Wednesday, September 18, 2013

Killoren on Pets, Livestock, and Narrative Value

[Guest post by David Killoren...]

I’m very grateful to Richard for allowing me this space to hash out some ideas about pets and livestock. I’ll try to be relatively quick here (although I recognize this post is a bit too long as-is). In the spirit of blogging, I will avoid footnotes, references, and acknowledgments. But I do want to say that I got a lot of great ideas and feedback from the folks at the Rocky Mountain Ethics Congress in Boulder this past August. If you’re one of those folks, please feel free to chime in in the comments below. (During the conference, I was trying to keep a list of the people who gave me feedback, but now I’ve misplaced the list.) When I have a new version of the paper on this topic, I hope that Richard will allow me to link it from this post. In the meanwhile, I eagerly welcome any objections, suggestions, off-the-wall ideas, references, etc.


In the recent past, the United States had (and perhaps still has) a racial caste system. White people occupied a higher caste. Nonwhite people occupied lower castes; there was not just one lower caste. For example, black people with lighter skin generally found it easier than darker-skinned black people to gain status, were considered eligible for a wider range of jobs, etc. So, we should say, the American caste system had several tiers: White people—or perhaps just white men, or white wealthy men, or whatever—were in the top tier, and other people were put into various lower tiers.

Something similar is true of our treatment of non-human animals.

Obviously, we treat human beings differently than we treat non-human animals. For example, we use non-human animals as a food source, but we do not (ordinarily) use human beings as a food source. There is a vast gulf between our treatment of human beings and our treatment of non-human animals, and this fact looms large in the thinking of moral philosophers and others who write and talk about animal issues.

Nevertheless, we should not overlook the fact that we do not treat all non-human animals in the same way. I suggest that we have a multi-tiered animal caste system, and that its shape is roughly as follows.

Human beings occupy the top tier. On the whole, in comparison to the way we treat non-human animals, we treat one another exceedingly well (although of course not all human beings are treated well, and not all are treated equally).

Pets, I suggest, are the main occupants of the second tier. (Other animals—e.g., various service animals, members of endangered species, etc.—might also be in this tier.) These animals are often treated almost as well as human beings. If we are good pet owners, we give our pets good food, protect them from the elements, care whether they have adequate opportunities for play and socialization, spend significant amounts of money on various medical treatments in order to improve and extend their lives, and so on.

[Side note on language: I’m using “pet” rather than “companion animal” just because (a) I want easy access to our moral intuitions, and (b) I think we can access our moral intuitions more easily if we use commonly-used language, and (c) I think “pet” is the commonly-used term (whereas “companion animal” is, I think, a piece of jargon). For the same reason, I will speak of “owners” here, even though I’d rather not presuppose that humans can legitimately own non-human animals.]

Livestock animals are in the third tier. We don’t treat livestock very well at all. We kill them in order to turn them into meat, and while they are alive we make them miserable (in modern factory farms) in order to ensure that their meat is plentiful and cheap. To treat a pet in this way, by contrast, is seen as very seriously wrong.

Nuisance animals (i.e., pests, such as rats) might occupy an even lower tier. We cruelly use livestock animals, and we do not much care whether they suffer, but their suffering is not our aim. By contrast, we sometimes regard the suffering and demise of pests as a worthy goal in and of itself. (When we see a mouse caught dead in a trap, we sometimes say: Good!)

So here’s one question: Is our treatment of livestock animals morally acceptable, given that we believe that it would be wrong to treat human beings in the same way? This question has gotten a lot of attention, and deservedly so.

Here’s another question: Is our treatment of livestock animals morally acceptable, given that we believe that it would be wrong to treat our pets in the same way? This question has gotten considerably less attention.

I’m interested in the latter sort of question here.


