Sei sulla pagina 1di 16

Health Care Anal

DOI 10.1007/s10728-009-0117-0

ORIGINAL ARTICLE

A Logical Analysis of Slippery Slope Arguments

Georg Spielthenner

 Springer Science+Business Media, LLC 2009

Abstract This article offers a logical analysis of Slippery Slope Arguments. Such
arguments claim that adopting a certain act or policy would take us down a slippery
slope to an undesirable bottom and infer from this that we should refrain from this
act or policy. Even though a logical assessment of such arguments has not received
much careful attention, it is of vital importance to their overall assessment because
if the premises fail to support the conclusion an argument is worthless. I partition
slippery slope reasoning by means of two dichotomies (reasoning under certainty vs.
uncertainty and one-step vs. multiple-step reasoning) into four general categories
and evaluate these in turn. The analysis reveals that slippery slope arguments are
logically fallacious.

Keywords Ethical reasoning  Medical decision-making  Practical reasoning 


Rational choice  Slippery slope arguments

Introduction

It is beyond doubt that in addition to theoretical reasoning there is practical


reasoning. We do not only want to work out what is the case but also what we ought
to do. One type of practical arguments is the so-called Slippery Slope Argument
(hereafter, SSA). Slippery slope reasoning is, however, a many-faceted activity and
there are thus many varieties of SSAs. The most common way of classifying them is
by distinguishing logical and empirical forms.1 Since my chief concern in this study
1
More about the distinction between different types of SSAs can be found in Enoch [10], den Hartogh
[9], Lillehammer [15], Lode [16, p. 1477], or Smith [23].

G. Spielthenner (&)
Department of Philosophy and Applied Ethics, The University of Zambia, Canada Court B3,
P.O. Box 32967, Lusaka, Zambia
e-mail: georg_spielthenner@yahoo.de

123
Health Care Anal

is with slippery slope reasoning in biomedical ethics and most SSAs in medicine
and in biology are empirical forms I shall restrict my attention to this type only.
Hence, in the remainder of this paper I mean by ‘slippery slope argument’ empirical
forms only.
There is, however, some difficulty in defining these SSAs since their nature is
variously understood by authors and many different arguments are sometimes called
‘slippery slope argument’. Some authors have proposed definitions that are either
too narrow,2 only stipulative (see [10], or are flawed for other reasons).3 Moreover,
attempting to give a precise definition seems to me misguided because the usage of
the term ‘slippery slope argument’ is so loose that there is no distinctive form of an
SSA. I am here in agreement with den Hartogh [9] and Lode [16]. The notion of a
slippery slope argument seems to be a so-called family resemblance concept. That
is, these arguments do not have a common property but are rather held together by a
network of resemblances. Hence, the term ‘slippery slope argument’ does not
denote a sharply circumcised concept, and SSAs are best seen as a family of related
arguments.
Nevertheless, for the purpose of this essay we can roughly characterise an SSA as
a practical argument that claims that permitting an act (or policy) will lead to, or
increase the likelihood of, an undesirable state of affairs and that this act (policy)
should therefore not be permitted. For illustration, consider this simple example:
‘‘The selling of human organs should be outlawed because allowing human organs
to be sold will inevitably lead to a situation in which only the rich will be able to
afford transplants.’’ According to the imagery of a ‘‘slippery slope’’, we should not
allow selling human organs on the ground that this would mean stepping on a
slippery slope and sliding all the way to its bottom—that is, to the undesirable state
of affairs that only the rich will be able to afford transplants.4
An objection may be raised at this point: According to deontological theories
(e.g., Kantian ethics), some actions are intrinsically bad (malum in se)—that is, they
are forbidden because of their intrinsic nature and not because of the undesirable
state of affairs they cause. In this essay, I do not deal with deontological reasoning
and it may therefore be objected that it has no relevance to deontologists.5 However,
it is worth stressing that most deontologists hold that the rightness of many acts is at
least partly determined by their consequences, which renders the paper relevant to
deontologists also. Furthermore, for reasons of space and the fact that the majority
of SSAs that can be found in the recent literature of ethics are not deontological, my
focus on consequentialist reasoning seems to be justified.
Despite the important role SSAs have played in many fields, they have a bad
philosophical reputation. Partly because many slippery slope claims are wildly

2
For example, the definition given by Volokh [26] refers to probabilistic versions of SSAs only.
3
See, for instance, the definitions of Schauer [19] or Schubert [20], which have an indeterminate logical
structure. It is not clear whether their form is a conditional or a biconditional.
4
Other metaphors that have been used to refer to this type of reasoning are ‘‘wedge argument’’, ‘‘thin
edge of the wedge’’, camel’s nose‘‘, and sometimes ’’domino‘‘ because an SSA resembles a row of falling
dominoes, with each event causing the next one in line (see [16]).
5
I am grateful to an anonymous referee at this journal for alerting me to this problem.

