Once You Start Using Slippery Slope Arguments, You re on a Very Slippery Slope

Transcription

1 Oxford Journal of Legal Studies, Vol. 21, No. 4 (2001), pp Once You Start Using Slippery Slope Arguments, You re on a Very Slippery Slope DAVID ENOCH Abstract Slippery slope arguments (SSAs) are, so I argue, arguments from consequences which have the following peculiar characteristic: They take advantage of our being less than perfect in making and acting according to distinctions. But then, once SSAs are seen for what they are, they can be turned against themselves. Being less than perfect at making the second-order distinction between distinctions we re good at abiding by and those we re bad at abiding by, we re bound to fail to make the distinction between good and bad SSAs. One can therefore construct an SSA, the conclusion of which is, that we ought not to use SSAs. After characterizing SSAs and constructing the SSA against the use of SSAs, I then explore its implications. 1. Introduction Despite the major role slippery slope arguments (SSAs) have always played in prudential, moral, political and legal argumentation, they have a bad philosophical reputation. Some texts even refer to them as slippery slope fallacies, or the slippery slope fallacy. Lately, however, philosophical discussions of SSAs have been more sympathetic. Contemporary texts dealing with SSAs either in general or with a specific SSA are rather ambivalent: SSAs are taken to be easily misunderstood and abused, and generally not very strong, but not necessarily bad or fallacious. In what follows, I will not take issue with this very general and not very interesting ambivalence towards SSAs. Rather, I will raise a problem for what I take to be the best version of SSAs. I want to show that a perfectly reasonable SSA can be constructed against the use of SSAs, and that there is therefore an air of self-defeat about the use of SSAs. In section 2 I characterize the kind of argument I will be discussing, and explain why it may be a reasonably good argument. Seeing that many different NYU Philosophy Department. For helpful conversations and comments on earlier drafts I thank Or Baron, Yaacov Ben-Shemesh, Theresed Bjorkholm, Oren Gazal, Pete Graham, Alon Harel, Andrea Iacona, Frances Kamm, Tziporah Kasachkoff, John Kleinig, Derek Parfit, Michael Ridge, William Ruddick, Josh Schechter, Brad Skow, Maya Steiniz and a reviewer for the OJLS. Earlier versions of this paper were presented at the Law Schools of Haifa University, Tel Aviv University and the Hebrew University. I thank the participants for the helpful discussion Oxford University Press

2 630 Oxford Journal of Legal Studies VOL. 21 arguments are sometimes called SSAs, clarifying the kind of argument I will be discussing requires distinguishing it from others. In order to spare the reader who is not particularly interested in the relevant literature the tedious comparative work, and in order to keep the flow of the main argument, I do this comparative work in an appendix. In section 3 I put forward my SSA against using SSAs, the implications of which are then discussed in section 4. In a concluding remark I comment on the applicability of my argument and of SSAs in general to politicians and philosophers. 2. Slippery Slope Arguments It has often been noticed that the term slippery slope argument is very frequently misused, and that not every argument that is presented as an SSA really is an SSA. 1 Furthermore, as will become clearer later in this section and in the appendix, SSAs purportedly genuine SSAs have been divided in the literature into many different kinds. It is therefore necessary to start by characterizing the kind of argument I will be raising an objection to in the next section. The characterization shortly to be given can be treated by the sceptical reader as a stipulation: it is this kind of argument that I will label SSA and raise an objection to, whether or not this is what other people have in mind when they think of SSAs. But I don t think it must be treated as a stipulation. If SSAs form an interestingly distinct kind of argument, and if they are not ridiculously bad, I think the following is the only way to understand what it is for an argument to be an SSA. Perhaps it will be helpful to start with an example. Take a version of the wellknown moral SSA against legalizing voluntary euthanasia. The proponent of this SSA concedes be it just for the sake of argument that allowing voluntary euthanasia is, in itself, morally unobjectionable. 2 The problem the proponent of the SSA is worried about is that allowing voluntary euthanasia may (or will) lead to the allowing of non-voluntary euthanasia, and that this in turn will lead to the allowing of involuntary euthanasia; this last action allowing involuntary euthanasia is taken to be morally unacceptable. This causal process is put forward as a reason not to allow even voluntary euthanasia, not to take the first step down the slippery slope. Why, though, should we believe that allowing voluntary euthanasia would result in allowing involuntary euthanasia? After all, the former is ex hypothesi morally permissible (ignoring the SSA itself) and the latter is impermissible; 1 E.g., G. Den Hartogh The Slippery Slope Argument, in H. Kuhse, and P. Singer (eds.), A Companion to Bioethics (1998) at 289; B. Williams Which Slopes are Slippery?, in B. Williams, Making Sense of Humanity (1995) at 214; D. Walton, Slippery Slope Arguments (1992); D. Lamb, Down the Slippery Slope Arguing in Applied Ethics (1988) at viii. 2 She may, of course, also have other reasons to object to allowing voluntary euthanasia, but if her SSA is to give any additional reason not to allow voluntary euthanasia, it must not be merely parasitic on other reasons. She must think, then, that even if there were no other reasons not to allow voluntary euthanasia, the SSA-reason would count against allowing it. For simplicity, then, in the text I assume that she concedes that allowing voluntary euthanasia is not in itself objectionable.