Our animal caste system, as it is currently put into practice, is not morally acceptable—not by a longshot. If you doubt this claim, then either you’re ignorant or I’m deeply confused about what happens to livestock animals in factory farms. I recommend Jonathan Safron Foer’s Eating Animals, which paints a gruesome portrait of the lives (and deaths) of livestock animals. It’s extraordinarily well-written for an activist book, and I believe it is mostly trustworthy.
We create and destroy a galaxy of livestock animals every eight years (or perhaps less): According to a relatively low-ball estimate, approximately 50 billion livestock animals die in factory farms each year; and according to a high-ball estimate, there are 400 billion stars in the Milky Way. If each of those animals’ lives is even a little bit bad, then our system of factory farms is responsible for immense badness. And the life of a typical livestock animal in a factory farm is, I believe, really quite bad. In light of the facts about the scale and depth of misery in factory farms, it is arguable (and has been argued) that our system of factory farms is the most harmful thing that human beings have ever made.

We can debate whether animals experience pain, whether their pain is bad, etc., but this is not the place for that. I am going to simply assume that our treatment of livestock animals is wrong and ought to change. But how should it change? Should our animal caste system be abolished or merely reformed?


According to a set of widely shared intuitions, non-utilitarian constraints on harm apply to human beings. For example, most people believe that it is morally wrong to kill one innocent person in order to harvest her organs to be distributed to five innocent people, even if this would be optimific (i.e., would do the most good in utilitarian terms).

Interestingly, however, at least in my own reflections, such constraints do not seem to apply to livestock animals. For example, it seems to me that it would be morally acceptable to kill one livestock pig in order to harvest her organs to be distributed to five livestock pigs, provided that this would be optimific.

In fact, it seems to me that it may be morally required (pro tanto) to harm livestock animals in cases in which this would be optimific. Consider a case in which a farmer kills a few diseased herd animals in order to prevent the disease from spreading to other members of the herd. We may think that the farmer in this case merely discharged her obligation, as a farmer, to protect the general health of her animals.

Yet I find that non-utilitarian constraints do seem to apply to pets. For example, it seems to me that it would be morally wrong for me to kill my dog in order to harvest his organs to be distributed to five other dogs, even if this would be optimific.

I think that these intuitions might be useful to those who want to argue in favor of reform (rather than abolishment) of the animal caste system. Reformers can say: “Given that non-utilitarian constraints apply to pets but do not apply to livestock animals, there is a moral difference between those two types of animals. Our treatment of these different types of animals, then, ought to reflect the moral difference between them. This means that we ought to have some sort of animal caste system (even if the caste system that we ought to adopt turns out to be very different from the caste system that we presently have).”

However, in order for the reformer to use the above intuitions to support her position, it would be necessary first to provide some sort of an explanation of how these intuitions might be correct. After all, intuitions can be mistaken.

So: Why would non-utilitarian constraints apply to pets without applying to livestock? What’s the morally relevant difference between these two types of animals?


One possibility is that there is a morally relevant species-level difference between pets and livestock. Pet species include dogs, cats, etc.; livestock species include pigs, chickens, cows, etc. Perhaps these different species are different in some morally relevant respect.

There are two problems with this suggestion. The first is that it is very hard to find a morally relevant difference between pet species and livestock species. Dogs and pigs, for example, are similar in many morally relevant respects—cognitively, emotionally, etc. Second, any given livestock species can be a pet species, and any given pet species can be a livestock species. There are pet pigs, pet chickens, pet cows, livestock dogs, livestock cats, etc.

Another possibility—which I mention because some people appear to endorse it—is that we have some kind of contract (i.e., a promise, agreement, or other sort of arrangement) with our pets to treat them (and not to treat them) in certain ways, whereas we have not made such a contract with our livestock.