123
Health Care Anal

exaggerated (e.g., that allowing euthanasia will lead to Nazi-style death camps),
they have been dismissed as paranoid and vacuous [26]. Moreover, they have been
denounced as arguments that have mainly demagogical power and can hardly have
decisive force and should therefore not play an important role in ethics [25]. Even
elementary logic textbooks put them on their list of ‘‘informal fallacies’’.
Yet despite the frequency of objections, slippery slope arguments persist in the
context of many practical discussions. SSAs are used extensively in legal contexts,6
but they are also employed in political philosophy or reasoning about environmental
problems (see [12]). In particular, SSAs play central roles in many debates in
medical ethics. They have been invoked in the debate on abortion,7 against the
legalization of voluntary euthanasia and physician-assisted suicide,8 and against
organ donation by unrelated donors (e.g., [13]). More recently, slippery slope
arguments have been put forward against stem cell research,9 gene therapy,10 in
vitro fertilization,11 cloning,12 new forms of eugenics,13 or prenatal diagnosis14—
and this list is by no means exhaustive.
There is thus no doubt that slippery slope reasoning is an important type of
reasoning in medical ethics. But given the disagreement surrounding SSAs—some
advance them to urge rejection of various practices, while others are highly critical
of such arguments—as well as the frequency with which SSAs are invoked, there is
a need to critically assess this type of argument. Slippery slope arguments can,
however, be assessed in numerous ways. For instance, an epistemic assessment is
concerned with the justification of the factual premises used in such arguments.

6
For instance, Codd [8] discusses the argument against allowing prisoners access to conjugal visits by
their partners. Further legal discussion can be found in Enoch [10], Lewis [14], Lode [16], Schauer [19],
Smith [22], or Volokh [26].
7
See, for example, the example given by Volokh [26] against opposing partial-birth abortion bans
because this might lead to a broader abortion restriction (p. 1031).
8
Singer [21] holds that the SSA is ’’undoubtedly the most widely invoked secular argument‘‘ against the
legalization of voluntary euthanasia and physician-assisted suicide (p. 538). See also Clark [6], Lode [16],
Nielsen [17], and Smith [23].
9
For example the argument ’’that even if research on embryos is not wrong in itself, it will open the way
to a slippery slope of dehumanising practices, such as embryo farms, cloned babies, the use of fetuses for
spare parts, and the commodification of human life‘‘ [18]; compare to this also Camporesi [5].
10
Compare this example, given by Bayertz et al. [1], ’’The application and development of somatic gene
therapy will lead to an increase in willingness to intervene in the germ-line cells …‘‘ (p. 467).
11
For instance, ’’In vitro fertilization is another step down the slippery slope towards the clear evil of
eugenics‘‘. Christianity Today, 2007. Available at: http://www.christianitytoday.com/ct/2007/march/24.
29.html (accessed 12 May 2008).
12
’’Cloning human embryos for research purposes will lead down a slippery slope to reproductive
cloning and finally to ‘‘designer babies’’. Available at: http://news.bbc.co.uk/2/hi/science/nature/
377046.stm (accessed 10 August 2007).
13
Compare this example, reported by Bostrom [3], ‘‘To forestall a slide down a slippery slope towards an
ultimately debased ’posthuman’ state, human enhancement technologies should be banned globally’’ (see
p. 202).
14
One argument in this context is that prenatal diagnosis may open the door for ‘‘sexing for all disorders
in which the mutation’s penetrance and/or the expression is lower or milder in one sex’’ (see [4]).

123
Health Care Anal

Most authors evaluate SSAs from this viewpoint only.15 In a valuational assessment
we ask what reasons the arguer has for claiming that the bottom of the slope is
undesirable or ‘‘morally unacceptable’’, and in a rhetorical appraisal we consider
how persuasive and influential a certain type of argument is.
In this essay, I have nothing to say about these types of assessment. I shall rather
confine my attention to a logical evaluation of slippery slope reasoning. A logical
investigation is concerned with the formal structure of arguments. In other words, in
a logical evaluation of slippery slope reasoning we want to know whether the
acceptance of the premises logically commits one to the acceptance of the
conclusion in the sense that it would be inconsistent to assent to the premises but
deny the conclusion. This issue has not received much careful attention16 although it
is of vital importance to the overall assessment of SSAs because if the premises fail
to support the conclusion an argument is worthless. A closer look at the logical
structure of slippery slope arguments seems therefore long overdue.
Before going on to a detailed consideration of SSAs, it is worthwhile to mention
that I will take it for granted in this study that practical arguments can be logically
conclusive. That is to say that we can validly derive a conclusion from its premises.
This view is by no means beyond dispute. Many logicians and philosophers endorse it,
but several writers have argued against it. However, a consideration of this issue
would take us beyond the confines of the present work and since I have argued for the
validity of practical arguments elsewhere,17 I shall not pursue this issue further here.
With those preliminaries out of the way, I now approach the main topic of this
paper. In what follows I shall assess empirical slippery slope arguments from a
logical point of view. For simplicity of exposition, I shall partition slippery slope
reasoning by means of the following two dichotomies: reasoning under certainty
versus uncertainty and one-step versus multiple-step reasoning. By combining these
two distinctions, we can group empirical SSAs into four general categories, which I
shall analyse now in turn.

One-step Slippery Slope Reasoning Under Certainty

SSAs in their most simple form are arguments that accepting a certain practice will
lead to an undesirable state of affairs, from which it is inferred that this practice
should not be allowed. Arguing this way is said to be under certainty because the
arguer has no doubt about what the outcome of accepting this practice will be and it
is a one-step SSA because it is claimed that accepting this practice will take us to the
15
See, for instance, Enoch [10] who asks, ‘‘Why, though, should we believe that allowing voluntary
euthanasia would result in allowing involuntary euthanasia?’’ (p. 630). Compare also Gay-Williams [11],
Lewis [14], and Singer [21].
16
Schauer [19] claims to ‘‘flesh out the logical structure’’ of SSAs (p. 363) but his study is not a logical
investigation. In general, if writers claim to give a logical assessment, they tend to confuse logical and
epistemic issues (see, for instance, [16, 22, 27]).
17
See Spielthenner [24]. It should be noted that my logical assessment will be based on an intuitive
understanding of when practical arguments are valid. The chief reason for this is that a formal account of
practical validity is an issue that is too complex to pursue in any detail here. Moreover, I think that my
arguments are intuitively plausible upon consideration.