3 WINTER 2001 Slippery Slope Arguments 631 surely we are going to make this distinction and avoid crossing the line between the permissible and the impermissible, aren t we? The proponent of the SSA is not convinced. She mistrusts our ability to make the proper distinctions. Once we allow voluntary euthanasia, she argues, we may (or will) fail to make the crucial distinction, and then we will reach the morally unacceptable outcome of allowing involuntary euthanasia; or perhaps even though we will make the relevant distinction, we will not act accordingly for some reason (perhaps a political reason, or a reason that has to do with weakness of will, or some other reason). I will use failing to abide by the distinction to stand for either failing to make the distinction, or making it but failing to act accordingly. The proponent of the SSA against voluntary euthanasia argues, then, that we are likely to fail to abide by the distinction between voluntary and involuntary euthanasia. I will call the (typically implicit) assumption that we are bad at abiding by the relevant distinction, an assumption soon to be explained in some detail, the Essential Premise. Generalizing, an SSA is an argument against an action (A) that is not in itself objectionable, based on the claim that performing it will lead to the performance of (B, which will lead to C, which will lead to the performance of... ) some other objectionable action (Z), because we are liable to fail to abide by the proper distinction between A and Z. 3 That the Essential Premise is something every proponent of an SSA is committed to can be easily shown: for Z, remember, is taken to be prima facie (even before this argument, that is) objectionable, whereas A is not; so there is a normative distinction between A and Z; and so if doing A may nevertheless lead us to do Z, we must be less than perfect at abiding by the relevant distinction, namely, the distinction between the objectionable and the unobjectionable. And that is all the Essential Premise says. SSAs, though, do not just presuppose the Essential Premise. Rather, the fact stated by the Essential Premise plays a causal role in the slippery-slope-process: when the causal story of how A ing leads to Z ing is told, a part of it is going to be that we have failed to abide by a relevant distinction. It is because the Essential Premise is true that A ing will lead to Z ing. To see this, notice that if it is argued that A ing may lead to Z ing by some other, non-standard, causal mechanism if, say, A ing will cause a button to be pressed, which will then cause our Z ing then we clearly don t have an SSA, but rather a non-slippery-slope argument from consequences (more on the distinction between SSAs and other arguments from consequences in the appendix). This is, then, the characteristic feature of SSAs: an otherwise unobjectionable A will lead to an objectionable Z because we are bad at abiding by distinctions, and in particular at abiding by the distinction between A and Z. 3 There is no reason why SSAs cannot be used to argue for an action that is otherwise objectionable, because performing it may increase the likelihood of our performing some other desirable action. But in what follows I will be assuming that SSAs are as they typically are arguments against certain actions. Nothing of importance will hinge on that.

4 632 Oxford Journal of Legal Studies VOL. 21 Notice that the Essential Premise does not state that we have no reason to distinguish between A and Z indeed, that we do have such a (prima facie) reason has been taken to be true ex hypothesi. Rather, the Essential Premise states that we cannot or are not likely to abide by the relevant distinction. In the terms Williams 4 uses, the Essential Premise denies an effective distinction (one that we are likely to abide by) midway down the slope, not a reasonable one (one that we have reason to abide by). The metaphor of a slippery slope may be of help here: what is distinctive of slippery slopes is not that we do not have a reason to stop midway down the slope indeed, very often we do have such a reason; rather, it is that, limited creatures as we are, we cannot (or are not likely to) do so. The Essential Premise states, again, that we are bad at abiding by the relevant distinctions. It does not have to though it might say we cannot abide by the relevant distinctions. Suffice it for it to be the case that we may fail to abide by the relevant distinctions, though the more probable it is that we will in fact fail, the better the SSA. Put generally, then, and other things being equal, the worse we are at abiding by the relevant distinctions, the better the relevant SSA. Who are we, the ones the Essential Premise says are less than perfect distinction-abiders? The answer may change, of course, with context. What can be said generally is that the proponent of an SSA mistrusts the distinctionabiding abilities of whoever it is who is likely to make the decision whether or not to Z. The relevant we may, then, refer to the relevant society (once we allow some intentional killings of innocent people, public opinion is going to fail to abide by the distinction between otherwise unobjectionable killings like voluntary euthanasia and objectionable ones like involuntary euthanasia), to the next generation (a generation raised in a society where some intentional killings of innocent people are allowed is not going to be able to distinguish between otherwise unobjectionable and objectionable ones), to a given institution (once we allow some intentional killings of innocent people, the courts are not likely to successfully distinguish between otherwise unobjectionable and objectionable ones), or even to oneself (I do not trust myself to abide by the relevant distinction). 5 Although putting forward an SSA involves questioning the distinction-abiding abilities of someone, then, it does not necessarily involve arrogance: those whose distinction-abiding abilities are questioned may include the proponent of the SSA herself. 6 4 Above n 1. 5 On occasion, Schauer seems to think that this is impossible (F. Schauer Slippery Slopes, 99 Harvard L Rev at 373 (1985)): I see the relevant distinction, and so there is no reason for me not to trust my distinctionmaking ability, at least with regard to this distinction. But this is false (and Schauer takes it back later on ibid at 374): I may have reason to believe that I will not make the relevant distinction when I ought to even though now I do: distinction-making abilities may change with time and circumstances. Furthermore, I may fail to abide by the distinction (even if I make it) because when the moment comes I will be weak-willed; and so on. Nevertheless, Schauer is right in drawing attention to the fact that there is something essentially third-personal about SSAs. Even when I mistrust my own discriminating faculties, I view myself, as it were, from the third-person perspective. 6 For a nice discussion of who is being arrogant with regard to whom when a lawyer uses an SSA in court, see Schauer, above n5at374.

