The following chapter will look at the `new theory of reference' originated by Kripke and Hilary Putnam, and attempt to show its relevance for a theory of essences and natural kind terms. I will briefly note Kripke's criticism of Russell's theory of descriptions when applied to names, and then move to the issue of how this affects essences.
Today the biggest challenge for any notion of real essences is the claim that
essences are relative to a certain theory in which they are described or to
a conceptual scheme, hence that they do
not give rise to de re necessities.[i] Dunlop claims that:
Essentialism is a doctrine which many philosophers find repugnant. It runs counter to a number of fashionable ideas. One is the idea that the only way of taking modality which makes it non-mysterious is to treat it as arising out of linguistic convention. On this view the modal values of an object's properties vary with the descriptions under which we are considering the object. Yet the essentialist takes the essence-accident distinction to be non-variable (p. 68).
So the objection is that essences only exist relative to a conceptual scheme; this is taken to show that necessities are imposed by the `mind' on the `things' of the world. As Quine put it, "necessity resides in the way in which we say things, not in the things we talk about".
I will defend Kripke and Putnam's
theory of essences and natural kinds. They
argue that essences are discovered, but that their a posteriori status
does not render them contingent rather than necessary. Our scientific
discoveries tell us about how things in the world really are. More precisely, I
will defend this view of essences against two influential versions of the
challenge described above: First, that scientific knowledge is defeasible,
hence cannot give us the kind of necessities that Putnam and Kripke claim.
Secondly, that psychological states, conceptual schemes, or theories determine
the modality of the world. The actual world in itself is modally innocent. If
either claim is true, then there is no sense to the claim that some things are
more necessary or essential than others.
Bertrand Russell's theory of reference (as spelled out in Introduction to Mathematical Philosophy) is meant to show how names can be analyzed into definite descriptions. The need for this theory stems from Russell's word- object (or `Fido'- Fido) theory of reference. On this account, in order for a word to refer to something, there must be an object associated with it. So in saying `dog' we can see that the `semantic value' of this term is the object it names (Luntley p. 32). Yet, as Russell knew, there are many terms that are meaningful but do not refer to anything. Take for instance the famous proposition below:
(1). The King of France is bald.
On Frege's account of logical form we would get Fa for this proposition; this way of uncovering the form includes the name that is in dispute. So, what are we to make of this proposition, given that there is no King of France? Furthermore, Russell accepted the principle of bivalence, which he called the law of excluded middle, which would seem to imply that either the sentence is true or it is false[ii]In (2') the negation has narrow scope, and is not the negation of (1); but in (2'') the negation has wide scope, since it applies to the claim as a whole. It is also true, hence shows that the law of excluded middle is not in doubt (cf. Alexander Miller sec. 2.8). .
Russell's solution was to argue that (1) can be translated as the conjunction
of the following claims:
(1.1). There is a King of France.
(1.2). At most one person is the King of France;
(1.3). And any such person is bald.
This shows that (1) is false since there is no King of France. Russell's solution also applied, he thought, to proper names and general terms: they too are disguised descriptions (1912 p. 54). For instance, `Pegasus flies' would seem to have the same logical form of (1) above appears to have the form Fa . But, given the analysis above, `Pegasus flies' can be shown to be false, since there is no Pegasus (even though in mythology he flies).
Kripke argues that Russell's account of the sense of names; consider his
example from Naming and Necessity
(1980 p. 6-7):
(2). Aristotle was fond of dogs
On Russell's account we would analyze this into three other statements that do not contain the name— hence, provide a way of showing the `meaning' of the explanans without including the explanandum. Let us suppose that `Aristotle' abbreviates the description `the last great philosopher of history', then on Russell's account we get the following:
(2.1) There is at least one last great philosopher of antiquity.
(2.2) At most one person was last among the great philosophers of antiquity;
(2.3). Any such person was fond of dogs.