123
Health Care Anal

bad consequence in a single step and not via a number of intervening steps. It may
well be that reasoning of this type is often elliptical in the sense that a reasoner just
does not mention the intervening steps although he believes there are some. But
many SSAs are one-step arguments and therefore we need to have a closer look at
them.18 Here is an example of such an argument:
We must not allow therapeutic cloning because this will inevitably lead to
reproductive cloning.
The logical structure of this type of SSA is clear: From stepping on the slope
(allowing therapeutic cloning, in our example) it follows logically that we slide to
the bottom of the slope (reproductive cloning), and from this it is inferred that we
must not step on the slope. For definiteness, let me put this in logical notation. Let
‘s’ stand for ‘We step on the slope,’ and ‘b’ for ‘We slide to the bottom of the
slope.’ From s (stepping on the slope) and s ? b (if we step on the slope we slide to
the bottom) it follows ‘b’ (we slide to the bottom). Put more formally: s, s ? b  b
is a valid argument. (On the left side of ‘’ (the ‘‘double turnstile’’) are the premises
of the argument and on its right side is the conclusion.)
A word of explanation will be in order: The term ‘follows logically’ is taken in
the strict sense of classical logic.19 I do not, however, hold that there exists a logical
relation between stepping on the slope and sliding to the bottom. This relation is
non-logical, symbolized here by ‘s ? b’. But from this non-logical relation,
together with the fact that we step on the slope, it follows logically that we slide to
the bottom. This structure is the reason why some authors in the field of practical
reasoning have regarded arguments of this type as valid (e.g., [2] or [7]).
I now proceed to the logical assessment of this type of slippery slope reasoning.
At first sight, it may appear that such SSAs are valid. But on closer examination it is
obvious that arguing in this way is fallacious. In what follows I shall explain why.
(1) The main reason, I suspect, why authors have often regarded this type of
reasoning as valid is that they have ignored the simple fact that drawing a
conclusion from an SSA requires comparing the values of available alternatives and
not only considering the bad outcome of a certain act or policy. Let us return to our
example. The putative fact that allowing therapeutic cloning will inevitably lead to
reproductive cloning does not entail the conclusion that we must not allow
therapeutic cloning. This is so for the simple reason that not allowing therapeutic
cloning might have consequences that are even worse.
In practical reasoning we argue for or against an act relative to a given set of
possible acts, commonly called alternatives.20 Alternatives represent the range of

18
Lode [16] holds that one-step arguments are not SSAs because the imagery such arguments suggest ‘‘is
more reminiscent of a cliff or wall than a slope’’ (p. 1477). I agree that one-step arguments are borderline
cases but since they are often used we cannot ignore them here.
19
Of course, I have been speaking somewhat loosely here because nothing follows logically from doing
something. My explanation must therefore be understood as shorthand for ’From the propositions that we
step on the slope and that we slide to the bottom if we step on the slope, it follows the proposition that we
slide to the bottom’.
20
For simplicity, it is here assumed that there are a finite number of alternatives and a finite number of
possible outcomes.

123
Health Care Anal

options we have in the circumstances; and one of the first tasks in a practical
problem is to draw up a list of options that are available. Arguing for or against a
policy requires comparing these alternatives, and we can only draw a conclusion
when we have assessed the relative merits of these options. In other words, a course
of action should not be avoided because it leads to a bad state of affairs but because
it is worse than another action that is available.
One problem of slippery slope reasoning is taking too narrow a focus by
concentrating on the bad outcome of an act instead of comparing this act to its
alternatives. Finding an appropriate balance among alternatives is often difficult, but
not trying to balance them is a mistake in reasoning.
Figure 1 is a so-called decision tree that provides a graphical representation of
the essence of one-step SSAs under certainty. This simple tree begins at the point of
decision (the square on the left), with three possible paths representing the
competing alternatives (for the sake of simplicity, only three have been drawn).
(Notice that in all figures the bold path represents the reasoning of proponents of
SSAs.)
At this stage it might be objected that sometimes we have no choice and then we
cannot (and therefore need not) consider alternatives. But if we really have no
choice—if we slip off the ladder or if our body is held immobile—then we do not
reason what to do. Reasoning whether to commit an act requires that we must think
that at least its omission is open to us. This may appear so obviously true as to be
hardly worth saying, but most advocates of slippery slope reasoning have
overlooked this fact, which is crucial for understanding the nature of practical
reasoning.
My claim that no practical conclusion can be drawn from the fact that an act has a
certain (good or bad) outcome will become more plausible when we consider now
what might be termed the conclusion-problem of SSAs.
(2) Authors are not agreed on what kind of normative conclusion is derivable
from an SSA and they draw therefore different conclusions. Some infer that we must
not step on the slope (e.g., [25], 42), others that we should not do so (e.g., [20], 262),
that we had better not stepped on the slope (e.g., [14], 197), or that it is
unacceptable to do this (e.g., [23], 224). One might think that this could be just slips
of the pen. As a matter of fact, however, it reveals a profound misunderstanding of
the nature of practical reasoning. As will appear from the discussion to follow,
practical reasoning can have different conclusions, depending on its logical