5 WINTER 2001 Slippery Slope Arguments 633 From what has thus far been said it follows that SSAs rely heavily on empirical evidence. Whether or not we (whoever the relevant we are) are bad at abiding by the relevant distinction, how bad we are at doing so, and how likely it is that given the Essential Premise A will lead to Z, are empirical questions; we are much better at abiding by some distinctions than by others, and the way to find out how likely we are to fail to abide by a specific kind of distinction is to engage in empirical psychological (or sociological, as the case might be) research. As has been widely noticed in the literature, then, SSAs are essentially empirical, 7 and so for an SSA to be plausible, the empirical premises it relies on must be plausible. This in itself is no flaw as in all arguments from consequences, of which SSAs form a subset, empirical premises are needed in order to establish a moral judgment. The observation that SSAs are essentially empirical is important, however, because it follows that SSAs that are not supported empirically are very weak indeed. In order to properly use an SSA, the proponent of the argument must do more than just present the argument; she must give us reason to believe that we really are likely to fail to abide by the relevant distinction, and that it is therefore likely that the bad outcome Z will follow. Are SSAs thus understood good arguments? Do they give reason not to do A? Seeing that SSAs are essentially empirical, the SSAs that are better supported empirically are better arguments, they give stronger reasons other things being equal not to do A. Some SSAs are bad ones in which the relevant instance of the Essential Premise is highly questionable, or at least not at all supported and others are fairly good. What should be clear, though, is that there is nothing fallacious about SSAs as understood here. 8 Some SSAs may be very good indeed, depending on the empirical support the relevant instance of the Essential Premise enjoys, and on how bad the dreaded outcome compared to the relevant alternatives is. To see that, it may be helpful to briefly consider the other side of SSAs, the method that may be called the device of stages. 9 Suppose I think that Z is a justifiable action, but I also know that most people disagree. Other things being equal, would it be rational 10 for me to be pleased with a decision by the majority to do A? Would it be rational for me to try and get them to do A? An example might help: in the bio-ethical context, an SSA is sometimes put forward as a reason not to engage in genetic research, because when the crucial moment comes, we are likely to fail to abide by the distinction between permissible and impermissible genetic manipulation. Let us suppose, then, that I think all genetic manipulation 7 See, for example, Schauer, above n 5 at ; J. Rachels, The End of Life (1986) at 173; Williams, above n 1 at 218; Den Hartogh, above n1at For similar claims, see W. van der Burg The Slippery Slope Argument, Ethics 102 (1991), at 59; G. W. Trianosky Rule-Utilitarianism and the Slippery Slope, The Journal of Philosophy 75 (1978) at 415; Walton, above n 1, ch 5; T. Govier What s Wrong with Slippery Slope Arguments?, Canadian Journal of Philosophy 7(2) (1982) ; Rachels, above n 7 at 173; Lamb, above n 1. A reminder: very different arguments are also sometimes called SSAs, and those may very well be fallacious (I discuss some of these in the appendix). The references in this note are to places where the authors argue that arguments very similar to how SSAs are characterized in the text are not necessarily fallacious. 9 See Walton, above n1at11,21 2 (and the references there). 10 In this paragraph I m assuming a means-end conception of rationality.

6 634 Oxford Journal of Legal Studies VOL. 21 is morally permissible, and that considering the possible benefits of genetic manipulation it should be legally permissible as well. Would it be rational for me to be pleased with a decision to engage in at least some genetic research, or to engage in negative (but not positive) eugenics? 11 Indeed, may it not be rational for me to try and use the device of stages, that is try to convince people first about the permissibility of theoretical genetic research, in the hope that it will then be easier to convince them that negative eugenics is also permissible, and then that so is positive eugenics? Whether or not this would be rational depends, of course, on the empirical evidence is it really likely that once convinced that negative eugenics is morally unobjectionable, people will more easily be convinced that so is positive eugenics? But I think it is perfectly clear that at least sometimes indeed, very often the device of stages is a good strategy. And whenever the device of stages is a good strategy for the proponent of Z justifying doing A first, an SSA is a good argument for the person objecting to Z, justifying not doing A. 12,13 Seeing that the device of stages is very often a reasonably good strategy, SSAs are very often reasonably good arguments. This is why raising a difficulty for them may be interesting. It is to doing that that I now turn. 3. A Slippery Slope Argument Against Using Slippery Slope Arguments SSAs, we ve seen, are, other things being equal, as good as we are bad at abiding by distinctions. And a specific SSA is as good as we are bad at abiding by the relevant distinction. Of course, we re better at abiding by some distinctions than by others, we re better at abiding by distinctions in some circumstances than in others, and some people and institutions are better at abiding by distinctions than other people and institutions (and, naturally, these factors may also interact there may be circumstances in which we re better at abiding by some distinctions but not others, and so on). It therefore follows that some SSAs are better than others. 11 There is a complication here: It may be rational for me to be pleased with such a decision, because of other direct benefits of negative eugenics, regardless of whether or not it will lead to positive eugenics as well. The question in the text is meant to bypass this difficulty: assuming the intermediate stages are of no value to me unless and insofar as they lead to the final stage, is it not rational for me to be pleased with such a decision? 12 Such a connection between the device of stages and SSAs has been noticed by Colwell (G. Colwell Slippery Slopes, Moral Slides and Human Nature, Informal Logic 17(1) (1995) 43 66, at 54) and Walton, who writes (referring to the device of stages this time as the gradualistic argument ): The slippery slope bears the same relationship to the gradualistic argument as modus tollens bears to modus ponens. (Walton, above n 1 at ) 13 The sentence in the text needs qualifying: if we have generally, or in a specific case reason to believe that as a result of a gradual process the unenlightened are more likely to become enlightened than the enlightened to become unenlightened, then the whenever in the text fails. For then, the device of stages may be rational for a proponent of Z (hoping that the gradual process will help enlighten the public as to the justifiability of Z) but an SSA may not be a good argument for the person objecting to Z, because what such an SSA will rely on is the danger that the public will in time fail to make the distinction that is to be made. Even when qualified according to this note (that is, when qualified such as not to apply to the kind of cases just described), though, the point in the text is sufficient to show that SSAs are often good arguments. I thank Frances Kamm for pressing me on the point discussed in this note.