The first claim is an existential onethat the thing described exists; the second claim is a uniqueness claim; and the third is the claim about what is predicated of the person-- `fond of dogs'. So, in uttering a sentence with a proper name, one will also (implicitly) be committed to the three descriptive assertions above; if one of the conjuncts is false, then the sentence as a whole will be too. Of course, in (1) above, the first description would fail: there exists no King of France. But in (2) everything seems to be in order there was at least one last great philosopher of antiquity, there was only one person who was the last great philosopher of antiquity, and he was fond of dogs. Therefore, it would seem, `Aristotle' refers to something and has been used correctly (that is, we know the sense of `Aristotle') in the above proposition.
Kripke argues that the description theory of names is false because, while it
suffices to give us a true understanding of the extensionally correct
conditions under which (2) would be true, it does not explain the truth of (2)
in conditions under which a counterfactual course of history, resembling the
actual course in some respects but not in others, would be correctly
described (1980 p. 6). Kripke says: "With respect to a counterfactual situation where someone other than Aristotle would have been the last great philosopher of antiquity, Russell's criterion would make that other person's fondness for dogs the relevant issue for the correctness of [2]!" ( 1980 p. 7). While the actual truth conditions of (2.1- 2.3) agree extensionally with (2), the truth conditions of the statement could vary since (2.1- 2.2) could pick out another person who was the last great philosopher of antiquity. If it is possible that we could find out that Aristotle was not the last great philosopher of antiquity, then there seems to be a problem with saying that we can identify a thing with its descriptions; we would seem to need a way of referring to Aristotle, or picking him out, in situations where some of his descriptions were changed.
Now we will outline Kripke's rigidity thesis, his notion of necessary truth,
and his causal theory of reference, before proceeding to essences. Kripke
claims that there is a problem with Russell's analysis of names as disguised
descriptions. Kripke offers a theory which provides a way of picking out
Aristotle from any other persons who may be descriptively similar to him. I
will outline his thesis of the rigid designation of natural kind terms and
names, and show how a view of essences is implied in such a theory.
Kripke's rigidity thesis is based on an intuitive point about how names work.
So, if a name stands for something, that is, it picks out a real person or
thing, then it must do so `rigidly'. This is because identity statements, for
Kripke, are necessary truths. The proof is as follows (1977 p. 67):
First, by Leibniz' law, for any objects, x and y, if x is identical to y, then any properties (F) that x has y will too.
(i). (x) (y) [(x=y) e (Fx e Fy)].
Also, obviously, all objects are necessarily self- identical
(ii). (x) ~ x=x.
And by substitution with (i) we get,
(iii). (x) (y) (x=y) e [~ (x=x) e ~ (x=y)].
So, Kripke concludes, we are warranted in asserting the following,
(iv). (x) (y) [x=y e ~x=y].
Kripke notes that since the existence of any particular thing is thought to be contingent, i.e. since it is possible that it might not have existed at all, (ii) could be false when it is applied to particular things. Kripke suggests that necessity here need only be interpreted weakly, so that in any world in which the object exists it is self- identical. But, as Kripke notes, even if we accept this weakened notion of necessity how do we reconcile this with (iv)? The description `inventor of bifocals' might not refer us to Ben Franklin. `The inventor of bifocals' could have been someone other than Ben Franklin. Kripke's point is that, while this descriptive property is contingent, if it is identical (substitutable) with Ben Franklin, then it is necessary since it is the same as saying that Ben Franklin is Ben Franklin. Kripke argues:
There is an object x such that x invented bifocals, and as a matter of contingent fact an object y, such that y is the first Postmaster General of the United States, and finally, it is necessary, that x is y. What are x and y here? Here, x and y are both Benjamin Franklin, and it can certainly be necessary that Benjamin Franklin is identical with himself (1977 p. 71).
Kripke claims that names and natural kind terms (which we will look at soon) are rigid designators, i.e., the objects/ kinds they signify are the same objects/ kinds in all possible worlds: "Let's call something a rigid designator if in every possible world it designates the same object, a nonrigid or accidental designator if that is not the case" (1980 p. 48). So, a rigid designator is a name that identifies the same person or thing across possible worlds or counterfactual situations. Kripke continues:
All I mean is that in any possible world where the object in question does exist, in any situation where the object would exist, we use the designator in question to designate that object. In a situation where the object does not exist, then we should say that the designator has no referent and that the object in question so designated does not exist (1977 p. 79).
