Академический Документы
Профессиональный Документы
Культура Документы
Ontological Reduction
Author(s): Dale Gottlieb
Source: The Journal of Philosophy, Vol. 73, No. 3 (Feb. 12, 1976), pp. 57-76
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2025896
Accessed: 28-12-2015 17:53 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Philosophy.
http://www.jstor.org
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
THE JOURNAL OF PHILOSOPHY
VOLUME LXXIII, NO. 3, FEBRUARY I2, I976
4- * _+
ONTOLOGICAL REDUCTION *
57
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
58 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 59
Numbers," in The W/ays of Paradox (New York: Random House, 1966), pp.
199-207, p. 203.
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
6o THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 6i
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
62 THE JOURNAL OF PHILOSOPHY
And yet, despite L-S*, the proof of Pythagoreanism fails. To see the
flaw, we must carefully distinguish between predicates (i) which
have purely numerical extensions, and (ii) which require commit-
ment only to numbers. As an example of a predicate satisfying (i)
but not (ii), consider '(ny) (x is y's height in inches)'. This predicate
has a purely numerical extension, but its extension is nonempty only
if there are things with heights in inches. Thus it commits us to the
existence of nonnumbers (if we assert that its extension is non-
empty), even though its extension is a subset of N. Thus we must
tighten up our requirement for translation: the effective procedure
must deliver predicates that not only have purely numerical ex-
tensions but also require commitment only to numbers. And this
L-S* cannot in general do. If S is arithmetically definable, L-S*
works fine; otherwise it requires the use of predicates that require
commitment to precisely the same objects we were trying to avoid.5
And since the set of true sentences of our over-all theory of the
world is not arithmetical, L-S* will not provide a numerical transla-
tion of that set.6
Of course, we could provide a translation for the set of our beliefs
about the world, and even for its logical closure.7 And it might be
thought that this is enough. After all, it seems that we are committed
only to whatever must exist for our beliefs to be true. But this is a
mistake. Consider the case of number theory. Imagine that, as a
response to incompleteness, a search is instituted for new axioms.
From time to time new axioms are added. Now L-S* gives us a
method of translating number theory into set theory based upon the
current choice of axioms. As that choice changes the translation
changes, but we will never be at a loss to translate (assuming we
never adopt a set of axioms that is not effectively enumerable). Is
the set T of true sentences of our total theory of the world. T is not arithmetic
(since, if it were, the set of truths of arithmetic, which is arithmetic in T, would
also be arithmetic, and Tarski's theorem refutes that), and so neither is T*.
Now even if T* is not arithmetic, the predicates M(Pi) will have extensions in
Nn, as the construction shows. However, they will require commitment to what-
ever we quantify over in specifying T*. Since T* is arithmetic in T, this reduces
to the problem of specifying T, i.e., the set of true sentences of our total theory.
How de we specify that? Well, if we take as primitive the predicate 'x is a true
sentence of our total theory', it's easy, but then our predicates M(Pi) will contain
primitive semantic terms. So we Tarski-ize truth; but that requires the use of all
the primitive predicates of our total theory, and hence we are committed to all
the objects we sought to avoid.
I am very much indebted to Charles Parsons' patient correspondence, which
enabled me to get straight on the status of L-S*.
6Thanks are due to Timothy McCarthy for very helpful discussion.
7Assuming we believe only a finite number of sentences, the logical closure of our
set of beliefs is finitely axiomatizable and so comes under Kleene's theorem 38.
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 63
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
64 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 65
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
66 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 67
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
68 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 69
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
70 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 71
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
72 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 73
of an observation sentence for Sam, but (B) does not. And the
difference between (A) and (B) does not turn on Sam's ignorance
of the fact that the correlation is one-one. For to know that the
correlation is one-one, Sam needs to know that the five particle-
events with which we started "mark the vertices of a full-fledged
four-dimensional hypersolid." Depending upon what particle events
are chosen, it may require observation, instrument theory, deep
physical theory, high-powered mathematics, etc. to certify that
choice. Even if Sam has this information, the fact that it is needed
in order to come to know (B) gives (B) an epistemic status far
removed from that of (A). It leaves (B) open to avenues of dis-
confirmation which are irrelevant to (A). A visual report of a non-
black crow can be challenged by appeal to lighting conditions and
the like; (B) can be challenged in addition by a critique of instru-
ment theory, etc.