7 WINTER 2001 Slippery Slope Arguments 635 Using an SSA in a case where we are very bad at abiding by the relevant distinction is in itself, let me concede at least for the sake of argument, unobjectionable. There is, after all, nothing fallacious about this kind of argument. Let A, then, stand for using SSAs when we re very bad at abiding by the relevant distinction. Using an SSA in a case where we are rather good at abiding by the relevant distinction is, of course, objectionable: for the relevant instance of the Essential Premise is false. 14 Let Z stand for using SSAs where we re rather good at abiding by the relevant distinction. And it seems reasonable to conjecture that there is a continuum of distinctions we are more and less good at abiding by, a continuum of circumstances in which we are more and less likely to abide by the right distinctions, and so forth. Let us use B, C, and so on in the obvious way standing for using SSAs where we re increasingly better at abiding by distinctions. How good are we at abiding by or even making the second-order distinction between distinctions we re good and those we re bad at abiding by? Can we trust our ability to distinguish between A and Z? This, of course, is largely an empirical question (more on that shortly), but for now let me conjecture that we re not at all good at abiding by such distinctions, mostly because we are bad at even making them. We are likely indeed, perhaps we are even bound to fail to make the distinction between distinctions by which we re bad at abiding and those by which we re good at abiding, and so we are likely to fail to make the distinction between good and bad SSAs. 15 Furthermore, as the discussion of some of the literature in the appendix makes clear, we are likely to fail to make the distinction between good SSAs and bad arguments that are not even genuine SSAs. Using good SSAs A is thus likely to lead to using bad ones (and other bad arguments) Z. It follows that although using good SSAs is in itself unobjectionable, we should not use them, because liable as we are to fail to distinguish between good and bad SSAs using good ones would lead to our using bad ones. We ought not to use SSAs, the argument concludes, even when they are good ones. This argument, of course, is an SSA: it uses an instance of the Essential Premise namely, that we are bad at making (and abiding by) distinctions between distinctions by which we re good at abiding and ones by which we re 14 The plausibility of the Essential Premise is not the only factor determining the quality of an SSA. The other relevant factor is how bad Z compared with the consequences of alternatives to A is. And there may be SSAs where Z is so much worse than the consequences of not A ing, that even if the relevant instance of the Essential Premise is not very plausible, and so it is not very likely that if we A we will Z, it would still be better not to A. In the text I for the most part avoid mentioning this complication; it should be understood as implicit. 15 Given the complication discussed in the previous note, the sentence in the text should be understood as slightly weaker: how good we are at abiding by the relevant distinction is a factor determining the goodness of the SSA, and so being bad at distinguishing between distinctions by which we are good and bad at abiding, we are likely to fail to abide by the distinction between good and bad SSAs. Of course, assuming we are very good at predicting the badness of alternative actions and that very often Z ing is much worse than the consequences of not A ing, we may be good at distinguishing good SSAs from bad ones even though we re not good at distinguishing between distinctions we re good at abiding by and those we re bad at abiding by. It is therefore a substantive largely empirical premise of my argument that we are not very good at distinguishing between SSAs where Z ing (compared to the alternatives) is bad enough to compensate for the possible implausibility of the Essential Premise, and where it is not. I thank Alon Harel for pressing me on this point.