Fig. 1 The structure of one-step SSAs under certainty

123
Health Care Anal

structure. I have argued for this elsewhere,21 and I shall be brief about this here
because the details are more than we need.
As I have already mentioned, there must be alternatives to stepping on the slope.
In our example, therapeutic cloning could be forbidden or allowed only under
specified conditions. Obviously, these alternatives have their own set of conse-
quences, and we can only infer that therapeutic cloning must not be adopted if all
alternatives have better consequences than allowing this practice.22 Authors who
hold that we must not step on the slope or that this is unacceptable seem to assume
that stepping on the slope is always worse than any of its alternatives. But this need
not be the case. Even if stepping on the slope starts us down to the bottom, this
outcome may still be better than the consequences of not stepping on the slope. If
not allowing therapeutic cloning has, overall seen, worse effects than allowing it,
then it follows the opposite conclusion that we ought to allow therapeutic cloning—
despite the fact that it leads to the allegedly bad consequence of reproductive
cloning. It should thus be clear by now that the fact that stepping on a slope leads to
an undesirable state of affairs does not ipso facto justify the conclusion that we
should avoid stepping on this slope.
(3) Suppose now that a proponent of slippery slope reasoning has considered
alternatives to stepping on the slope and has found that their consequences are better
than the bottom of the slope. His argument is then still not unqualifiedly valid
because slippery slope reasoning is defeasible.23 Let me explain this. It is a
characteristic feature of practical reasoning that it is defeasible or, in the technical
jargon, non-monotonic. This means, if a conclusion follows validly from given
premises, it need not follow from a larger set of premises, even if the original
premises are included. In other words, saying that a conclusion follows defeasibly
means that it follows relative to a given set of options and their outcomes.
Considering additional alternatives or new consequences may ‘‘defeat’’ the
argument. Let a single example serve to illustrate this: You want to fly to Rome
tonight, and you can get seats on Air France and Lufthansa. Since the fare on Air
France is lower (and there are no other differences) you conclude that you should
take Air France. But then you learn that KLM offers a flight to Rome also, for an
even lower fare than Air France (again, by hypothesis there are no other differences).
This additional premise ‘‘defeats’’ the original argument. It no longer follows that
you should take Air France; the new conclusion is that you should take KLM.
It is particularly important to note that SSAs are defeasible. Proponents of this
type of argument tend to propound one reason against stepping on a slope and they
seem to think that this settles the matter. Apparently it does not occur to them that
their reasoning can be defeated by new options or consequences not yet considered.
It goes without saying that in slippery slope reasoning we normally cannot be sure
that some options have not been omitted, and an argument that includes the worry of
21
See my ‘‘Instrumental Reasoning Reconsidered’’ [forthcoming].
22
Observe, that we can use any of a number of phrases to express such conclusions because ordinary
language does not uphold a sharp distinction between different deontic modals. For example, I could have
said just as effectively that we canot allow therapeutic cloning or that this is out of the question.
23
In a more traditional terminology, these argument have been called prima facie valid or pro tanto
valid, but it is now an established terminology to call them defeasible.

123
Health Care Anal

sliding down a slope (but does not consist in this concern) is therefore at best
defeasibly valid.

Multi-step Slippery Slope Reasoning Under Certainty

In the preceding section, I indicated that one-step slippery slope reasoning is the
exception rather than the rule. In the present section, my purpose is therefore to
analyse slippery slope reasoning under certainty that consists not only of the top of
the slope and its bottom but of any number of intervening steps also. For illustration
consider this example:
Accepting somatic gene therapy will increase the willingness to intervene in
the germ-line cells, which will eventually cause us to accept genetic
enhancement also.
To see why this kind of reasoning is only apparently valid, one needs to get clear
what its basic logical features are. Let us therefore have a closer look at its logical
structure.
According to this type of reasoning, stepping on the slope will lead to an
intervening step i1, which will lead to i2,…, in, which in turn will lead to the
undesirable bottom of the slope, and from this it is concluded that we should avoid
stepping on the slope. Some logical notation can make this clearer. Let ‘in’ mean
‘We slide to an intervening step n’ and let i1, i2, i3,…, in be the sequence of
intervening steps of an SSA. A multi-step SSA under certainty has then the
following structure: s ? i1, i1 ? i2,… in-1 ? in, in ? b, and since ‘b’ stands for
the undesirable bottom of the slope, it is concluded that ‘s’ (stepping on the slope)
should be avoided. Figure 2 illustrates the structure of this type of SSA.
From a logical point of view, we need now to distinguish two cases: (1) The
intervening steps can be valuationally neutral. That is to say that the arguer does not
regard them as desirable or undesirable; he worries only about the bottom of the
slope. In this case, the intervening steps can be ignored because from the premises
s ? i1, i1 ? i2,…, in-1 ? in, in ? b it follows s ? b and from this premise and s
(stepping on the slope) it follows b. In other words, if the intervening steps are
valuationally neutral, we can reduce a multi-step SSA with any number of
intervening steps to a one-step SSA, and the comments made in the previous section
apply to this type also.

