Professional Documents
Culture Documents
net/publication/262259980
CITATIONS READS
2 142
1 author:
Mehmet Elgin
Mugla Sitki Kocman Üniversitesi
13 PUBLICATIONS 156 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
How can A Priori Laws Contribute to the Empirical Content of a Theoretical System? View project
All content following this page was uploaded by Mehmet Elgin on 19 November 2014.
Abstract
Some philosophers argue that the legitimacy of special sciences depends on
their containing laws and that only laws these sciences can have are ceteris
paribus laws. I argue that there is a minimal condition for ceteris paribus
laws and this minimal condition states that if “ceteris paribus A’s are B’s”
express a lawful relation, then “ceteris paribus A’s are B’s” and “ceteris
paribus A’s are not B’s” should not both be true. I argue that if there are
genuine ceteris paribus laws, then ‘ceteris paribus’ should be a proper part
of a law statement in which it occurs. However, if we interpret ‘ceteris
paribus’ in this way, the containing sentence cannot be a law because it does
not satisfy a minimal condition for ceteris paribus laws. Consequently, I
argue that even if there are ceteris paribus sentences, it is unlikely that there
are ceteris paribus laws.
Introduction
With few exceptions, philosophers think that general propositions
in the special sciences and general propositions in physics and chemistry
have important differences. While at least some of the generalizations in
physics and chemistry satisfy the universality condition that laws require,
it is thought that generalizations in the special sciences don’t satisfy this
condition—i.e., for any special science generalization there is always
some defeater. If it is true that laws don’t have exceptions, then it looks
as if the special sciences don’t have laws. For some philosophers
(Stephen Schiffer (1991), James Woodward (2000, 2001, 2002), Earman
& Roberts (1999), Earman, Roberts & Smith (2002)), this conclusion is
correct (never mind the argument), but others find it quite difficult to
accept. For they think that the following argument (or something like it)
is correct:
1. At least sometimes the special sciences (Biology, Economics,
Psychology, Sociology, etc.) provide good explanations.
2. Providing good explanations requires citing laws.
3. Therefore, there are laws in the special sciences.
4 Special Sciences and Ceteris Paribus Laws
© Φ Philosophical Writings
Mehmet Elgin 5
not held fixed, then the counterfactual “if there was an extra planet in the
solar system, it would describe an ellipse” would come out false. Then,
the point is that if what is held fixed varies from context to context then
even strict laws may violate the minimal condition, so the minimal
condition may not be necessary for cp laws after all (I owe this point to a
anonymous referee). The point, indeed, explains why cp laws that fail to
satisfy the minimal condition also fail to support counterfactuals. I think
the anonymous referee want to argue that since even strict laws would
violate the minimal condition, it should not be necessary for cp laws
either. However, strict laws would not fail to support counterfactuals.
The reason is this: when we formulate strict laws, say law of inertia, their
antecedent conditions don’t vary from context to context. For example,
we can formulate law of inertia and law of gravitation as follows: “if
there are no forces acting on the object, then…” and “if the gravitational
force is the only force acting on the system then…”. In these examples,
it is clearly specified which possible world we live in, but in the case of
cp laws this is not the case, and this is why they fail to support the
minimal condition. My point is this: there is a way of formulating strict
laws in a way that specifies exactly which possible world we live in and
this formulation would safe them from failing to support the minimal
condition, but it is impossible to do the same with cp laws.
Hausman (1992, 134-135) argues that the cp clause picks out
different things in different contexts. However, the minimal condition
entails that such sentences cannot be laws. For, if the cp clause picks out
different things in different contexts like “I”, then the minimal condition
should not be introduced as a necessary condition. I have offered
independent reasons why the minimal condition should be accepted as
being a necessary condition for any law (so in particular for ceteris
paribus laws). In any case, if the cp clause is like “I”, then there is a
question about whether indexical terms can appear in laws. It may be said
that “it” in “if there are no forces acting on the object, then it will
continue its motion with uniform velocity if it is in motion and if it is at
rest then it will remain at rest” is also an indexical term but this does not
prevent it from appearing in this law. However, when we symbolize this
sentence in quantifier logic, we symbolize ‘it’ in the consequent as ‘x’
that ranges over all objects, so ‘x’ that represents ‘it’ in the consequent of
this conditional occurs as a bound variable. Therefore, ‘it’ in the above
sentence does not really behave like an indexical term. Now consider the
following sentence: “if everyone attends, I will be surprised”. In this
example, the main connective is the conditional and ‘I’ in the consequent
© Φ Philosophical Writings
Mehmet Elgin 7
is a constant not a variable. If the cp clause is like the latter kind, then its
appearance in law statements will be problematic since laws are
supposed to hold invariantly. The other problem is that referents of
indexical terms are determined, in part, by what the speaker intends to
refer to in using these terms, but this should not be the case with laws.
© Φ Philosophical Writings
8 Special Sciences and Ceteris Paribus Laws
© Φ Philosophical Writings
Mehmet Elgin 9
say that glass is fragile and that this feature is not readily lost. To say
that iron filings will arrange themselves around a bar magnet in a
specified pattern, cp, is to say that magnets exert a certain sort of force
on iron filings, a disposition magnets do not loose while remaining
magnets (Lipton 1999, 164).