There are three problems with this suggestion. First, even if I have made a contract with my own pet, I do not seem to have made any such arrangement with others’ pets. Yet it seems to me that non-utilitarian constraints apply to all pets, not just to one’s own pets. That is, it seems that it would be wrong for me to kill someone else’s pet, even in certain cases in which doing so would be optimific. Second, I am skeptical of the idea that it is even possible to enter into a contract with a non-human animal, given that contracts are normally made using complex language. Third, even if it is possible to make contracts with non-human animals, it seems unlikely that such contracts would exist in all or even most of the relationships that we have with our pets. (When one brings an animal into one’s home, feeds it, gives it care and attention, etc., and thereby makes it into a pet, this does not seem to entail the existence of any kind of contract. Compare: I can bring a homeless person into my home and feed her, etc., without making any contract with her.)

A variant of this suggestion is the idea that human beings, not individually but collectively, have made contractual arrangements with other various pet species. For example, it may be proposed that humanity has struck some kind of bargain with dogs: We feed and protect them, and in exchange, they do jobs for us (e.g., guard our sheep) and entertain us. I am skeptical of this idea simply because I do not understand how an entire species might have bargained with another species in any meaningful or morally significant way.

A third possibility: The relationship between pets and humans has some morally important feature—and this feature, whatever it is, is missing from the relationship between livestock and humans. This strikes me as a promising proposal. And I believe that some version of this proposal has been the focus of most philosophers who have tried to grapple with the moral status of pets. I won’t go into the reasons why I am unsatisfied with other versions of this proposal that I have seen. Instead, I’ll just describe the version of the proposal that I’m hoping to be able to accept.


Before trying to explain the moral gap between pets and livestock, let us first consider the moral gap between humans and livestock. Why would it be the case that non-utilitarian constraints apply to humans but not to livestock? Here is one possibility, which I will call the Narrativity Hypothesis. The hypothesis can be stated in three parts.

(1) Humans do two morally significant things: (a) We have positive and negative experiences; and (b) by reflecting on our past experiences and anticipating our future experiences, we integrate our experiences into a self-told life story. By contrast, livestock animals do only one morally significant thing: They have positive and negative experiences. Livestock animals lack the intellectual abilities necessary in order to integrate their experiences into self-told stories.

(2) A positive experience has positive hedonic value (and a negative experience has negative hedonic value). Human lives and livestock lives both possess hedonic value. However, when a human being integrates her experiences into a self-told life story, she imbues her life with an additional kind of value: narrative value. Livestock lives lack this sort of value.

(3) Hedonic value is morally fungible: it is morally permissible (pro tanto) to destroy (or prevent) a certain amount of hedonic value in order to produce (or salvage) a greater amount of value. By contrast, narrative value is not morally fungible: it is morally wrong (pro tanto) to destroy (or prevent) a certain amount of narrative value even if doing so would produce (or salvage) a greater amount of value.

Now, when I kill a livestock animal in order to save the lives of five livestock animals, I destroy some hedonic value in order to salvage a greater amount of value. This is pro tanto permissible, according to the Narrativity Hypothesis. By contrast, when I kill a human being in order to save the lives of five human beings, I destroy narrative value as well as hedonic value, and this is pro tanto wrong, according to the Narrativity Hypothesis, even if it doing so salvages a greater amount of value. Thus the Narrativity Hypothesis can explain why non-utilitarian constraints apply to humans but not to livestock animals.

What about pets? The Narrativity Hypothesis, in and of itself, does not contain the resources to explain why non-utilitarian constraints would apply to pets. After all, a pet, like a livestock animal, lacks the intellectual abilities necessary in order to integrate its experiences into a self-told life story. So, it seems, the Narrativity Hypothesis suggests that pets’ lives should possess only a morally fungible type of value—namely, hedonic value, not narrative value.

But here’s an idea: Pets cannot do the work of integrating their lives into stories—but maybe human beings can do the necessary work for their pets.

I integrate the experiences of my life into a self-told story by being a close observer of my own life. Well, pet owners are usually close observers of their pets’ lives. This is how it is with my dog and me. I have an Italian Greyhound. His name is Amtrak. I know where Amtrak was born and I know something about the circumstances of his birth. I know approximately how old he was when I adopted him and how old he is now. I have watched as he has struggled at some things, succeeded at some things, failed at some things, learned from mistakes, passed milestones, etc. I make short-term and long-term plans for him. I foresee a future in which he will senesce, slow down, and die.