Fig. 2 The structure of multi-step slippery slope reasoning under certainty

123
Health Care Anal

(2) In most cases, however, intervening steps are not valuationally neutral. On the
contrary, they tend to be disvalued and all the more so the farther down the slope
they are. To clarify this variety of SSA, it is useful to take advantage of the
following fact of logic: From the premises {s ? i1, i1 ? i2,…, in-1 ? in, in ? b}
it follows s ? (i1 ^ i2 ^…^ in-1 ^ in ^ b). That is to say, from ‘s’ (stepping on
the slope) it follows the conjunction of all intervening steps (no matter how many
they are) and the bottom. It is therefore possible to reduce this variant of multi-step
SSA to a one-step SSA with multiple consequences. Let me illustrate this by
referring to our example. Instead of saying that somatic gene therapy will increase
the willingness to intervene in the germ-line cells and this will cause us to accept
genetic enhancement also, we could simply say that accepting somatic gene therapy
has the two consequences of increasing the willingness to intervene in the germ-line
cells and causing us to accept genetic enhancement.
Let me now turn to the logical appraisal of such arguments. It is almost a truism
that biomedical decision problems involve multiple and often conflicting values.
When advocates of slippery slope reasoning face such problems they are prone to
the following two kinds of errors in reasoning: (1) They tend to focus only on the
bad consequences of stepping on a slope. That is, they concentrate on the
undesirable bottom and ignore its positive outcomes. This is a mistake because
practical problems with multiple consequences cannot be resolved by focusing on
any one consequence. Upon reflection, this should appear obvious.
(2) Proponents of SSAs do, of course, sometimes admit that stepping on a slope
has good and bad effects. Commonly, however, they argue then that we should
nonetheless not step on the slope because the undesirable bottom would outweigh
the positive effects.24 Let me explain why this is an error of reasoning. If an
alternative (e.g., stepping on a slope) has the consequences c1, c2,…, cn (one of them
is the bottom of the slope), we do not compare the (dis) values of these
consequences. What we rather need to compare are entire constellations of
consequences. For definiteness, let me put this slightly more technically. Suppose
that we have two alternatives a and b, the consequences of a are (ca1, ca2,…, can)
and the consequences of b are (cb1, cb2,…, cbn). We do now not compare, say, ca1 to
ca2 or cb2 to cb3—arguing, for instance, that the negative ca1 outweighs the positive
ca2—we rather need to compare the compounds (ca1, ca2,…, can) and (cb1, cb2,…,
cbn) in order to determine the values of the alternatives a and b—as shown in Fig. 3.
We may then find that the alternative a is better than b.
In short, in practical reasoning we need to assess the relative merits of available
alternatives and not the relative merits of the various consequences of these
alternatives. There are different techniques of comparing constellations of
consequences. But this is a subject that is fortunately outside the scope of this paper.
I now want to apply the lesson we have learnt in the preceding paragraph to the
example given at the beginning of this section. Suppose that the arguer has a positive
attitude towards germ-line therapy (after all, it can eliminate unwanted alleles from
the gene pool) but is worried about genetic enhancement. It is now faulty reasoning to

24
A case in point is Nielsen’s [17] claim that ‘‘when the risked harm of a course of action outweighs its
expected benefits, it ought to be avoided’’ (p. 17).

123
Health Care Anal

Fig. 3 Assessing the relative merits of available alternatives

argue that we should not accept somatic gene therapy because it leads to both germ-
line therapy and genetic enhancement and the disvalue of genetic enhancement
outweighs the positive value of germ-line therapy. We rather need to compare the
compound of germ-line therapy and genetic enhancement to the consequences of
alternatives to somatic gene therapy; and we can only derive the (defeasible)
conclusion that we should not accept somatic gene therapy if at least one of these
options has a better overall outcome than accepting somatic gene therapy.

One-step Slippery Slope Reasoning Under Uncertainty

In the preceding sections, we dealt with slippery slope reasoning when, for practical
purposes, we can know the consequences of each alternative before acting. But
certainty is hard to come by and most of the time, we make do with less. We must
therefore turn now to the more realistic situations in which we do not know what the
consequences will be until after acting—no matter how much time and thought we
expend. Such decisions are said to be made under uncertainty.25 In a biomedical
context, decision problems normally exist in an uncertain environment, and not
knowing exactly what would happen if a particular course of action were to be
adopted adds a new layer of complexity to slippery slope reasoning. But in many
situations uncertainty is unavoidable and effective decision-making requires then
confronting it by judging the likelihood of different outcomes and assessing their
possible results. In this section, I shall discuss one-step SSAs under uncertainty.
Consider the following example:
We should not allow stem-cell research. If this practice is allowed to get a
foothold, scientists may start taking volunteers’ sperm and eggs to create
human embryos for the sole purpose of destroying them for research.

25
Obviously, there can be combinations of certainties and uncertainties in one SSA—e.g., when one step
down the slope is regarded as certain but others only as likely. However, I think it is safe to set such cases
aside because my arguments can easily be adapted for such mixed forms.