8 636 Oxford Journal of Legal Studies VOL. 21 bad at abiding in the role this premise characteristically plays in SSAs, namely, it plays a part in the causal story to be told in explaining why doing A (using good SSAs) will or may lead to doing Z (using bad SSAs). The argument in the previous paragraph is, then, an SSA the conclusion of which is that we ought not to use SSAs. In the next section I discuss some implications of this apparently self-defeating nature of SSAs. Before I do that, though, two further points are in order. First, as already mentioned, SSAs are a special kind of argument from consequences. As in all arguments from consequences, drawing attention to the bad outcomes of one course of action is not enough; one has to show that the alternative courses of action don t have just as bad (or even worse) consequences themselves. It is in doing this balancing that how bad the bad consequences the SSA anticipates are is of importance. In the SSA against voluntary euthanasia, for instance, the bad consequences supposed to follow are extremely bad, and indeed proponents of SSAs in general try to paint as dire a picture of the consequences of the disputed action as possible. What, then, are the bad consequences my SSA against using SSAs warns against? What would be the consequences of using bad SSAs, and how bad would these consequences be? First, of course, using bad SSAs will cause our having what may very well be false beliefs we will believe the conclusions of a bad argument, erroneously thinking that it is a good one. That in itself may be considered a bad outcome (though perhaps not as bad as some other outcomes often warned against by proponents of SSAs). But, more than that, using bad SSAs will lead to our acting immorally, imprudently and irrationally. If, for instance, the SSA against voluntary euthanasia is a bad one (because we re actually rather good at abiding by the distinction between voluntary and involuntary euthanasia), then failing to notice that it is a bad argument might lead to the unnecessary suffering and humiliation of numerous terminal patients who would rather die than go on suffering. The bad consequences of using bad SSAs are very severe: avoiding all the good things we could do once noticing that (bad) SSAs do not give reasons not to do them. As some of these bad consequences are morally bad, the argument above may very well support the conclusion that we morally ought not to use SSAs. 16 Even more than that, however, may be at stake. In the political sphere, SSAs are usually used by those trying to defend the status quo. 17 Furthermore, SSAs are often used to hinder further discussion of what may be very important issues, for it is often argued that allowing a discussion to be conducted is already the 16 It may be worth emphasizing that the morally bad consequences my SSA against SSAs warns against do not merely involve the avoidance of doing some good; as the example in the text shows, they may very well involve a positive wrong-doing. Not respecting the requests of autonomous creatures regarding their own lives (or deaths) thereby causing them great indignity is at least if the SSA against voluntary euthanasia fails, and assuming there are no other reasons not to allow euthanasia acting wrongly. I thank Frances Kamm for pressing me on this point. 17 Lamb, above n 1, at vii, notices this point. Den Hartogh (above n 1 at 281) thinks this is even a necessary characteristic of (at least paradigmatic) SSAs. For reasons I do not need to specify here, this seems to me wrong. But it is certainly clear that SSAs are usually used to defend the status quo. Jeremy Bentham already noticed this characteristic of SSAs (which he refers to as Fallacy of Distrust, or What s at the Bottom? ) some two centuries ago (J. Bentham in H.A. Larrabee (ed.), Handbook of Political Fallacies (1952) at 100).

9 WINTER 2001 Slippery Slope Arguments 637 first step down the slope. 18 SSAs, then, are used not just in order to justify a conservative conclusion in a discussion, but also in order to justify not holding (and not allowing) a discussion in the first place. The dangerous result of using bad SSAs, then, may be letting the defenders of the status quo win the day by preventing further discussion of contested issues. I take this to be a serious political danger. Thus, my SSA against using SSAs does warn against seriously bad consequences. It should therefore be taken seriously. The second point mentioned earlier has to do with the empirical basis for my SSA against SSAs. Earlier I argued that SSAs are essentially empirical, and that, other things being equal, the more plausible the empirical ground the better the relevant SSA. How plausible, then, are the empirical claims that we are bad at abiding by the distinction between good and bad SSAs (and between good SSAs and other non slippery slope bad arguments), and that using good SSAs will therefore increase the likelihood of us using bad ones? In the absence of relevant research-data (or, at least, the absence of such data I m aware of), I hope the following speculation will not be too objectionable. Almost every single text that discusses SSAs emphasizes that SSAs are very often exaggerated and abused. It is commonly noticed that even arguments that are clearly empirically ungrounded or that are fallacious (whether or not they merit being called SSAs) are often put forward as purportedly good SSAs. Isn t this all the empirical evidence that is needed in order to conclude that we really are bad at making the relevant distinction, the distinction between good and bad SSAs? Not quite. One may concede that people very often use bad arguments, bad SSAs (or bad arguments wrongly thought of as SSAs) included, without conceding the empirical premises my argument trades on. For my argument to run through, it is necessary that (part of) what causes people s use of bad SSAs is the use they re making of good SSAs and their being bad at distinguishing between good SSAs and bad ones. Evidence that people use bad SSAs is not enough, of course, to establish this claim. Let me note here, then, that this is the nature of the empirical claim my argument depends on, and let me conjecture that it is true. As support for this conjecture, I want to use a method discussed in the previous section: an SSA is a good argument for someone dreading the bottom of the slope if and only if 19 the device of stages is a good strategy for someone who wants to get to the bottom of the slope, and tries to get an initially reluctant audience to agree. Suppose, then, that I want to get you to think even of bad SSAs that they are good arguments (perhaps because I am an ardent defender of the status quo across the board). Will it be rational for me to try to get you, as a first stage, to use good SSAs, because that may very well lead, eventually and through many stages, to your using bad SSAs? This too, of course, is largely an empirical question, but it is one I think most of us would be rather confident 18 Walton (above n1at22)notices this point. 19 But see n 13, above.

