely because after her accident the human individual that she had been in fact no longer existed. Thus Karen Quinlanís life ended with her accident, although a particular living human organism causally continuous with her body survived the accident. Nor of course was Karen Quinlan the same person before and after her accident, precisely because after her accident the person that she had been also in fact no longer existed.

This brings me to the concept of a person in the moral sense. The basis of my analysis is Harry Frankfurtís well-known and influential hierarchical volitional theory of persons, according to which "one essential difference between persons and other animals is to be found in the structure of a personís will." Frankfurtís account begins with the notion of a desire. A desire is a felt need (as opposed to an actual need: not all felt needs are actual needs) for something, or a preference for something, or a wish for something. To desire X is to want X; and to desire to X is to want to X. According to Frankfurt, some animals have not only "first-order desires," which are ordinary direct desires for things or events, but also "effective first-order desires." Effective first-order desires are desires that move (or will move, or would move) the animal all the way to action. An effective first-order desire is the same as an animalís will or first-order volition. First-order desires may or may not be accompanied by "second-order desires": to want (not) to want X, or to want (not) to want to X. If so, some of the second-order desires may be directed to the determination of precisely which first-order desire is the effective first-order desire, or the animalís will and first-order volition; and such desires are "second-order volitions." Whatever their order-level however, all desires or volitions can be either conscious or non-conscious. For convenience however I will concentrate exclusively on conscious desires and volitions.

According to Frankfurt all and only persons have second-order volitions, because all and only persons care about the constitution of their wills. By contrast to persons, creatures that are "wantons" have effective first-order desires, but they either lack second-order desires (hence they cannot care about the constitution of their wills) or if they have second-order desires they nevertheless lack second-order volitions (hence they can but in fact do not care about the constitution of their wills). Again according to Frankfurt, all non-human animals, all human infants, and some human adults are wantons. Finally, a person has freedom of the will if and only if she can determine by means of a second-order volition precisely which among her first-order desires is the effective one. This is also known as "identification" or "decision"; otherwise persons have unfreedom of the will. Wantons have neither freedom of the will nor unfreedom of the will, simply because they are not persons.

I disagree with Frankfurtís hierarchical volitional theory of persons on two relatively minor but still fairly significant points.

First, I doubt that Frankfurtís notion of personhood adequately captures the broadly-speaking Kantian sense of personhood, according to which persons are what I will call "higher-level" or "Kantian" rational animals, that is, animals capable of norm-guided logical or practical reasoning, moral self-legislation, and reflective self-consciousness. Higher-level or Kantian rationality involves the possession of capacities for the recognition of necessary truth, a priori belief or certainty, and non-instrumental, altruistic, non-hedonistic, non-consequentialist obligation or the "categorical Ďoughtí." By sharp contrast, what I will call "lower-level" or "Humean" rationality involves only the possession of capacities for the recognition of contingent truth, a posteriori belief, and instrumental, self-interested, hedonistic, consequentialist obligation or the "hypothetical Ďoughtí." All animals with a capacity for higher-level or Kantian rationality also possess a capacity for lower-level or Humean rationality, but not the converse. Toddlers and bonobo apes, for example, are Humean or lower-level rational animals but not Kantian rational animals. Animals with a capacity for rationality in the higher-level or Kantian sense are constrained in their intentional agency by the Categorical Imperative or at least by some other suitably universal, non-instrumental, altruistic, non-hedonistic, and non-consequentialist moral principles. By contrast, animals with merely a capacity for rationality in the lower-level or Humean sense are constrained in their intentional agency only by (at least some of) the axioms of rational choice theory.