Those of us raised in the joint shadow of the Vienna Circle and British
Empiricism were taught that these kinds of concepts [power or capacity
concepts] must not appear in science. I was puzzled about this from
early on since it seemed to me that many of the most important concepts
I learned in physics are power or capacity concepts, force being the
first, simple example (Cartwright 2002).
© Φ Philosophical Writings
Mehmet Elgin 11
© Φ Philosophical Writings
Mehmet Elgin 13
the traditional sense) makes it the case that x will not be time
translationally invariant (Sober 1999). Let’s suppose that there are two
physical states that realize M (this is just to keep things simple, the same
argument will apply to the cases where M is realized by more than two
states); then, Pr(B | M) = Pr(B | M&P1)Pr(P1 | M) + Pr(B | M&P2)Pr(P2
| M). If Pr(B | M& Pi) Pr(B | Pi) (which Sober says captures the idea
that physics is causally complete), then Pr(B | M) = Pr(B | P1)Pr(P1 | M)
+ Pr(B | P2)Pr(P2 | M). It isn’t plausible to think that Pr(P1 | M) or Pr(P2
| M) will be the same at all times since individuals with M at one time
might mostly have P1 while at other times they might mostly have
P2. Given that these probabilities change values through time, there are
only two ways that Pr(B | M) will be time translationally invariant. One
is that Pr(B | P1) = Pr(B | P2). However, it isn’t plausible to think that
this exact equality obtains. For example, let B = S will go to the movies
tonight and let M = S believes this afternoon that a movie is playing and
wants to go. If P1 and P2 are two physical realizations of that
belief/desire pair, then Pr(B | P1) won’t be the same as Pr(B | P2) since
P1 might be a physical state that would make the behavior quite unlikely
(because it also makes a heart attack probable) while P2 might be some
other physical state that would make the behavior likely. On the other
hand, if Pr(B | P1) and Pr(B | P2) differ in value, then for Pr(B | M) to be
invariant, they must change in value in such a way that their changes and
those of Pr(P1 | M) and Pr(P2 | M) are all “coordinated” so that Pr(B | M)
remains constant through time. This is even more implausible. Thus,
given these probability relations and the traditional understanding of the
multiple realization thesis, Pr(B | M) will not be time translationally
invariant.
© Φ Philosophical Writings
14 Special Sciences and Ceteris Paribus Laws
highly dubious that there is a unique property that ceteris paribus in all
cp claims picks out. It should be clear that the ceteris paribus clause
refers to different things when the contents of two statements that employ
the cp clause are different. Consider, for example, “ceteris paribus
smoking causes lung cancer” and “ceteris paribus human behaviors are
goal oriented”. Cp refers to different things in these two examples. It is
difficult to reconcile the minimality condition and the fact that the cp
refers to different things in different contexts. The former requires more
constraints on the cp clause while the latter requires less.
Lipton, and Cartwright try to find a unique property that governs
the concept of ceteris paribus in all contexts. They suggest that ceteris
paribus in “cp if A then B” refers to the ideal conditions under which the
lawful relation between A and B holds. However, there are no uniquely
defined conditions that are ideal. Since ideal conditions will vary in
different sciences, ‘ideal conditions’ will pick out different properties in
different sciences. This means we are back to the problem discussed in
the previous paragraph.
Either the accounts of cp laws satisfy the minimum condition or
they don’t. When they satisfy this condition the trivialization result
follows. When they don’t satisfy this condition, it is dubious that such
statements be laws (for reasons discussed in section 2). Consequently,
given the minimum condition and the nature of cp statements, it is highly
dubious that we will ever find cp laws.
I can imagine certain people protesting by saying that one has to
start somewhere. Unlike physics, other special sciences are still in the
process of development. So, it may be that physicists have a pretty good
idea how to state ideal circumstances but scientists in other special
sciences are still trying to figure out how to state ideal circumstances and
laws that apply in such circumstances. And, so-called cp laws are perhaps
the predecessors of real laws in these sciences. Earman, Roberts and
Smith (2002) discuss this possibility. Doesn’t it make sense to use cp as a
placeholder in those generalizations even though we are not sure how to
describe the ideal circumstances in which they apply? This may be
methodological advice. However, the problem with cp laws is not
whether cp is useful in discovering real laws, but whether there are cp
laws. Thus, even if it turns out that it is a good strategy to use cp as a
placeholder in generalizations that may turn out to be laws, this does not
answer the question of whether there are cp laws. This may be an
argument why one needs it for epistemic reasons, but it does not say
© Φ Philosophical Writings
Mehmet Elgin 15
anything about whether there are laws where cp is itself a part of the law
statements.1
Bibliography
1
I thank Elliott Sober for reading earlier drafts of this paper. I also thank Ellery Eells,
Berent Enç and Dan Hausman for reading the very early draft of this paper.
© Φ Philosophical Writings
16 Special Sciences and Ceteris Paribus Laws
Mehmet Elgin
Muğla Üniversitesi
Felsefe Bölümü
Kötekli/Muğla
Turkey
melgin@mu.edu.tr
© Φ Philosophical Writings