123
Health Care Anal

The logical anatomy of such SSAs is obvious: We should not allow a certain
practice (stem-cell research, in our example) because allowing it is likely to start us
down a slippery slope, at the bottom of which is an unacceptable state of affairs
(wantonly creating human embryos only to destroy them for research).
Although the logic of SSAs under uncertainty is more controversial than the
logic of reasoning under certainty, there are—apart from the problems which we
have already considered—two additional logical mistakes, which need to be
discussed in this connection. Proponents of slippery slope reasoning tend to focus
on the bad possible outcome only, and they are liable to make mistakes when they
combine the values of consequences with their probabilities. Let us consider these
issues in turn.
(1) According to one-step reasoning under uncertainty, we may slide down to the
bottom if we step on the slope. From this it follows, however, that we may not slide
down. That is to say, stepping on the slope may have other, possibly positive,
consequences and it is not known before acting what outcome it will have. It is here
advantageous to introduce some mathematical notation. If P denotes the possibility
that an event is happening, then it is a theorem of probability theory that 1 - P is
the probability that this event will not happen. Let us return to the example at the
beginning of this section. If there is, say, a probability of 0.4 that allowing stem-cell
research will lead to creating human embryos for the purpose of destroying them for
research, then there is a probability of 0.6 that this will not happen.
One of the pitfalls proponents of slippery slope reasoning may watch out for is
over-focusing on the negative. To avoid bad consequences, I submit, they give
disproportionate attention to the possible downside of an option, ignoring the fact
that the upside potential may far outweigh the downside risk. To derive a valid
conclusion from an SSA we need to consider the full range of possible outcomes,
not only the ones with bad consequences.
(2) As I have already emphasized, we can only validly derive a conclusion from a
practical argument if we compare the alternatives available to the agent. This
requires estimating their respective values. Suppose that you want to draw the
conclusion that one of your options should be chosen. You know the value of their
consequences and the probabilities of the different possible outcomes. In order to
determine the relative values of your options you need to combine the probabilities
with the values of their consequences. Authors are not agreed on how this can be
done, but one established method (which requires quantifying probabilities and
values) is computing the so-called expected value of each alternative. The expected
value is the sum of the products of the probabilities times the values of the
consequences. For illustration, suppose that an alternative has three possible
outcomes, as shown in Fig. 4.
Figure 4 is a slightly more complex decision tree. The alternative ‘‘stepping on
the slope (s)’’ leads to a fork (the circle) indicating uncertainty. Each possible
outcome of this uncertainty is represented by a branch leading out from the fork;
and each of these branches leads to different consequences, which are summarized
at the tips of the tree.
The symbol P(b) represents the probability of sliding to the bottom, and V(b)
represents its value. The probabilities of the two other outcomes are P(x) and P(y)

123
Health Care Anal

Fig. 4 The structure of one-step slippery slope reasoning under uncertainty

and the values of their respective consequences are V(x) and V(y). The sum of the
products of the probabilities and the values of their consequences is the expected
value (EV). If we write this succinctly in mathematical notation, we obtain the
following: EV = P(b) 9 V(b) ? P(x) 9 V(x) ? P(y) 9 V(y). Be it noted that I do
not commit myself to this method of estimating the value of an alternative. But it is
clear that an agent can derive a normative conclusion only after estimating the
values of the available alternatives by combining the values and probabilities (not
necessarily by multiplying them) of their consequences.
Let us apply this to our example. Allowing stem-cell research has different
possible outcomes; the creation of embryos solely for research is only one of them.
By combining the values of these outcomes and their probabilities, we can estimate
the value of allowing stem-cell research. However, even if this value is negative,
this does not settle the matter because we need to estimate the values of alternatives
to allowing stem-cell research also; and we can only conclude that we should not
allow stem-cell research if at least one of these alternatives is better than allowing it.

Multi-step Slippery Slope Reasoning Under Uncertainty

In most SSAs it is tacitly assumed that there are two or more intervening steps
between the top of the slope and its bottom. If the link between these steps is
probabilistic we obtain a multi-step SSA under uncertainty. Generally speaking,
according to this type of SSA, we should avoid stepping on the slope because this
may lead to the intervening step i1, which may lead to i2,…, in which finally may
lead to the undesirable bottom of the slope. To illustrate this, let us consider a
simple example.
The legalization of voluntary euthanasia is likely to lead to legalizing non-
voluntary euthanasia, which in the end might result in allowing involuntary
euthanasia also. We should therefore not allow voluntary euthanasia.
On the face of it, reasoning of this kind may be regarded as correct practical
reasoning. But once its basic logical feature is clear, we shall realize that it is faulty.
For ease of exposition, let us confine ourselves to an analysis of our two-step
example, which is sufficient to reveal the logical problems of multi-step SSAs (see
Fig. 5).

123
Health Care Anal

Fig. 5 The structure of multi-step slippery slope reasoning under uncertainty

(1) According to the argument, stepping on the slope (i.e., legalizing voluntary
euthanasia) is likely to lead to the intervening step of legalizing non-voluntary
euthanasia (in Fig. 5 represented by i). However, since legalizing non-voluntary
euthanasia is only one possible outcome, stepping on the slope must have other
possible outcomes also (:i in Fig. 5). For instance, voluntary euthanasia might
become a legal practice without taking us down the slope. A proponent of the
argument above will doubt this, but the fact that he assigns the intermediate step
only a certain probability logically commits him to admit that it may have other
(positive) outcomes also; and, as I have already noted, an arguer needs to estimate
the value of these other possible outcomes also and take their value into
consideration.
(2) Thus far, our example was not different from one-step reasoning under
uncertainty. But the example claims that sliding to the intermediate step may take us
down to the bottom of allowing involuntary euthanasia. Once more, if it is possible
that we slide to the bottom, it is also possible that we do not slide there (as shown in
Fig. 5 by the branches leading out from the second fork).26 It is, for instance,
possible that involuntary euthanasia remains illegal and only voluntary and non-
voluntary euthanasia are permitted. In any case, a proponent of the argument under
consideration has to admit this possibility and he has to estimate the (dis)value of
this possible outcome, weighted with its probability.
Let me pull this together. In (1) it was claimed that legalizing voluntary
euthanasia is likely to lead to the intervening step ‘‘non-voluntary euthanasia’’ (Pi),
but may have other outcomes also (P:i). (The symbol ‘:i’ stands for ‘We do not
slide to the intervening step.’) Estimating the value of legalizing voluntary
euthanasia requires now combining the expected value of this intervening step ‘i’
(that is, EVi) and the expected value of the other possible outcomes—in our
example, not sliding to the intermediate step (EV:i). For the sake of brevity and
clarity suppose that the arguer estimates this value by adding the products of the
probabilities times the values of the possible outcomes. The expected value of
legalizing voluntary euthanasia is then calculated by the following procedure:
Pi 9 EVi ? P:i 9 EV:i.
However, since the intermediate step makes sliding to the bottom only likely and
has therefore other possible outcomes also, we need to estimate its expected value
26
I exclude here the case where this possibility is 1 because if P = 1 we have a reasoning under
certainty.