10 638 Oxford Journal of Legal Studies VOL. 21 in answering positively. If that is so, and if we would be right in giving such an answer, my SSA against using SSAs is empirically very plausible indeed. 4. Implications and Objections In the previous section I have given what I take to be a perfectly reasonable SSA against using SSAs. What follows from that? In subsection A I try to answer this question, and in subsection B I discuss some possible objections to my argument. A. Implications of the SSA against Using SSAs The argument of the previous section an SSA against using SSAs serves to show that there is something inherently unstable about using SSAs (this statement will have to be qualified shortly). It shows that taking all SSAs seriously is a strategy that cannot be consistently deployed, because there is an argument that undermines it and that must, according to this very strategy, be taken seriously. One way to see this point is to think of my SSA against using SSAs as an ad hominem dilemma directed at the proponent of, say, an SSA against voluntary euthanasia: either SSAs give a strong reason not to perform the contested action, or they don t; if they don t, your argument is undermined; if they do, my argument gives a strong reason not to use yours; therefore, you have either failed to give a good reason not to allow voluntary euthanasia, or the reason you have given is a reason we have strong reason to ignore. As already hinted, this would be too quick. SSAs differ, I ve said several times, in their plausibility, depending on the plausibility of the relevant instances of the Essential Premise and on the badness of the feared outcomes (as well as on the goodness or badness of the alternative courses of action). Seeing that some SSAs are better than others, a proponent of an SSA against allowing voluntary euthanasia may argue that hers is a better SSA and mine is worse. She may then deploy the strategy of taking only some SSAs seriously; as long as hers is better than mine, this would be a consistent strategy. I want to put forward here, then, two weaker conclusions. First, a proponent of a specific SSA must face the challenge of presenting an objection to my argument that is not also an objection to hers. Second, my SSA against using SSAs sets an upper bound on the strength that can be consistently attributed to SSAs. Let me briefly discuss these points, starting with the second. The stronger the reason you think SSAs generally give to avoid the contested action, the stronger the reason you must on pain of inconsistency think my SSA gives not to use SSAs. So the stronger you generally take SSAs to be, the stronger the reason you must concede you have not to use SSAs. Once faced with the SSA against using SSAs, then, the aficionado of SSAs must qualify her attitude towards them. She must insist that some of them (including the one against using SSAs) are not that good, or that all SSAs are defeasible (and that the one against using SSAs is in fact defeated), and so on. At the very least,

11 WINTER 2001 Slippery Slope Arguments 639 then, my argument sets an upper bound on how enthusiastic one can rationally be about (all) SSAs. This, though, is not a very interesting conclusion, for it has often been noticed that SSAs are defeasible, and that some are better than others. Fortunately, then, more follows from the argument in the previous section. The other conclusion mentioned earlier runs as follows. The proponent of a specific SSA must find a way to distinguish between her SSA and mine, or else she cannot consistently argue that hers justifies its conclusion and mine does not. Note that the case at hand is not a simple case of conflicting reasons that need to be weighed against each other. My argument, if sound, gives a secondorder, exclusionary, 20 reason, a reason to ignore (or to accord less weight to) reasons given by (typically, but not necessarily, first-order) SSAs. So in order to defend an SSA it is not sufficient to show that it is stronger than mine. Rather, it is necessary to defend a consistent strategy that takes the relevant SSA but not the SSA against using SSAs seriously. In order to do that, showing that the first-order SSA is stronger than my second-order one is necessary, but not sufficient. A reason must also be given why this difference makes a difference; why, that is, it is justified, for some strength-threshold in between the strength of the first- and second-order SSAs, to take seriously arguments stronger than, but not weaker than, that threshold. If that cannot be shown, my SSA still gives a reason to ignore the reason given by the relevant first-order SSA. There are two general ways the proponent of SSAs may be able to face this challenge. First, she may argue that the bad consequences her SSA warns against (compared with the consequences of alternative courses of action) are much worse than the bad consequences my SSA warns against (compared with the consequences of the alternatives). Or, second, she may argue that the instance of the Essential Premise her argument trades on is empirically more plausible than the one my SSA trades on. 21 That she may be able to face this challenge, however, does not show the challenge is not there to be faced. And unless the proponent of an SSA can thus distinguish between her SSA and mine, my argument shows that her argument must be ignored. The following is an interesting implication of this general point. When arguing for the Essential Premise, the proponent of an SSA must not simply argue for the general thesis that we are bad at making and abiding by distinctions. 22 This thesis, if true, would support my SSA as much as hers. To face the challenge discussed above, then, she must show why the particular instance of the Essential 20 In the sense introduced by Joseph Raz. See J. Raz, The Authority of Law (1979), part If the plausibility of the relevant instances of the Essential Premise (and of the claim that A ing will therefore cause Z ing) is put in probabilistic terms, and the badness of the relevant consequences (compared with the consequences of the alternatives) is quantified in terms of, say, utility, and assuming risk-neutrality, then the challenge the proponent of a specific SSA faces is to show that the product of the badness of the consequences her argument warns against and the probability of the instance of the Essential Premise relevant to her argument is larger than the product of the badness of the consequences my argument warns against and the probability of the instance of the Essential Premise relevant to my argument. 22 Schauer (above n 5 at 376) notes that if SSAs are to be more than just claims that we ought to be careful because we can t predict the future, then the one putting forward the SSA must be able to show why we are more likely to fail to abide by the relevant distinction in this case than usually.