Now some animals are persons in Frankfurtís sense, and also rational in the lower-level or Humean sense, but not rational in the higher-level or Kantian sense. Anyone who has looked after a toddler knows that it is possible for an animal to care very deeply about the constitution of her will but not (yet) be capable of norm-guided logical or practical reasoning, moral self-legislation, or reflective self-consciousness. Again, a toddler can identify or decide in the Frankfurtian sense by using a second-order volition to determine precisely which among her first-order desires is her effective first-order desire. But the identifications or decisions of toddlers are at best momentary or temporary, and do not occur consistently or over an extended period of time. As everyone knows, toddlers are fickle and willful. In this way, toddlers are capable of many different sorts of complex affects and complex emotional states, but they are not higher-level or Kantian rational animals. So toddlers are persons in the Frankfurtian or hierarchical volitional sense, but not persons in the Kantian sense. This subtlety in the notion of personhood can be conceptually finessed by distinguishing between K(antian)-persons and F(rankfurtian)-persons. All K-person are F-persons, but only some F-persons are K-persons.

The K-person vs. F-person distinction between can be further elaborated by further unpacking the subtle differences between K-persons and toddlers. As everyone knows, toddlers exemplify the fact of phenomenal consciousness and are also minimally self-conscious in the sense that they can recognize themselves in a mirror and make simple judgments about some of their own mental states. They may also know a few words of their native natural language. They have simple beliefs, and if they do know a few words of their language, they can carry out some simple inferences. They want things. Also they usually know what they want, and they care a great deal about getting what they want. They can mentally cause movements of their own bodies by means of their volitions. They know their own names. They are intellectually curious, have capacious memories and wonderful imaginations, and are sometimes highly insightful. But they are also extremely naïve and uncritical, highly emotional, highly inconsistent in their behavior, and above all fickle and willful, and certainly cannot be held personally or morally responsible for their actions. They are incapable of conceiving their own lives as a whole. They cannot engage in retrospective or prospective self-interpretation and self-evaluation, explicit step-by-step deliberation about immediate action, or long-term planning. They do not know the normative significance of their own deaths--that persons must strive to make their individual finite lives into meaningful and coherent wholes. They need to be most carefully looked after, gently but firmly told what to do, loved unconditionally, and at all times protected from the vicissitudes of an often unfriendly and violent world. They are not autonomous in the Kantian sense, or capable of moral self-legislation. Otherwise put, toddlers are F-persons because they engage directly in the process of constituting their wills by means of second-order volitions, but they are not (yet) rational agents in the higher-level or Kantian sense of rationality.

What then is the basic difference between K-persons or rational animals in the higher-level or Kantian sense, and F-persons? The answer is that K-persons or rational animals in the higher-level or Kantian sense are inherently non-instrumental-reasons-giving or categorically normative-reflective animals, while F-persons are not inherently so. In other words, rational animals in the Kantian or higher-level sense are inherently capable of giving non-instrumental reasons for their beliefs and actions, and thus are inherently capable of being categorically normatively self-guided in their beliefs and actions, whereas at least some F-persons are not capable of this. Or in still other words, rational animals in the Kantian or higher-level sense have a capacity for a priori epistemic and practical justification and in particular for a priori epistemic and practical self-justification, whereas at least some F-persons do not.

Second, I disagree with Frankfurt about the scope of the class of wantons. In particular, in view of strong evidence from cognitive ethology, I believe that at least some non-human animals--and in particular, apes--are F-persons.

These two minor critical extensions of Frankfurtís theory are of fundamental moral importance. If I am correct that both K-persons and F-persons are persons in the moral sense, then it follows that rationality in the higher-level or Kantian sense is not a necessary condition of personhood (although of course it remains a sufficient condition). On the contrary, the necessary and sufficient condition of personhood is hierarchical volitional emotional complexity, or the capacity for second-order volitions, whether or not this is accompanied by a capacity for norm-guided logical or practical reasoning, moral self-legislation, and reflective self-consciousness. This directly implies the existence of a class of persons between mere animals and higher-level or Kantian rational animals. And if, in turn, as many philosophers and non-philosophers alike believe, an animal has a right-to-life if and only if it is a person, this directly implies that it is the psychological capacity for having complex emotional states and in particular second-order volitions, that necessarily and sufficiently triggers a right-to-life, not higher-level or Kantian rationality.