123
Health Care Anal

(EVi) too. I have already hinted in (2) that this requires combining the value of the
bottom V(b) with the probability that we slide to the bottom given that we have slid
to the intermediate step P(b|i) and combining this with the value of not sliding to the
bottom V(:b) and its conditional probability P(:b|i). (The symbol ‘P(b|i)’, read
‘‘the probability of b, given i,’’ represents the probability of b occurring, assuming
that i has already occurred, or will occur.) Writing this in mathematical notation, we
obtain: EVi = P(b|i) 9 V(b) ? P(:b|i) 9 V(:b).
(3) Suppose now that the arguer has estimated the value of legalizing voluntary
euthanasia. No matter whether this value is positive or negative, this estimation does
not warrant drawing a normative conclusion because such an inference requires
comparing the value of this option to its alternatives; and, as I have already
emphasized, an arguer can only validly infer that we should not legalize voluntary
euthanasia if there is at least one alternative that is better than it.
The upshot of the preceding paragraphs is as follows: Focusing only on the
undesirable bottom is an error of reasoning because the bottom is only one possible
outcome of stepping on the slope. The value of stepping on the slope needs to be
estimated by combining the values of all possible outcomes (weighted with their
probabilities), and even this estimation does not warrant inferring a conclusion
because a valid inference requires a comparison of stepping on the slope with its
alternatives.27

Nonprobabilistic Slippery Slope Reasoning Under Uncertainty

There are degrees of uncertainty. In a biomedical context we are often facing


problems that are so vague that we cannot assign probabilities. That is the extreme
end of uncertainty. For instance, how likely is it that stem cell research will lead to
reproductive cloning, that gene therapy leads to eugenics, or that legalizing
voluntary euthanasia will take us down the slope to involuntary euthanasia? It is
clearly extremely difficult to settle here on a set of numerical probabilities and
therefore we may need to rely on nonprobabilistic decision criteria, which are based
solely on the value of the outcomes and ignore their probabilities.
This brings up the question of the logic of deciding in such circumstances. There
is much disagreement on this. Various logical criteria have been proposed, but none
has firmly established itself. Occasionally, one alternative dominates the others.
Roughly, that is to say that at least one of its outcomes is better than the outcomes of
its alternatives and no outcome of this alternative is worse than the other outcomes.
In such exceptional cases it is rational to choose this dominant option. But in most

27
Critics of slippery slope reasoning sometimes call attention to the fact that the probability of sliding to
the bottom is the smaller the longer the slope is (e.g., [5]). If we assume that we slide to the bottom only if
we step on the slope, this probability P(b) is P(i1) 9 P(i2|i1) 9 … 9 P(b|in). For instance, in the case of a
three-step slope with P(i1) = 0.4, P(i2|i1) = 0.6 and P(b|i3) = 0.3, the probability of sliding to the bottom
is only 0.072. It is worth noticing, however, that this is only indirectly relevant because it makes the value
of stepping on the slope less negative. It is not directly relevant because we do not compare the bottom to
the other possible outcomes of stepping on the slope, but stepping on the slope to its alternatives.

123
Health Care Anal

situations there is no option that dominates all alternatives and then a balancing of
alternatives is unavoidable.
The so-called Maximin Principle is often recommended for such cases. The
rationale underlying this principle is to find for each alternative the worst outcome
and then choose the option for which the worst outcome is best.28 Clearly, when
only the set of possible outcomes is known but no probabilities can be assigned, it
does not make sense to put more weight on one outcome than another. This gives
the Maximin Principle some initial plausibility since it holds that in such cases it is
reasonable to consider the worst outcomes only, and then choose the option for
which the worst outcome is the best.
Let us reconsider the example given at the beginning of the previous section
under the realistic assumption that the arguer holds that there is some risk of sliding
from voluntary to involuntary euthanasia but is unable to quantify this likelihood.
The reasoner may hold that the worst outcome of legalization is involuntary
euthanasia (the bottom of the slope), and from this he may conclude—apparently in
accordance with the Maximin Principle—that we should not legalize voluntary
euthanasia. Proponents of slippery slope reasoning have, I suspect, an affinity for the
Maximin Principle because it represents an extremely risk averse approach to
practical reasoning by looking only at the worst outcomes of each alternative and
ignoring the values of its other outcomes. But it is exactly this characteristic that has
caused much criticism in the literature. The exclusive focus on the worst outcomes
makes it all too easy to construct examples for which the Maximin Principle gives
unreasonable results.
Moreover, it must be emphasized that slippery slope reasoning is not even in
accordance with this controversial principle. Since the Maximin Principle tells us to
choose the option where the worst outcome is the best, it requires a comparison of
alternatives. But as I have repeatedly pointed out in this study, by arguing that
stepping on the slope will (or may) lead to the undesirable bottom and inferring
from this that we should not step on the slope, slippery slope reasoning falls short
even of this minimum requirement of practical reasoning.
Let us take stock. This unremittingly negative discussion has shown that slippery
slope arguments are logically fallacious. The (putative) fact that stepping on a slope
will (or may) start us down a slope does not allow deriving the conclusion that we
should avoid stepping on this slope because practical reasoning requires a
comparison of alternatives and we can only validly infer a normative conclusion
as a result of such a comparison. Contrary to how it may appear, however, I do not
deny the importance of emphasizing the danger that adoption of certain policies
may take us down a slope to a state of affairs that we want to avoid. Authors often
assume, or write as if they assume, that such concerns are unrealistic. In my view,
however, there is often a considerable risk of sliding down a slope. Nonetheless, this
possible slide is never more than one thread of a practical deliberation. Ignoring this
is one of the greatest hindrances to good practical reasoning.

