Feb 14, 2011

How slippery is the slope?

The slippery slope fallacy is rather slippery. We often make claims about a slippery slope, and also accuse others of making the fallacy of the slippery slope. Which is it, a legitimate argument, or a fallacy?

It depends. We should distinguish several kinds of slippery slope arguments, neither of which is necessarily fallacious.

Let's suppose we're discussing eating dessert before dinner. You are in favor, and I am opposed. There are three kinds of slippery slope arguments I might make.

Tendency arguments: X tends toward Y
(1) Eating dessert before dinner tends to lead to eating lots of sweets.
(2) One shouldn't eat lots of sweets.
(3) Therefore, one shouldn't eat dessert before dinner, since one shouldn't do what tends to lead to something else that one shouldn't do.
This is a pretty reasonable prudential argument. (1) may be disputed, but the most controversial bit is drawing the conclusion (3) from the premises (1) and (2).

Necessity arguments: X always results in Y
(1) Eating dessert before dinner always leads to eating lots of sweets.
(2) One shouldn't eat lots of sweets.
(3) Therefore, one shouldn't eat dessert before dinner, since one shouldn't do what always leads to something else that one shouldn't do.
This argument reverses the weaknesses of the first: (1) is implausible (though potentially true as an empirical matter), whereas the inference from the premises to the conclusion is more sound.

Parity arguments: X is on par with Y

(1) Eating dessert before dinner is on par with eating too many sweets. (That is, if it's acceptable to eat dessert before dinner, then it's also acceptable to eat lots sweets.) 
(2) It's not acceptable to eat lots of sweets.
(3) Therefore, it's not acceptable to eat dessert before dinner, since it's not acceptable to eat lots of sweet and that is on par with eating dessert before dinner.
As with the second argument, the action is all in the first premise. The claim that two action types are on par from the perspective of morality or prudence is a substantive claim, and will likely to be controversial.

These are three varieties of the slippery slope argument. Each variety has some sound instances, and some unsound instances: neither one is always fallacious.

Having diagnosed the structures of these arguments and their distinctive weaknesses, we can see that accusing an interlocutor of making a slippery slope argument is a rather delicate matter. The first thing to do is get clear about which form of the argument the opponent is making. It's very common to mistake Parity arguments for Tendency or Necessity arguments. In those cases, the person making the allegation of a slippery slope is himself committing the strawman fallacy, since those versions of the argument are often weaker. The second thing to do is to pinpoint the disputed premise. For example, one sometimes hears the following argument against abortion:

(1) Abortion is just like infanticide. (Or: If abortion is acceptable, then so is infanticide.)
(2) Infanticide is wrong.
(3) Therefore, abortion is wrong.

This is an example of a Parity argument. While some philosophers dispute (2), the allegation of a slippery slope here is about the first premise. Once we identify that premise as controversial, the real work starts. (1) is a substantive moral claim. To rebut it, one points to features that distinguish abortion from infanticide. Once we start down that road, we are able to have a much more productive discussion than if we had simply rejected the argument as committing a slippery slope fallacy. Whether it does so is a substantive question.

So, the slippery slope fallacy is a slippery beast, and is perhaps best not thought of as a fallacy at all.


  1. Hi Matt,
    Interesting clarification. I'd started writing a long comment which for some reason could not be posted and is now lost in cyberspace.

    Don't you think (2)-premises may as importantly be disputed in countering slipper slope fallacy (SFF) allegations? I mean, those who're taxed with committing SSF are often those who'd claim that (2) is not only disputable but clearly wrong or dogmatic *and* that, paradoxically, they agree with the allegation that (1) is true. That is, for instance, infanticide may not be as wrong as is commonly thought precisely because, in some circumstances and under specific descriptions, it is quite similar to abortion.

    Of course, this does not rule out substantial disputes w.r.t. (1), e.g. is a fetus a person, assuming every infant is one; or does a fetus stand in all the same relevant relations to the moral community as every infant; etc. But, if the concept of person is revised so as to cover some non-humans without necessarily covering all humans, then it might morally make sense to assert (1), while disputing (2). And it might as well be the case that, assuming (2), some may soundly make parity arguments without committing a fallacy, and these arguments could in fact be made by those who, on your interpretation, make the allegation of SSF.

  2. Two additional points taken from McMahan's "Ethics of Killing".

    A) See p. 454-455. There's an argument for extending the use of adult stem cell to grow organs (for medical research purposes or to save a person's life). Following a tolerance principle (not unrelated to a vagueness problem of identity between a set of body parts and the body itself, especially in transplantation cases), it seems that,

    if (1) it is permissible to grow one organ;
    then (2) it is permissible to grow two organs;
    if (2),then (3) it is permissible to grow three organs;
    if (n-1),
    then (n) it is permissible to grow n organs, where n is the adequate number of body chunks whose make up an adequate description of a body.

    Does this argument from vagueness rest on a slippery slope? Or, are SSF related to vagueness problems?

    B) See p. 486. McMahan reverses the charge of SSF against the argument from the possibility of abuse that would commend forbidding killing in cases of self-defense. The SSF, McMahan argues, can be averted provided we erect safeguards, that is, if we know where to draw a line between permissible and impermissible killings.

    This line, it seems to me, may help rebut parity (1)-premises as well as tendency ones in the arguments you mention. That is, the parity assertion, as you rightly point out, begs a question, namely, that of permissibility. But we may equally be tempted to dispute (2)-premises on the same ground.

    Thus McMahan does not reject the idea that extensions of permissible killings have tendency-like slippery slopes. But: (i) they're not, as such, fallacies, since "possibilities of abuse" may be relevant to permissibility, hence to infering (3) from (1) and (2); (ii) it could be agreed by both camps that slippery slopes do not count *against* the arguments that are alleged to induce them, but *for* the erection of safeguards.

    Interestingly, McMahan has a different argument against SSF allegations:

    "It seems, moreover, that resistance to euthanasia has its own slippery slope. If those who advocate euthanasia are in danger of becoming less sensitive to killing, it seems equally true that those who seek to deny others the option of euthanasia are in peril of becoming inured and desensitized to the suffering of others and more willing to tolerate it in all areas of life." (486)

  3. Oops: "whose ADDITION make up an adequate..."

  4. Nicolas, thanks for your comments, you raise some very interesting points.

    I think I agree, as you say in your first post, that in some cases the response to the abortion argument (for example) will target (2) rather than (1). That's a good point.

    Your discussion of McMahan is quite interesting as well. Off the cuff, I think I'd tend to disagree that so called "possibilities of abuse" are relevant to permissibility, but that's a somewhat delicate issue. McMahan's tu quoque regarding a slippery slope for euthanasia opponents is an interesting point that I hadn't thought of before.