Perhaps, when I observe and reflect on my dog’s life, I do the cognitive work of integrating his life into a story. And perhaps, in doing this for my dog, I give his life a certain kind of narrative value, which we might call narrative value by proxy—a type of value that his life would lack if it weren’t for his relationship with me.

Let us say that the Extended Narrativity Hypothesis is the original Narrativity Hypothesis above, plus the supposition that (a) human beings, by relating to their pets in a certain way, imbue their pets’ lives with by-proxy narrative value, and (b) by-proxy narrative value is not morally fungible. The Extended Narrativity Hypothesis implies that non-utilitarian constraints should apply to the lives of pets.


I wonder whether the following might true: An individual S is a person in the moral sense iff S is a bearer of narrative value.

If this were true, then the Extended Narrativity Hypothesis would be an extended personhood hypothesis—a moral analog of the idea of the extended mind. Personhood ain’t all in the head, according to this view, because a pet dog (for example) can be a person in virtue of (a) the dog’s capacity for positive and negative experiences (which is in the dog’s head) plus (b) a human owner’s capacity to integrate the dog’s experiences into a story (which isn’t in the dog’s head).


One advantage of the Extended Narrativity Hypothesis is that it can explain why non-utilitarian constraints apply to humans and to pets but not to livestock.

Another advantage is that it can explain why non-utilitarian constraints should apply to certain severely disabled human beings who are intellectually similar to livestock animals. Such people normally have caretakers. The caretaker of a severely disabled person can do the work of integrating the disabled person’s life into a story, thus giving by-proxy narrative value to the disabled person’s life.

I want to mention two big disadvantages of the Extended Narrativity Hypothesis:

First, the Extended Narrativity Hypothesis implies that non-utilitarian constraints should apply to some livestock animals. If a farmer has a close relationship with her animals, then she might inadvertently give by-proxy narrative value to her animals’ lives. This seems more likely to happen in a small farm setting than in a factory farm. Thus, I suspect, proponents of the Extended Narrativity Hypothesis will need to say that the killings that happen at small farms are often morally problematic—and that the killings that happen in factory farms are not morally problematic in quite the same way. This is counterintuitive because we normally regard small farms as far better, from a moral point of view, than factory farms. (The Extended Narrativity Hypothesis proponent can still say, however, that factory farms are very morally problematic, given that factory farms produce a much larger amount of negative hedonic value than positive hedonic value.)

Second, the Extended Narrativity Hypothesis implies that non-utilitarian constraints should not apply to some human beings. If a severely disabled human being lacks the ability to integrate her life into a self-told story, and if no human caretaker does the necessary integrative work for her, then the disabled human’s life is left devoid of narrative value. (The Extended Narrativity Hypothesis proponent can still say, however, that the disabled human’s life in such a case can possess hedonic value, and this type of value can be morally important even though it is morally fungible.)

So, it seems, the proponent of the Extended Narrativity Hypothesis is going to have to bite some bullets. However, I think these bullets are edible if bad-tasting, and may be worth biting if we are very interested in defending reform (rather than abolition) of the animal caste system. I think the Extended Narrativity Hypothesis, or something like it, might be the reformer’s best hope. But of course, before we can know whether the Extended Narrativity Hypothesis is viable, much more needs to be said than I can say here.


If we were to adopt the Extended Narrativity Hypothesis, what sort of animal caste system would we want to have?

First, we would have to acknowledge that livestock animals’ lives do possess value. In particular, their lives possess hedonic value. Hedonic value is morally fungible, according to the Extended Narrativity Hypothesis, but this does not imply that hedonic value is morally unimportant. Something can have a large amount of fungible value. (Imagine a treasure chest containing $100 million dollars. This chest is very valuable—even though we would gladly destroy it in exchange for a treasure chest containing $101 million dollars.) Indeed, it is sometimes better to lack non-fungible value; it can be quite inconvenient to be the sort of being whose life possesses a non-fungible type of value. (For example, if I am one of five people who need an organ transplant, I might wish that the value of human lives were morally fungible.)