Closely related to the right to life is what I will call the right not to suffer. The right not to suffer is the right not to be unjustly caused to suffer. Its moral rationale is that since (other things being equal) a personís permanent death is always a bad thing for that person, and since the only morally sufficient reason for the moral permissibility of euthanasia or mercy killing--whether voluntary or involuntary, or active or passive--is the prevention of personal suffering, the only thing that can be worse for a person than to die or to be killed is to be caused to suffer. So if there is a right not to be unjustly killed (i.e., the right to life), then there must also be a right not to be unjustly caused to suffer. In this way, just as the psychological capacity for having complex emotional states and in particular second-order volitions necessarily and sufficiently triggers a right-to-life, so too it necessarily and sufficiently triggers the right not to suffer.

Now I believe that all toddlers, as well as some abnormal humans, including many victims of Alzheimerís disease, autism, cerebral palsy, Downís syndrome, and spina bifida, are F-persons but not K-persons. If I am right about this, then it follows that all toddlers and many victims of Alzheimerís disease, autism, cerebral palsy, Downís syndrome, and spina bifida have a right-to-life and also a right not to suffer by virtue of their hierarchical volitional emotional complexity, despite their being incapable of norm-guided logical or practical reasoning, moral self-legislation, and reflective self-consciousness, not to mention their also being pre-linguistic animals or partially linguistic animals. Furthermore, if I am also correct that not all non-humans are wantons, and that there are some non-human F-persons (e.g., apes), then it also is directly implied by this fact together with the above points that some non-humans have a right-to-life and also a right not to suffer by virtue of their hierarchical volitional emotional complexity, despite their being incapable of norm-guided logical or practical reasoning, moral self-legislation, and reflective self-consciousness, not to mention their also being partially linguistic or non-linguistic animals. Or more bluntly put, I am saying that morally we should care every bit as much about apes (and possibly other primates, as well as whales and dolphins) as we already do about toddlers and about many victims of Alzheimerís disease, autism, cerebral palsy, Downís syndrome, and spina bifida. But time and energy spent morally worrying about normal human fetuses prior to the onset of sentience or phenomenal consciousness, or about bats, cats, dogs, horses, cows, etc., or about humans that are merely and permanently sentient, in persistent vegetative states, or anencephalic, are probably wasted time and energy. Now I want to try to prove this explicitly.

V. The Moral Comparison Principle

My argument for the Moral Comparison Principle or the MCP, as I mentioned in section II, uses four basic premises: (1) the mental-mental gap; (2) the sharp distinction between pain and suffering, along with the corresponding concept of a person in the moral sense; (3) the multiple realization thesis; and (4) the structure-restricted correlation thesis. I have already explicated (1) and (2). So before we get to my argument for the MCP proper, I need to say something about (3) and (4).

Both the multiple realization thesis and the structure-restricted correlation thesis arise out of philosophical debates about functionalism. Functionalism holds that mental properties are not intrinsic or necessary and non-relational properties of something, but are instead functional properties: extrinsic or contingent and relational properties of something. More precisely, functional properties are patterns of occurrent or dispositional causal transitions from inputs to outputs, applying to the external and internal states of physical machines or living organisms. Standard examples of functional properties are the properties instantiated by sequences of digital computations in a Turing machine, and the properties instantiated by those neurobiological processes in the brains and central nervous systems of animals that are apt to cause behavior. According to materialist functionalism, functional properties are second-order physical properties that are supervenient on first-order physical properties: so the materialist functionalist holds that mental properties are identical to a certain special sort of second-order physical property.