It might seem that a description could be a rigid designator too, if we
stipulate that it refers to a certain person. So, we could use the description
`the man who won the US election in 1968 ` to refer to Richard Nixon in a rigid
manner. Here we intend the name to be interchangeable with the description; of
course, this is not usually done, since we often say that it is possible that
someone other than Nixon might
have won the `68 election. Here we would still want to refer to Nixon, while
`won the `68 election' refers to someone else.
To further explicate the necessity of individual identity, let us contrast
Kripke's view with David Wiggins' claim that sameness of material origin and
material make up is not a condition of identifying individuals in other worlds.
Kripke's point was that we need to be able to find (identify) the person we are
talking about when many of his qualities have been altered. Wiggins argues that
if we need more than the sortal specification f
which answers a `what is it question'
to identify the counterfactual object, then any of the descriptions
which identify the object will do; but no one description is favoured over
another, since any of them can be used to identify a certain object (Dunlop p.
83). So, while any of the descriptions can potentially be used to identify an
object, none are privileged. While this method of Wiggins' may help us find the object (say, Julius
Caesar), because it eschews material origin and make up we may lose the
identity of Caesar. We can identify Caesar by the `property' of having been
stabbed by Brutus; but if the only limit on counterfactual speculation is
Caesar's sortal property man,
then we may lose him in our conjectures. For instance, on Wiggin's account, it
seems possible to change enough of Caesar's properties so that he becomes
identical with a Brutus in another world. If we change enough of Caesar's
non-sortal, contingent properties, Caesar becomes identical with Brutus, and
vice versa. Yet, this would make Caesar's identity contingent, and we would
lose him as the person he really is, since he is now the same as Brutus in this
other world. So, Kripke is correct to argue that there must be some essential
property that helps us to identify individuals, in addition to the mere sortal f.
Kripke holds that names, if they refer at all, do so necessarily; names are rigid designators since (i)
identical objects are necessarily identical; (ii) true identity statements
between rigid designators are necessary; and (iii) identity statements between names with the same reference in an
actual language are necessary (Kripke 1980 p.4). Kripke says that (i) and (ii)
are self-evident theses of philosophical logic, but (iii) follows because of
our intuitions of what names do. Some philosophers, such as Frege, have held
that identity claims are between names and not the objects themselves. This
seems to be what Kripke calls a `metalinguistic' thesis: identity statements
such as "Cicero is Tully" are read as "`Cicero' and `Tully' are names for the
same person" (1977 p. 90). If this were
true, then Kripke's claim that names (as terms only) rigidly designate persons
would be contingent, since one of the names might turn out to signify someone
else. Kripke realizes that the latter case, if this is indeed how identity
functions, is false, for we could have used the two names to designate
different entities. To outline his counterintuition, Kripke presents a thought
experiment regarding `identity' and `schmidentity' (1980 p.108). Even if identity is only a metalinguistic relation
between names, such as `Tully' and `Cicero', Kripke claims we would still be
left with the problem of shmidentity which is between an object and itself. The
problem `is Tully shmidentical with Cicero' still arises apart from
considerations of the terms themselves. So, Kripke's intuition is that the
identity relation between a thing and itself keeps reappearing and, so,
requires an answer; such an answer he believes to have given with his theory of
rigid designation. When we know the identity is between Cicero and Tully, and
both are the same object in material and origin, then it is not the case that
the identity is contingent which it could possibly be if we only are relying
on terms alone.
Kripke claims that general or natural kind terms are closer to proper names
than heretofore thought. A general name, on Mill's account, could be analyzed
as a conjunction of serially necessary and jointly sufficient properties which
pick out the extension of the term; so, `humanity' would be `rationality',
`animality' and certain physical properties (1980 p. 128). But Kripke objects
that natural kind terms are not short for the properties a dictionary would
take to define them. Some properties can be asserted necessarily of kinds of things. Kripke claims that there are
essential properties which are discovered by scientific investigation. The
natural kind terms which signify these essences function as rigid designators.