Given that livestock animals’ lives possess value, we would need to restructure our food system in a way that is respectful of that value. We would permit the destruction of livestock animals’ lives only in cases in which a greater amount of value is produced or salvaged than prevented or destroyed. I imagine that this could be the case, for example, in a farm where the animals have happy lives and are killed quickly and with minimal pain. This is not the case in the factory farm system that exists currently.

Second, we would need to make sure that we are not routinely killing animals whose lives possess by-proxy narrative value. This is already the case for pets, at least for the most part (although euthanasia of pets raises an interesting issue here). But we would have to ensure that it is the case for livestock as well. We would need to think very carefully about the conditions in which an animal’s life can acquire by-proxy narrative value, and we would have to make sure that those conditions do not obtain in farms where animals are killed for food.


A good-faith argument for reform along the lines of the above would require much more than what I’ve said here. I haven’t even really given an argument for the Extended Narrativity Hypothesis; I’ve only tried to explain why that hypothesis ought to be attractive to those who prefer reform rather than abolishment of the animal caste system. But this post is already too long, so I’ll just stop here. What do you think?


  1. Hi David, thanks for this very interesting guest post! A few thoughts...

    (1) You write: "it seems to me that it would be morally wrong for me to kill my dog in order to harvest his organs to be distributed to five other dogs"

    Just to clarify: Does this intuition persist even when the "five other dogs" are also your pets? A consequentialist version of your animal caste system might say instead that we just ought to give extra weight to the interests of our pets (and, to a lesser extent, others' pets).

    (2) If there were an omniscient being (who saw the "life story" of every living thing), would that mean that even livestock would have narrative value by proxy, according to the Extended Narrativity Hypothesis? It doesn't seem to me that knowledge alone turns a creature into one's "pet" (so to speak). So I wonder if the missing element is also what's intuitively doing some important moral work.

    (3) Conversely, it isn't clear to me that pet-owners always construct life-stories for their pets. (Galen Strawson has argued that not all people even construct narratives about themselves.) But it still seems that they should care about their pets in the same way as the rest of us.

    (4) What do you think of the "skeptical" view that we don't really have any special obligations to pets, and the temptation to think otherwise derives from our tendency to anthropomorphize? (Such a skeptic might still accommodate some related character evaluations, e.g. that it would seem to take a distressing callous person to not care especially about cute kittens, etc.)

    1. [David Killoren responds:]

      1. Yes, this intuition does persist when the five other dogs are also my pets, but I feel torn. That case, to me, looks like the case in which I have six human children, five of my children need transplants, and the healthy one is a suitable donor. In either the pet variant or the child variant, it seems to me wrong (pro tanto) to kill the one to save the five, although I feel pulled in the other direction too.

      In connection with this, I'd like to mention a complication that arises from self-sacrifice.

      I think it is permissible for me to sacrifice my life in order to give my organs in order to save five others, to jump on a hand grenade in order to save five others, etc. This view, however, seems to be in tension with the Extended Narrativity Hypothesis. After all, when I kill myself to save five others, it would seem that I destroy some narrative value, which is supposed to be non-fungible. So it seems I am committed to say that killing myself to save five others is wrong (at least pro tanto). But I want to say that killing myself to save five others is not wrong (not even pro tanto).

      Here is one way to resolve the difficulty. Perhaps killing myself to save five others does not actually destroy any narrative value, because my self-sacrifice is part of my self-told story. Killing myself, according to this view, is like an author bringing her novel to an end. If the end of a novel is good, this can actually increase its value, rather than destroy it.

