Meaning, Extension, and Dispositions
Abstract
Semantic externalism maintains that the meaning of an expression may
not always be determined by a description (or a cluster of descriptions) a
speaker associates with the expression, or, more generally, what is in
speaker’s mind (narrowly considered), but that the physical and social
environment may also play some role in it. The discussion has focused on proper
names and natural kind terms. Externalism with respect to general terms is
apparently much more controversial than with proper names (see e.g. Unger 1983),
and many seem to be rather skeptical about the whole metaphysics of natural
kinds assumed by the standard arguments for externalism (e.g. Mellor 1977,
Crane 1981, Dupré 1981, Unger 1983, Bach 1987). It has
been argued that one can find genuine natural kind terms, if anywhere, only in
science, not in ordinary language, and that in ordinary language general terms
such as ‘tiger’ or ‘fish’ do function (roughly) as descriptions (e.g. Dupré 1981). One has repeatedly replied to
Putnam’s Twin Earth thought-experiment by the suggestion that ‘water’ had, when
nobody knew the chemical constitution of water, in its extension both H2O
and XYZ. Apparently the discussion has come to a deadlock.
The arguments of Kripke, Putnam and others rest on the intuitively
plausible realist idea that perceptual seemings are
not necessarily as
fine-grained as the reality that causes them. As a clear opposite of this idea, recall that in the hey day of logical
positivism one subscribed the translatability
thesis, the requirement that all cognitively meaningful talk must be
explicitly definable in observational terms, and consequently, the idea that
every meaningful term is thus synonymous with some observational description. One
can actually view this as a version of descriptivism. This thesis was given up as a consequence of Carnap’s
(1936) observation that dispositional
terms do not admit of explicit definitions in terms of observational terms,
but nevertheless are clearly cognitively meaningful. This conclusion also rebuts
the corresponding variant of descriptivism.
A more general perspective now
suggests itself: what if we temporarily put aside the disputed natural kind
terms and consider instead dispositional terms and look if this could shed some
new light on the internalism/externalism-debate. Dispositional terms are
general terms which have certain virtues when compared to natural kind terms.
First, ordinary language contains indisputably many dispositional terms. Second,
there is now a rather wide agreement on certain basic question concerning the
metaphysics of dispositions. Therefore
they provide a less controversial ground for considerations in the philosophy
of language.
One standardly distinguishes two opposite
views on dispositions, phenomenalism
and realism. According to phenomenalism
(e.g. Ryle 1949), dispositions are nothing over and
above the holding of certain conditionals – or, for our present purposes, the
disposition ascription is synonymous with the associated conditional (or,
perhaps, a cluster of conditionals). Essential for phenomenalism
is the claim that dispositional ascription says absolutely nothing about e.g. a
class which is said to be fragile, when it is not dropped. In other word, when
a fragile class a and
a non-fragile class b are not being
dropped, there need be no factual difference between them. Realism disagrees. It claims that if a is fragile and b is not, they must differ in some
factual respect whether they are being dropped or not. The essential content of
realism is often expressed by saying that dispositions must have bases, that is, there must be a factual property (or property-complex)
possessed by an item causally active in bringing about the manifestation of the
disposition. Phenomenalism denies this claim.
Although phenomenalism
was once very popular, it is nowadays hard to find anyone defending it, and
denying the existence of factual bases of dispositions. It is generally
considered as inadequate. Realism, in some form, seems to have become the received
view on dispositions. Indeed, phenomenalism faces
several difficult problems. First, there is the problem of so-called multi-track
dispositions, that is, the fact that the same disposition may be manifested in
many different ways. The simple conditional approach entails that different
dispositions corresponding different conditionals – but this is highly implausible!
Further, Armstrong (1973) has presented the compelling principle that for every
true contingent proposition there must be something in the world which makes
the proposition true. Phenomenalism, however,
contradicts this intuitive thesis.