Materialist functionalism is committed to the truth of both the multiple realization thesis and the structure-restricted correlation thesis. The multiple realization thesis asserts that one and the same functional property can be instantiated in many different actual or possible physical individuals, types of organism, natural kinds, and compositional stuffs. This is verified by the fact that the very same computational software can be instantiated in many different sorts of hardware, and that the very same type of neurobiological process (e.g., digestion) can be instantiated in many different species of animals (e.g., humans and cats). The structure-restricted correlation thesis, on the other hand, asserts the relativization of the instantiation of mental properties to species-specific or even sub-species-specific physical structure types, or as Kim puts it:

If anything has mental property M at time t, there is some physical structure type T and physical property P such that it is a system of type T at t and has P at t, and it holds as a matter of law that all systems of type T have M at a time just in case they have P at a time. In other words, mental properties occur in animals under specific physical conditions in a lawful way, and this lawful regularity is found across whole species or sub-specific groups sharing the same physical constitution. If mental properties are functional properties, then the structure-restricted correlation thesis is a necessary condition of the multiple realizability of mental properties. So if materialist functionalism is true, then the structure-restricted correlation thesis follows automatically. But even if mental properties are not functional properties, nevertheless the structure-restricted correlation thesis still seems to be true, because all that it says is that X (i.e., an animal) has a mental property at a time only if X has a certain physical structure and there is some lawful connection between instantiations of that mental property and basic properties of that physical structure, and this seems unexceptionable. Indeed, it is important to see that the structure-restricted correlation thesis is perfectly compatible with various denials of materialism, whether reductive materialism or nonreductive materialism. Even if mental properties are neither identical with nor in any sense supervenient on physical properties, there can still be a structure-restricted correlation between mental properties and physical properties in animals.

How can we isolate a given structure-restricted correlation? The correct answer, I think, is the one offered by Kim, namely that we isolate it by finding the causal-functional characterization of that mental property, and then correlating the causal-functional role picked out by that characterization with a certain species-specific or sub-species-specific neurobiological constitution in animals. Again however, it is crucial to remember that we need not identify mental properties with functional properties in order to do this: all we need is to have a causal-functional characterization of the relevant mental property. Consider, for example, the causal-functional characterization of the mental property of being in pain that I sketched in section III: pain is the animalís subjective experience of tissue damage or neurobiological systemic disruption within its own body; pain is subject-dependent, multiply realizable, and multiply functional; and pain has a necessary connection (other things being equal) to animal behavior. The causal-functional role of pain is multiply realized in humans, apes, bats, and so-on, and therefore determines a set of structure-restricted correlations between the mental property of being in pain and different species-specific or sub-species-specific neurobiological constitutions.

This point leads to one last concept that we need before we get to my argument for the MCP: the concept of a "schematization" of a mental property. The basic idea behind the schematization of a mental property is that animals of different species or sub-species will typically subjectively experience the same sorts of things--e.g., colors, sounds, or pain--somewhat differently, precisely because their neurobiological constitutions are somewhat different, and also in some sort of systematic relation to the various differences in neurobiological constitution. Again, schematization is the way that different types of animal body directly affect the specific character of phenomenal consciousness, generally in the form of a distinctive species-specific body-image that gives a correspondingly distinctive underlying phenomenal spatiotemporal organization to the animalís sensations and affects. Slightly more abstractly put, a mental property is schematized just in case the specific subjective experiential character of its instances is regularly and systematically modified and shaped (via a body-image) by the multiple realizations of its corresponding causal-functional role under different structure-restricted correlations. My thesis is then that necessarily all phenomenal consciousness is schematized. This is what I will call the schematization principle. Human pain and bat pain are both pain, in the sense that they each play the same causal-functional role in the human species (homo sapiens) and the bat species (michrochiroptera): but in virtue of the schematization principle it seems very likely that bat pain is sharply different from human pain, not merely neurobiologically, but also phenomenologically. Indeed it is precisely schematization, I think, that implicitly drives the basic intuition that Nagel uses to motivate the mental-mental gap. We lack any sort of adequate conceptual understanding of what it is like to be a bat precisely because the subjective experiences of humans and of bats are schematized very differently, and because the differing contents of schematization across different species or sub-species are given directly only to the phenomenally conscious members of those very species or sub-species.