      But this raises a new problem for me. If the person telling the story can end the story without destroying its value, and if (as I've claimed) the pet owner is the person telling the story of the pet's life, then it would seem that a pet owner can kill her pet without destroying the narrative value of the pet's life. In that case, the Extended Narrativity Hypothesis would only explain why it is wrong to kill other people's pets, but would not explain why it is wrong to kill one's own pet. I have some ideas about how to deal with this problem, but I am still not sure exactly what to say.

      Another way to resolve the difficulty is to say that it is only wrong to destroy someone else's non-fungible value. In the case where I sacrifice my own life for greater good, I destroy my own non-fungible value; but in the case where I sacrifice my dog's life for greater good, I destroy his non-fungible value (although it is value that I gave to him by narrativizing his life). I do not like this option (because I want to say that moral value doesn't belong to anyone) but maybe others will.

      [to be continued...]

    2. 2. I really like this thought. I for one am not sure it's so bad to say that God's knowledge of every animal's life would put all animals into the same moral category as pets. I might be willing to bite that (conditional) bullet. But suppose we want to say that mere knowledge about the brute facts concerning an individual is not enough to imbue the individual with narrative value. Then what else is required?

      I want to deny that the missing element is anything like love of the individual, or fondness for the individual, etc., because I want to be able to condemn optimific killing of pets even in cases in which the owner hates the pet.

      By the way, I also want to deny that narrative value can be observer-independent (as I think Jeff Sebo believes) because I want to say that non-pets (e.g., livestock and wild animals) lack narrative value.

      One possibility is that the observer has to appreciate the individual as a candidate for narrative value. This wouldn't solve the God problem, though, because presumably God would appreciate every conscious animal as a candidate for narrative value. So, maybe the God problem is unavoidable and I just have to hope it's not fatal.

      3. Yes, this is a real problem. I am going to have to set a pretty low bar for the construction of a life-story, and I will have to say that pet owners who manage to avoid narrativizing their pets' lives can permissibly treat their pets like livestock.

      4. I feel pretty strongly tempted toward the skeptical view you mention. Its simplicity is very appealing and it can accommodate much of what we want to say. But I think the Extended Narrativity Hypothesis accommodates *more* of the things that really devoted pet owners want to say, so I think we pet lovers should be hoping to be able to defend it and we should consider the skeptical view mainly as a last resort.


  2. Some thoughts I have:

    -The idea that something becomes much more valuable if we attach narrative significance to is a good explanation for the way people regard their bodies. Technically speaking, my body isn't "me," my brain is me, my body is just a large flesh-robot my brain pilots. But people tend to attach great narrative significance to their bodies. This nicely explains why harming someone's body is usually much worse than damaging their physical possessions.

    -The concept of "Narrative Value" reminds me strongly of Derek Parfit's concept of "Global Preferences," where we have preferences for how our life should go as a whole. Like the narrative value theory, Global Preferences allows some non-fungibility of value. In a slightly modified version of Parfit's classic example, if someone tortures you for a brief period of time, then gives you an overwhelming desire to take a drug, and then supplies you with a lifetime supply of the drug, that person has made your life worse, not better. The satisfaction of the newly created desire cannot compensate for the pain of torture, because that new desire is not in line with your global preferences.

    - I'm not sure our intuitions about the organ donor case are caused by the non-fungibility of value. When considering that case I prefer to do what Richard recommended in an earlier post and "naturalize" it (that is, imagine a scenario with the exact same consequences, but where no one is responsible for them, everything happens naturally). For instance, if six people or six pets all get a fatal illness at once, which would be better, if one of them spontaneously recovered, or if five of them did? "Five" seems obviously correct to me, indicating that my intuition in that case is more about moral responsibility or something, rather than about how fungible value is.

    -However, there are other "non-fungibility" cases where my intuitions don't change when they are naturalized. For instance, in "replacement" scenarios I consider it gravely immoral to kill someone, even if doing so allows you to replace them with someone whose life will have just as much utility. If I consider a scenario where the "replaced" person is struck by lightning rather than deliberately killed, it still seems just as bad. So I think you are right that some values are non-fungible, even if I don't like the particular example you chose.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)