Realism, in itself, does not yet
rule out the view that the meaning of a dispositional term is nevertheless
given by its corresponding conditional, or at least that the conditional
correctly determines the extension of the dispositional term. There are,
however, arguments based on the possibility that a disposition is untestable even in principle, which undermine all
conditional accounts. The most well know seems to be C.B. Martin’s (1994)
thought-experiment of ‘electro-fink’. It shows that the simple conditional
account is wrong. The meaning of a
dispositional term is not given by the conditional traditionally associated
with it.
Some, e.g. Lewis (1997) and Mumford (1998), have developed more sophisticated
conditional approaches in order to deal with Martin’s arguments. Fara (2001), however, argues convincingly that conditionals
are simply not suited to the task of stating the truth conditions of
disposition ascriptions. His argument is based on what he calls “masking
problem”. That is, in Martin’s ‘fink’ case, the object is caused to lose the
disposition. But it is also possible that only the manifestation is being inhibited,
or masked, but that the object keeps its disposition. Mark Johnston (1992) gives an example of a
fragile class cup with internal packing to stabilize it against hard knocks. Even
thought the cup would not break if struck the cup is still fragile. The cup has an intrinsic property – weak
crystalline bonding, say – which “accounts” for its disposition to break (Fara 2001).
What is relevant to the theory of reference is
that the intrinsic property which accounts for a disposition may well be
unknown to the speakers – all they know is a way it may be in some occasions
manifested. It seems quite clear that
such an incomplete knowledge, and the associated conditionals, are not sufficient to determine the correct extension of the
dispositional term in question. But if so, dispositional terms may provide a
good case for semantic externalism.
Descriptivism comes in many forms,
depending on the constraints one puts for the admissible descriptions. And some
sort of constraint is necessary to assure that description
theories avoid the Fregean objections that plague
simple Millianism and motivate descriptivism (cf.
Braun 2003). The usual assumption that meanings must be learnable also sets
certain limits. It has been often required that the description uses only observational
predicates. It may be also required that the property expressed by the
description is purely general, or that the properties involve only individuals
with whom the speaker is directly acquainted. It has been traditionally assumed
that an admissible description is a truth-functional combination (usually, a
conjunction) of permissible predicates.
More generally, it may be required that a competent speaker is in
principle able to recognize the entities which belong to the extension of an
expression she understands, or that all there is to the meanings must be
manifested in observable use.
Whether or not something counts as a
strong argument against descriptivism obviously depends on many such choices. It
may be possible to liberalize descriptivism to the extent that it can avoid
practically any possible counterexample. But such unconstrained versions are
likely unable to satisfy the needs that originally motivated descriptivism. And
some natural constraints at place, it seems that the received wisdom on
dispositions seriously undermines descriptivism.
References
Armstrong, D. (1973), Belief, Truth and Knowledge.
Bach,
K. (1987) Thought and Reference,
Braun,
D. (2003) “Names and natural kind terms”, in: Ernest Lepore
and Barry C. Smith (eds.) Handbook of Philosophy of Language, Oxford University Press,
Forthcoming.
Carnap, R. (1936), ‘Testability and meaning’, Philosophy of Science 3: 439-53.
Crane, T.
(1991), ‘All the difference in the world’, The
Philosophical Quarterly 41,
1-25;
Dupré, J. (1981), ‘Natural kinds and biological taxa’, The
Philosophical Review 90, 66-90.
Fara,
M. (2001), Dispositions and Their
Ascriptions, Doctoral
Dissertation, Department of Philosophy,
Lewis, D. (1997), ‘Finkish
dispositions’, The Philosophical Quarterly 47: 143–158.
Martin, C. B. (1994), ‘Dispositions and
conditionals’, The Philosophical Quarterly 44(174): 1–8.
Mellor, D. H. (1974), ‘In defense
of dispositions’, Philosophical Review 83: 157–181.
Mellor, D. H. (1977), ‘Natural kinds’, British Journal for the Philosophy of
Science 28, 299-312.
Mellor, D. H. (2000), ‘The semantics and ontology
of dispositions’, Mind 109(436): 757–780.
Mumford,
S. (1998), Dispositions,
Prior,
E. (1985), Dispositions.
Ryle, Gilbert 1949) The Concept of Mind.
Unger, P. (1983) ‘The causal theory of reference’, Philosophical Studies 43, 1-45.