Here now, finally, is my argument for the MCP.

    1. Assume Nagelís conception of phenomenal consciousness as what it is like to be for an organism, understood now as what it is like to be for an animal.
    2. Phenomenally conscious states like pain have corresponding causal-functional characterizations that describe their causal-functional roles.
    3. The causal-functional role of pain is multiply realized under different species-specific or sub-species-specific structure-restricted correlations.
    4. This in turn implies the schematization principle as applied to pain.
    5. Given the schematization principle and Nagelís mental-mental gap, there is good reason to believe that bat pain is not only phenomenologically very different from human pain but also in fact conceptually inaccessible for us.
    6. We have good reason to believe that a subjective experience can be as morally significant as human pain only if it is phenomenologically quite similar to human pain. This is because our reason for believing that the subjective experience of pain in our species is even sometimes morally significant is necessarily based on first-person evidence. But the less phenomenologically similar a given subjective experience-type E1 (say, bat pain) is to another subjective experience type E2 (say, human pain), then the less reason we have to believe that E1 has all or even any of the consciousness-based properties that E2 has.
    7. So since there is good reason to believe that bat pain is not only phenomenologically very different from human pain but also in fact conceptually inaccessible for us, then there is good reason to believe that bat pain is not as morally significant as human pain.
    8. All and only persons can suffer.
    9. Suffering has fundamental moral significance, because personhood in the moral sense entails both the right to life and also the right not to suffer.
    10. Now some humans, and also some non-humans, are persons in the moral sense and therefore can suffer. By contrast, since bats are not persons, bats are capable of having only bodily pain, and cannot suffer.
    11. Some human or non-human animals are not persons but will in the natural course of their genetic development become persons. To the extent that these animals have already begun to acquire some of the psychological sub-capacities that are jointly constitutive of personhood and the capacity for suffering, there is good reason to believe that they are more morally significant than bats, since bats will never become persons or suffer in the natural course of their genetic development.
    12. Therefore, there is good reason to believe (i) that the pain of human or non-human animals that are persons is more morally significant than the pain of bats, and also (ii) that the pain of human or non-human animals that will become persons in the natural course of their genetic development is more morally significant than the pain of bats.

    13. This preliminary conclusion is what I will call the bat-restricted moral comparison principle. It is crucial to see that this principle could not be restricted to apes. This is because (as evidence from cognitive ethology strongly indicates) apes are persons, and also because there is little or no reason to think that the schematization principle and Nagelís mental-mental gap will entail that the pain of apes is phenomenologically very different from human pain or conceptually inaccessible to us, in view of the fact that (as evidence from cognitive neuroscience strongly indicates) the neurobiological constitution of apes is very similar to that of normal humans. Indeed, the degree--if not the precise kind--of neurobiological difference between normal apes and normal humans seems to be not appreciably larger than the degree of neurobiological difference between many humans who are victims of Alzheimerís disease, autism, cerebral palsy, Downís syndrome, or spina bifida, and normal humans.

      Now take the bat-restricted moral comparison principle and substitute for the first occurrence of Ďbatsí the phrase Ďany non-human species of animals that are non-personsí and substitute for the second occurrence of Ďbatsí the phrase Ďany non-human species of animals that will never become persons in the natural course of their genetic developmentí. These substitutions are justified by the concept of a person in the moral sense, together with the thesis that all and only persons can suffer, together with the thesis (shared by Bentham and Singer) that the capacity for suffering confers fundamental moral significance on an animal. Making those substitutions, we get

    14. Therefore, there is good reason to believe (i) that the pain of human or non-human animals that are persons is more morally significant than the pain of any non-human species of animals that are non-persons, and also (ii) that the pain of human or non-human animals that will become persons in the natural course of their genetic development is more morally significant than the pain of any non-human species of animals that will never become persons in the natural course of their genetic development.

