Here's stereotype of how the dialectic goes. Quid tells Quo that she should give up meat, and when Quo says "I'd rather not; I like meat", Quid tells her that it's cruel to slaughter your fellow creatures for your own benefit (especially when the benefit is so trivial as getting a meal that you like better). After all, nobody would think that it's O.K. to kill and roast your fellow humans, even if it turns out that roast man-flesh is a truly delicious meal. Quo isn't convinced; she argues that slaughtering humans and slaughtering animals is different, and when Quid demands to know why, Quo remarks that humans have some distinctive mental capacity that gives humans a direct moral standing that animals, like plants and rocks, don't have. (Which property Quo picks isn't relevant here. Common candidates include self-consciousness, reflective reason, abstract thinking, moral agency, and some other stuff.)
Quid comes back with the famous "Argument from Marginal Cases". There are a lot of different ways to gloss the argument, depending in part on the details of the pro-meat argument that it's responding to, but here's a schematic gloss: pick any mental property that you like that might explain a difference in moral standing, and tag it Morally Relevant Rationality (MRR) for convenience. No matter what property you pick, one of two things will be true. Either (1) all humans will have MRR, but many or most animals will have it too (e.g.: responsiveness to pain, formation of desires), in which case MRR will be too broad to justify slaughtering animals; or else (2) no animals will have MRR, but there will be at least some humans who don't have it either, in which case MRR will be too narrow to live up to our expectations of a rational ethical theory. Horn (1) of the dilemma is clear enough; to account for horn (2), Quid points out that for pretty much any distinctive mental property that you can find in paradigmatic cases of humanity (healthy adults with no congenital defects and a normal upbringing), there will be at least some "marginal cases" -- infants, the comatose, people with brain lesions, the severely mentally retarded, feral children, etc. -- who don't have it now, or lost it, or never had it to begin with. But any ethical theory that entails that we could ethically slaughter infants, the comatose, people with brain lesions, the severely mentally retarded, feral children, etc. for food is monstrously wrong. Therefore there are no reasonable candidates for MRR that do the work anti-vegetarians want them to do. Therefore, put down that steak.
The argument's attractive because, among other things, it saves you from the hard philosophical work of having to respond to each concrete suggestion for MRR, or having to engage with the specific arguments for the connection between MRR and moral standing. After all, if the dilemma really does cover all plausible candidates then Quid can use the argument as a schematic for a response to any candidate for MRR and any account of the connection between it and moral standing. The problem, though, is that there's a perfectly good response to the Argument from Marginal Cases on the record -- the call it Argument from Species Normality -- and, as far as I know, ethical vegetarians in the literature haven't yet successfully responded to it. In fact, as far as I know, ethical vegetarians in the literature haven't even understood it.
Here's how the response goes. Quo concedes that Quid has pointed out a genuine difficulty. But, she says, there's a way out. Here's how. Take horn (2) of the dilemma, and choose some mental property that paradigmatic humans have but animals don't. It's true that not all humans have that property. But there is another, closely related property that all and only humans (including all "marginal cases") do have: each and every one is human. This may seem trifling or crude, but suppose that Quo goes on to point out that being human means (among other things) being a member of a species whose paradigm cases have MRR, that is, one of the kind of creature for whom it is normal to have MRR. So even though there will be humans who individually lack MRR, Quo contends that enjoying the human form of life -- even if, in a particular case, the mental faculties that are involved in that form of life are undeveloped or frustrated or damaged (perhaps irreparably) you are obligated to treat them differently from the way you would treat an animal; for animals the place occupied by MRR isn't empty or inaccessible; it just doesn't exist.
The canonical vegetarian response to the Argument from Species Normality seems to be to misunderstand it. One way to misunderstand it is to bring out the doctrine ethical individualism, as if it explained anything; the idea is that "being a member of the same species as x, y, and z" is a pure Cambridge relation; why should that external relation have any burlier ethical consequences than "having the same name as x" or "being the third person born after y." Why should having these kind of "properties" have any ethical bearing at all? Wouldn't making ethical distinctions based on them be a case of arbitrary (and therefore unjustifiable) group privilege? (People who like ghastly neologisms sometimes call this "speciesism," by analogy with "racism" and "sexism.")
But this is a mistake, and I think it's a mistake that's no less crude for being so common. Making it may be the result of making some other mistakes about the logical form of statements about living creatures and the natural kinds they belong to, of the sort discussed by Michael Thompson in The Representation of Life; as Thompson shows, philosophers tend to try to understand statements about living creatures and their distinctive forms of life with a much too narrow idea of what sorts of logical features statements can have, and they tend to systematically get things wrong by excluding, or ignoring, the kinds of teleological facts that Thompson calls "aristotelian categoricals." For example, consider the natural-historical statement, "the domestic cat has four legs and a soft coat of fur." This is a true general statement. And it is not to be refuted by pointing "But look at poor Tibbles, who has been shaved and maimed in a tragic accident!" The reason that it is true is not that all domestic cats have four legs and a soft coat of fur (poor Tibbles doesn't). Nor is it that some or many or most or even the overwhelming majority of domestic cats have these features. (Even if some mad scientist released a virus that caused all the domestic cats in the world to permanently shed their fur coat, "the domestic cat has a soft coat of fur" would still be true.) It's that cats are the kind of creature that has four legs and soft fur under normal conditions, even if tragic circumstances have altered those conditions for some cats, or even for all cats temporarily. (If they altered conditions for all cats permanently, that might constitute a change in the form of life for cats as such. But that's not important here.)