The warp of epistemic status engendered by the need for the
knowledge that the correlation is one-one is not restricted to observa-
tion sentences. To know the translation of any sentence of the form
'(3x) , Fx' will require knowing that the correlation is one-one. If
knowledge of the original sentence does not require the relevant
science, its epistemic status will shift in translation by the need for
that science. Thus there will be sentences at many levels of our
knowledge-sentences known by observation, the projection of ob-
served regularities, so-called "analytic" truths-whose epistemic
status will be lost in translation. The result is that (C2) is violated
throughout our theory of the world.
Another instance in which epistemic status will be affected by
Harman's reduction is explanation; in particular, microreductions
will be lost. If the sentences that provide microreductions are robbed
of their ordinary referents, the explanation will normally be de-
stroyed. Take the molecular theory of gases as an example. When it
is coupled with the kinetic theory of heat, we get the familiar ex-
planation of the gas laws. Indeed, it is the fact that the molecular
theory of gases provides an explanation of the gas laws, instead of
merely having the gas laws as deductive consequences, that gives
such strong support to that theory. Many (I am among them)
believe that the explanatory power of this theory is due to its
enabling us to predict how gases will behave via a theory of the mate-
rial constituents of the gas. This is not to say that all microreductions
-i.e., all cases of deducing the behavior of an entity from the be-
havior of its parts-explain. (Think of explaining the flight of a
baseball into the stands by appeal to the flight of its elementary
particles to the same place.) Rather, it is held that in some cases and
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
74 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
ONTOLOGICAL REDUCTION 75
be applied to proposed reductions of considerable interest, in spite
of its dependence upon unresolved problems of epistemology for its
ultimate clarification. In addition, the case of lost explanation pro-
vides a starting point for the rejection of any Pythagorean reduction.
Recall that the violation of (C2) by the translation of the molecular
theory followed from (i) the necessity of interpreting the explanation
provided by that theory as a case of microreduction, and (ii) the
fact that in a numerical universe there are no microreductions. (i)
and (ii) together constitute a quite general argument against any
translation of physical theory into a theory whose ontology is purely
numerical or set-theoretical.
V. TOWARD A CRITERION FOR ONTOLOGICAL REDUCTION
Let's stop a moment to take stock. We have seen that (i) the criteria
for ontological reduction advocated by Goodman and Quine do not
succumb to blanket Pythagoreanism, but nevertheless (ii) they are
incorrect, sanctioning as they do the unicorn-reduction of numbers
to sets. Furthermore, (iii) their failure is shared by any extensional
criterion, since (iv) (C2) is a necessary condition for the acceptability
of a reduction, and (C2) appeals to nonextensional features. Having
thus cleared the field of the errors of our forebears, what can be done
toward providing an acceptable criterion-i.e., a condition which is
both necessary and sufficient and which will integrate into an ac-
count of the ontological consequences of the successful reductions?
We already have (C2) as a necessary condition; so it is a natural first
step to try it out as sufficient as well. Unfortunately, some sticky
questions immediately appear whose treatment I can only hint at
in the briefest way.
First, we must decide what feature of ontological reductions we
want (C2) to be sufficient for. We must be careful not to confuse the
legitimacy of a reduction with its over-all desirability. We may admit
the legitimacy of various set-theoretical reductions of number
theory, and yet reject them because set theory is more poorly under-
stood than number theory. We may admit the legitimacy of a re-
duction of events to objects, properties, and times, and yet reject
the reduction because we abhor properties. The legitimacy of a
reduction of Fs to Gs is what entitles us to claim that "for the
purposes at hand . . . no commitment to [EFsbeing different from
Gs] is necessary." To conclude that we ought, all things considered,
to drop the Fs in favor of their G-counterparts requires in addition a
careful consideration of "all things" i.e., the impact this would
have on the simplicity, unity, explanatory power, etc. of our total
theory. Obviously, (C2) is not a sufficient condition for the over-all
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions
76 THE JOURNAL OF PHILOSOPHY
This content downloaded from 161.23.84.10 on Mon, 28 Dec 2015 17:53:09 UTC
All use subject to JSTOR Terms and Conditions