    15. This intermediate conclusion is what I call the cross-species moral comparison principle.

      Now, finally, take the cross-species moral comparison principle and substitute for both occurrences of Ďany non-human species of animalsí the phrase Ďany species of animalsí. These substitutions are justified by anti-speciesism as applied to humans, or the rejection of the thesis that some species-specific difference between humans and non-humans is the sole or primary determinant of the moral significance of animals or moral discriminations between animals. If by some cosmic accident it turned out that no humans were persons, and that all persons were non-humans, then the basic rationale for the MCP would hold just the same. Making those substitutions, we get

    16. Therefore, there is good reason to believe (i) that the pain of human or non-human animals that are persons is more morally significant than the pain of any species of animals that are non-persons, and also (ii) that the pain of human or non-human animals that will become persons in the natural course of their genetic development is more morally significant than the pain of any species of animals that will never become persons in the natural course of their genetic development.
And this last conclusion is the MCP.

VI. Conclusion

The ethical payoffs of the MCP are considerable. If I am right, then we have good reason to extend serious moral consideration to some non-human animals and in particular to apes, but no good reason to treat all animals equally. We also have good reason to think that the arbitrary late-term abortion or torture of normal human and ape fetuses, the arbitrary infanticide or torture of normal human and ape infants, the arbitrary killing or torture of toddlers and normal young apes, the arbitrary killing or torture of many victims of Alzheimerís disease, autism, Downís syndrome, and spina bifida, the arbitrary killing or torture of normal adult apes, as well as the arbitrary killing or torture of K-persons of any species, are all impermissible. But we have no good reason to think that the abortion of human fetuses prior to the onset of sentience or phenomenal consciousness, or the killing of merely and permanently sentient humans or of non-sentient humans, including anencephalic infants and humans in persistent vegetative states, are impermissible.

It should be noted however that the MCP does not entail that human or non-human animals that are non-persons, like bats, are without any moral significance at all. The capacity for suffering confers fundamental (or basic-rights-entailing) moral significance on an animal, and the fact that an animal is not capable of suffering entails that it is less morally significant than any animals that are capable of suffering: but this fact does not wholly undermine its moral significance. Animals that are not persons and are never going to develop into persons, and therefore are incapable of suffering, do not have intrinsic non-instrumental moral significance. But they might well have extrinsic or instrumental moral significance, and possibly of a high degree. For example, it seems very wrong to torture non-human animals that are merely sentient or phenomenally conscious (e.g., bats) because such torture strongly tends to promote the brutal treatment of non-human animals that are or are going to become persons, just as it seems very wrong to torture human animals that are merely sentient or phenomenally conscious (e.g., those in the final stages of Alzheimerís disease, or extreme victims of cerebral palsy or spina bifida) because this strongly tends to promote the brutal treatment of human animals that are or are going to become persons. But it does not seem wrong, other things being equal, painlessly to kill human animals in the final stages of Alzheimerís disease, or extreme victims of cerebral palsy or spina bifida. And the same goes for non-human animals that are merely sentient or phenomenally conscious.

Here, finally, is another interesting consequence of my argument for the MCP. If animal ethicists want to hold that "pain is pain is pain, wherever it is experienced," then they must defend a reductive materialism of the old-fashioned, now generally rejected, "type-physicalism" variety. In other words, they must reject both the multiple realization thesis and the structure-restricted correlation thesis, which are accepted by virtually everyone who works in mainstream contemporary philosophy of mind, whether materialist or non-materialist, as well as rejecting Nagelís widely-accepted mental-mental gap argument. But even if animal ethicists do opt for these highly implausible views in the philosophy of mind, they still cannot help themselves to the direct inference from animal pain to animal suffering without committing the Bentham-Singer Fallacy.