These distinctions make a difference. In particular, it's important to see how aristotelian categoricals make a difference for what we can say about individual members of the species. They make a difference with respect to value, in obvious cases. If Tibbles has two legs, that's a tragedy; it's something wrong with poor Tibbles. If I have two legs, that's normal; that's how humans are. The fact that we recognize my two-leggedness as normal and Tibbles' two-leggedness as a defect is tied up with the fact that I'm supposed to have two legs and that Tibbles is supposed to have four. And that's a biological fact about Tibbles herself, not just a Cambridge relation to other members of her species. It goes similarly with humans who suffer from a cognitive defect. The point of the appeal to normality isn't relational or statistical; the point is to show how all human beings, each and every one of them, individually has a particular intrinsic property. That property isn't just sharing a species with other humans who do have MRR. It's having, individually, a faculty for MRR (whatever that may be), and we're supposed to be able to exercise it, even if in particular cases that faculty is not yet developed, or inactive, or frustrated, or irreparably damaged. A human being that can't comprehend language or engage in reflective reasoning has a something wrong with her (that's what calling it a "disability" or a "defect" means); a pig that can't comprehend language or engage in reflective reasoning is just living how the pig lives.
(Actually that's not quite right. An infant that can't understand language or reflectively reason isn't abnormal or defective. But that's no more difficult to deal with than the fact that people who are sleeping don't exercise MRR while they're unconscious. The normal condition in the human form of life is that infants will develop MRR over time. Not so pigs.)
I think this is also closely related to the other common response to the Argument from Species Normality -- what we might call the Argument from Freak Intelligence. Here's how James Rachels put it (in "Darwin, Species, and Morality", Animal Rights and Human Obligations, Tom Regan and Peter Singer, eds., p. 100):
This idea--that how individuals should be treated is determined by what is normal for their species--has a certain appeal, because it does seem to express our moral intuition about defective humans. "We should not treat a person worse merely because he has been so unfortunate," we might say about someone who has suffered brain damage. But the idea will not bear close inspection. Suppose (what is probably impossible) that a chimpanzee learned to read and speak English. And suppose he eventually was able to converse about science, literature, and morals. Finally he wants to attend university classes. Now there might be various arguments about whether to permit this, but suppose someone argued as follows: "Only humans should be allowed to attend these classes. Humans can read, talk, and understand science. Chimps cannot." But this chimp can do those things. "Yes, but normal chimps cannot, and that is what matters." Is this a good argument? Regardless of what other arguments might be persuasive, this one is weak. It assumes that we should determine how an individual is to be treated, not on the basis of its qualities, but on the basis of other individuals’ qualities. This chimp is not permitted to do something that requires reading, despite the fact that he can read, because other chimps cannot. That seems not only unfair, but irrational.
I think Rachels is obviously right here; it would be wrong to treat the hyperintelligent chimp that way on those grounds. (A fortiori, it would also be wrong to slaughter a hyperintelligent pig and eat it. The fact that normal swine are not as intelligent as human beings wouldn't be a reasonable ground for denying it a right to life.) Species membership isn't a good grounds to make a distinction in cases of freak intelligence. But he's just wrong to draw the conclusion that the case of humans with cognitive defects has to be symmetrical with the case of animals with freak intelligence. Perhaps the problem is that Rachels is still thinking of the appeals to species or kind as if they were just appeals to membership in a set with some particular defining characteristic. If that's all that was at stake, then trying to make an ethical distinction based on it surely would be arbitrary. But that's nota all; the Argument from Species Normality gains whatever force it has by appealing to faculties or potentialities that each individual humans, as a human, has. Think of it this way: a pig with freak intelligence has, ex hypothesi, manifested MRR. Actuality entails potentiality, so the pig has the faculty for MRR and it makes sense to demand that it get the same moral level of consideration you give to your fellow humans (whatever the right level for that is). But that does not entail that you also have to demand that animals with normal cognitive abilities for their species get the same level of consideration you give to your fellow humans with severe cognitive defects (whatever the right level for that is). Non-actuality doesn't entail non-potentiality, and you have to distinguish between the cases where a faculty is present but unexercised, damaged, frustrated, undeveloped, etc. and those in which there isn't any faculty to lament the damaging of at all. And one of the reasons that you would give for making distinctions of this sort just is that there's a difference between a healthy adult pig with the cognitive abilities of a pig and an adult human with the cognitive abilities of a pig. If you aren't approaching the world of life with a rich enough conceptual framework to recognize these kind of teleological facts, then you probably need to enrich your conceptual framework before you can sensibly deal with the notion of goodness at all.
Now, is it true that the distinctively human faculty for MRR, whatever that is, even if it's not being exercised, and even if it can't be, in a particular case, really does make for a burly difference in moral standing between humans and animals? Probably, but is it enough of a difference to justify slaughtering and eating the animals even though you'd never consider treating "marginal case" humans that way (and would rightly be punished harshly for doing so)? I doubt that it is true. But I think to give a good reply to the Argument from Species Normality, you'll need some kind of argument that specifically engages with the details of the particular faculty that's suggested to play the role of MRR, and the details of the account that's given to connect it to moral standing, in order to show that it doesn't make enough of a difference, or doesn't make the right kind of difference, that's needed to justify the incredible suffering inflicted by the meat industry. Trying to avoid the messy part of the argument by skipping over the details and making an appeal to marginal cases just won't get you anywhere that you should want to go.