28
The term ‘‘maximin’’ is derived from the Latin term ‘‘maximum minimorum’’ which means ‘‘the
maximum of the minima’’—and it is exactly this maximum that should be chosen, according to this
principle.

123
Health Care Anal

References

1. Bayertz, K., Paslack, R., & Schmidt, K. W. (1994). Summary of ‘‘gene transfer into human somatic
cells. State of the technology, medical risks, social and ethical problems: A report.’’ Human Gene
Therapy, 5, 465–468.
2. Black, M. (1964). The gap between ‘is’ and ‘should’. The Philosophical Review, 73(2), 165–181.
3. Bostrom, N. (2005). In defense of posthuman dignity. Bioethics, 19(3), 202–214.
4. Bredenoord, A. L., Pennings, G., Smeets, H. J., & de Wert, G. (2008). Dealing with uncertainties:
Ethics of prenatal diagnosis and preimplantation genetic diagnosis to prevent mitochondrial disor-
ders. Human Reproduction Update, 14(1), 83–94.
5. Camporesi, S. (2007). The context of embryonic development and its ethical relevance. Biotech-
nology Journal, 2, 1147–1153.
6. Clark, M. (1998). Euthanasia and the slippery slope. Journal of Applied Philosophy, 15(3), 251–257.
7. Clarke, D. S. (1985). Practical inferences. London: Routledge & Kegan Paul.
8. Codd, H. (2007). The slippery slope to sperm smuggling: prisoners, artificial insemination and human
rights. Medical Law Review, 15, 220–235.
9. Den Hartogh, G. (2001). The Slippery Slope Argument. In H. Kuhse & P. Singer (Eds.), A Com-
panion to Bioethics (pp. 280–290). Oxford: Blackwell.
10. Enoch, D. (2001). Once you start using slippery slope arguments, you’re on a very slippery slope.
Oxford Journal of Legal Studies, 21(4), 629–647.
11. Gay-Williams, J. (1996). The Wrongfulness of Euthanasia. In R. Munson (Ed.), Intervention and
Reflection. Basic Issues in Medical Ethics (5th ed., pp. 168–171). Belmont: Wadsworth.
12. Hansen, J. E. (2005). A slippery slope: How much global warming constitutes ‘‘dangerous anthro-
pogenic interference’’? Climate Change, 68, 269–279.
13. Levinsky, N. G. (2000). Organ donation by unrelated donors. New England Journal of Medicine,
343(6), 430–432.
14. Lewis, P. (2007). The empirical slippery slope from voluntary to non-voluntary euthanasia. Journal
of Law, Medicine & Ethics, 35, 197–210.
15. Lillehammer, H. (2002). Voluntary euthanasia and the logical slippery slope argument. Cambridge
Law Journal, 61(3), 545–550.
16. Lode, E. (1999). Slippery slope arguments and legal reasoning. California Law Review, 87, 1469–
1543.
17. Nielsen, W. H. (1987). The slippery slope argument against the legalization of voluntary euthanasia.
Journal of Social Philosophy, 18(1), 12–27.
18. Sandel, M. J. (2004). Embryo ethics—the moral logic of stem-cell research. New England Journal of
Medicine, 351(3), 207–209.
19. Schauer, F. (1985). Slippery slopes. Harvard Law Review, 99(2), 361–383.
20. Schubert, L. (2004). Ethical implications of pharmacogenetics—do slippery slope arguments matter?
Bioethics, 18(4), 361–378.
21. Singer, P. (2003). Voluntary euthanasia: A utilitarian perspective. Bioethics, 17(5–6), 526–541.
22. Smith, S. W. (2005). Evidence for the practical slippery slope in the debate on physician-assisted
suicide and euthanasia. Medical Law Review, 13, 17–44.
23. Smith, S. W. (2005). Fallacies of the logical slippery slope in the debate on physician-assisted suicide
and euthanasia. Medical Law Review, 13, 224–243.
24. Spielthenner, G. (2008). The logical assessment of practical reasoning. SATS–Nordic Journal of
Philosophy, 9(1), 91–108.
25. Van der Burg, W. (1991). The slippery slope argument. Ethics, 102(1), 42–65.
26. Volokh, E. (2003). The mechanisms of the slippery slope. Harvard Law Review, 116(4), 1026–1137.
27. Wright, W. (2000). Historical analogies, slippery slopes, and the question of euthanasia. Journal of
Law, Medicine & Ethics, 28, 176–186.

123

Potrebbero piacerti anche