12 640 Oxford Journal of Legal Studies VOL. 21 Premise her argument trades on is empirically plausible, and furthermore why it is more plausible than the particular instance my SSA trades on. She must show, in other words, why we are more likely to fail to abide by the distinction between, say, voluntary and involuntary euthanasia than the one between good and bad SSAs (or at least the one between this good SSA and bad SSAs), or why the we relevant to her Essential Premise are worse at abiding by the distinction relevant to her argument than the we relevant to my Essential Premise are at abiding by the distinction relevant to my argument, and so on. Anything less than that will just not do. B. Objections In this subsection, I want to discuss several possible objections to my SSA against using SSAs (one possible objection that my argument is based on empirical speculation has already been discussed; I have conceded that this is so, and tried (in section 3) to show that the relevant empirical conjecture is very plausible). In discussing the first two objections I want to do more than just show how they can be answered; I want to show that they are really particular instances of objections that apply generally to SSAs (trying to object to my argument can thus be a productive way of finding general difficulties with SSAs). As has already been noticed, then, the challenge for the proponent of a specific SSA is not just to raise an objection to my SSA, but rather to raise an objection to my argument that is not equally an objection to the SSA she puts forward (this is what the last objection tries to do). (i) Elitism and becoming better at abiding by distinctions When using the SSA against using SSAs, I in effect warn an audience not to use even good SSAs, because that will lead them irrationally to use bad SSAs. There seems to be something disturbingly élitist about such a move. 23 Furthermore, it may be argued that the right thing to do in such circumstances is not to avoid using good SSAs, but rather to improve our abilities to distinguish and abide by the distinction between good and bad SSAs. Anything else, it may be argued, would be an insult to the intelligence of the audience one is trying to convince. This objection is not, I think, strong. Elitist arguments may nevertheless be good ones, and the fact that it would be nice if we were to improve our distinctionabiding abilities does not give a reason to ignore the fact that as things actually stand we are liable to fail to abide by some distinctions (saying this is perfectly compatible, of course, with saying that we ought to try and become better distinction-abiders). If taking that into account amounts to insulting the intelligence of one s audience, then insulting the intelligence of one s audience may, under certain circumstances, be the right thing to do. 23 Though not necessarily: as noted in section 2, the relevant we may be people or institutions not in your audience (in which case the argument still seems elitist, only now the audience are considered a part of the élite), or the relevant we may even include oneself (in which case the argument is not élitist).

13 WINTER 2001 Slippery Slope Arguments 641 Be that as it may, this objection is equally an objection to all SSAs. 24 Every SSA assumes the fallibility of the relevant decision makers; every SSA is thus likely to be elitist, and perhaps even to insult the intelligence of the relevant audience. This objection, then, cannot save the proponent of a specific SSA, because it cannot serve to distinguish between her SSA and mine. And, as was emphasized earlier, this is what would be needed if a specific SSA is to be saved. (ii) A way out midway down the slope? Convinced by my arguments that using good SSAs is dangerous, someone may still object along the following lines: All you ve shown is that we should be more strict in setting the threshold for acceptable SSAs than we would otherwise be. This conclusion is of some interest, but it is weaker than the one you try to defend. It does not justify not using SSAs. Setting a higher standard for the acceptability of SSAs does not mean setting the highest possible one. There is nothing wrong with this objection: it just serves to emphasize once again what the empirical premises my argument relies on are. My argument only gives a reason to avoid all SSAs if we are likely to fail to abide by any more subtle distinction midway down the second-order slope. But note, as Williams does, 25 that a similar point applies to all SSAs: if there is a point somewhere down the slope where we might stop well ahead of reaching the objectionable Z, then the SSA does not give a reason not to take the first step down the slope. Once again, then, the objection fails to distinguish between the relevant firstorder SSA and mine. (iii) An argument against using any arguments? Given that we are less than perfect at making and abiding by the distinction between good and bad arguments in general (and not just between good SSAs and bad SSAs), isn t it true that using good arguments may lead because we are less than perfect at making this distinction to our using bad ones? So isn t my argument really an SSA against using arguments, and not just against using SSAs? And is that not all that is needed in order to show that there must be something wrong with my argument? 26 Fortunately, my argument cannot be generalized in the way suggested by the objection. To see this, consider again the instances of the Essential Premise that are relevant for my argument and for its generalization. My argument relies on us being bad at distinguishing between good SSAs and bad SSAs (and other 24 Den Hartogh (above n 1 at 286) notices this general (perhaps problematic) feature of SSAs. Van der Burg (above n 8 at 61) makes a similar point in saying: Very often reasonable distinctions may be made effective. Whitman also says that the thing to do may be to improve our ability to make the right distinctions (J. P. Whitman The Many Guises of the Slippery Slope Argument, Social Theory and Practice 20(1) (1994) at 90). And Kleinig quotes from a debate about an SSA warning against policemen and policewomen accepting free coffee (because this may lead to more seriously objectionable gifts), where one person objecting to this SSA argues that it insults the intelligence of the officers. (J. Kleinig, The Ethics of Policing (1996) at 178.) 25 Above n1at I thank Andrea Iacona, Josh Schechter and Frances Kamm for this objection.

14 642 Oxford Journal of Legal Studies VOL. 21 bad arguments that are sometimes erroneously thought to be SSAs). The generalization relies on our being bad at distinguishing between good arguments and bad arguments. Now, had we been just as bad at making (and abiding by) the latter distinction as we are at making the former one, the objection in this subsection would have been well taken. But we are not, and it is not. I conjecture, then, that we are worse at distinguishing between good and bad SSAs than we are at distinguishing between good and bad arguments in general. At the very least, there are many kinds of arguments where we are much better at distinguishing between the good ones and the bad ones than we are with regard to SSAs. And so my argument is much stronger when applied to the usage of SSAs than when applied to arguments in general, or to those kinds of arguments in particular. Of course, this is an empirical claim (hence the use of the word conjecture ). But it is, I think, a rather plausible one: after all, SSAs have their bad reputation for a reason. And this empirical conjecture becomes even more plausible when it is remembered that we may refer to whoever is likely to make the relevant future decisions. I think there is strong inductive evidence that (some) political institutions are exceedingly bad at making the distinction between good and bad SSAs, much worse than they are at making the distinction between good and bad arguments more generally. Perhaps, though, this is not a satisfactory reply to the worry expressed by the objection. For even if we are indeed worse at distinguishing good from bad SSAs than we are at distinguishing good from bad arguments in general, my argument still commits me to the following possibly disturbing counterfactual: had we been as bad at distinguishing good from bad arguments in general as we (actually) are at distinguishing good from bad SSAs, it would have been the case that we ought not to use any arguments. 27 Avoiding using SSAs, I have argued, would have an all-things-considered good effect; it is better not to use even good SSAs than to use good and bad ones alike. But can it be seriously suggested that avoiding using arguments in general would have an all-things-considered desirable effect? If not, as seems to me plausible, then my argument doesn t generalize to an argument against using arguments in general even in the counterfactual circumstances imagined: even had we been as bad at distinguishing between good and bad arguments in general as we actually are at distinguishing good and bad SSAs, we would still have had reason to use (some) arguments, though not SSAs. Perhaps, though, we can imagine an even more surprising set of circumstances, where the all-things-considered effect of using arguments in general is undesirable, indeed as undesirable as that of using SSAs; am I not then committed to it being the case that we ought not to use arguments in general? Here I answer simply Yes. But in such a case, I submit, this conclusion is no longer at all counterintuitive. 27 I thank the reviewer for the OJLS for pressing me on this point.

15 WINTER 2001 Slippery Slope Arguments On Politicians and Philosophers: A Concluding Remark SSAs are very commonly thought to be better when applied to institutional decisions and to politicians as the relevant decision makers than when applied directly to the moral questions themselves as discussed by philosophers. 28 Philosophers, it is often noted, are better at making distinctions and perhaps also at abiding by them (though this is far more controversial), and so are less likely to fall victim to the kind of mistake the failure to distinguish between A and Z that SSAs trade on. Furthermore, critical philosophical discussion is more likely to lead to the participants making the right distinction than, say, a parliamentary procedure. If my SSA against using SSAs is a good argument a surprising result follows. For who is more likely to fail to make the distinction between good and bad SSAs politicians or philosophers? Once again, I think, politicians are more likely to miss the distinction. If that is so, my argument gives a very strong reason not to let SSAs have influence in the political arena indeed, a much stronger reason than the one it gives not to use SSAs within a philosophical discussion. Not only, then, are the dangers of using SSAs more severe in political discussions than in philosophical ones (letting the defenders of the status quo hinder discussion); but also these dangers are much more likely to occur when SSAs are used in the political sphere. 6. Appendix: Different Kinds (and Accounts) of SSAs In order to further clarify what I take SSAs to be, it may be useful to briefly distinguish my kind of SSAs from some others that are to be found in the literature under that name. Throughout this appendix I assume that SSAs form an interestingly distinct kind of argument, and that they are not all obviously bad. That a purported SSA is really just an uninteresting instance of another general kind of argument therefore serves as evidence that it is not a genuine SSA after all, as do reasons to believe that a purported SSA is obviously fallacious. I do not wish to defend this assumption here. It is, I think, a sound methodological assumption to make when trying to characterize SSAs. First, SSAs are not sorites-type arguments. An example of such arguments, called conceptual SSAs by Govier, logical SSAs by Walton and L 2 (the second kind of logical SSAs) by van der Burg 29, is the following: 30 a 10-year-old boy is a person; from the time of conception to the time when a child is 10- year-old, there is no sharp cut-off point for which it is reasonable to argue that before it the child (or fetus) is not a person, and after it it is; therefore, from conception the fetus is a person. That this is not a good argument has been 28 See, e.g., van der Burg, above n8at61 2; Den Hartogh, above n1at286 7, and the references there. 29 See Govier, above n8at303; Walton, above n 1, ch 2; van der Burg, above n8at Discussed, e.g., by C. Li The Fallacy of the Slippery Slope Argument on Abortion, Journal of Applied Philosophy 9(2) (1992)