First of all, though, we can fix the reference of a natural kind term by
pointing at a sample, or samples, of a kind of
thing. For example, if there is a bunch of samples of yellow metal on a
table, then it is possible to fix the reference of `gold' by pointing at those
samples. If someone were to ask `what is gold' it would be possible to answer
by pointing at a sample and saying `this'. (We will be looking more at
ostensive definitions shortly). This process can be termed the `baptism'. Of
course, as Kripke notes, some or all of the samples on the table might be
fool's gold, so there needs be a test to tell us if this stuff gold (p. 136).
By `gold' people usually mean the valuable substance that is scarce, in
contrast to iron pyrites which looks like gold but is not valuable or scarce.
Science can tell us what the essence of gold is (namely, atomic # 79) so that
we can identify it more strictly in the future. But all of this testing comes
after the baptizing of the substances, and is a matter of investigating the
nature of the baptized stuff.
Furthermore, the meaning of the initial
baptism (of `gold') can be carried on by others by a causal chain; when others
choose to refer to the same thing that the initiator did, then they are using
the term rigidly. The person must intend to use the term with the same
reference.
An important part of Putnam and Kripke's theory is the argument that
psychological states do not determine intension (meaning), because they do not
even fix extension. Putnam claims that traditional theories of reference have
taken meaning to be tantamount to a psychological state. There are two
assumptions of this view. Firstly, that one's knowing the meaning of a term is
just being in a certain psychological state[iii]. And
secondly, that the meaning (intension) of a term determines its extension. So,
on this view, the psychological states determine the extension of terms just as
much as the intensions do (p. 221). The argument can be outlined as follows
(where f and g are psychological states, and ext is an extension):
f=g e ext (f) = ext (g)
f=g
ext (f) = ext (g)
If the two relevant sorts of psychological states, f and g, are the same, they have the same extension--- they mean or signify the same (e.g.) extramental object(s). But what if we can have a person on Earth who uses `water' to mean a tasteless, colourless liquid, and another person on Twin Earth who also uses the term `water' to refer to a liquid with the same superficial properties. When scientists from Earth visit Twin Earth they find that what Twin Earthlings call water is not H2O, but another substance, XYZ. Here we would seem to have a case where the inhabitants in each world have the same psychological state when they use the term `water', but there is a different extension. What each person means to do is to refer to the substance that he is familiar with; but, then, it would seem that it is the actual extension that is determining the meaning, not the psychological state of the agent. So, (where S1 = mental state of Earthling; and S2 = mental state of Twin Earthling) by Modus Tollens:
(1) S1 = S2 e f=g
(2) f=g e ext (f) = ext(g)
But, ext (f) … ext (g).
So, f … g (& S1 … S2).
Putnam's thought experiment shows that the extensions can differ while the psychological states are the same. Two different persons can be in the same psychological statethink that `water' refers to the liquid stuff they are each familiar withbut actual have their words mean different things, since the meaning is determined ultimately by the extension, which differs in each world. Putnam rejects (1), since he has argued that the psychological states are the same, but that the extensions are different. These meanings are not inside but are (at least partly) outside our heads.
This theory of how we use natural kind terms can be further clarified by
looking at Putnam's claims about rigidity and indexicality. Putnam claims that
we can teach someone what we mean by a natural kind term in two ways: either
ostensively or descriptively (pp. 230-231). Whether the term is used rigidly,
or to abbreviate a bunch of descriptions, will make a difference to what is
signified. If we ostensively `define' it then we just point it out; but if we
take a natural kind term to mean a set of descriptive properties, then to
identify the substance is to find the proper set of descriptions. For instance, the term `water' on Earth
refers to the substance H2O, while `water' on Twin Earth refers to the substance XYZ. In these worlds
people respectively use the word `water' to refer to H2O
and XYZ. Both kinds of liquid have the
same superficial, or descriptive, qualities even though they are at bottom
different kinds. There seem to be two plausible ways to describe the situation.
We can either claim that `water' is world- relative in its meaning, in that
water means H2O
in this world and XYZ in Twin Earth, or that it means H2O in all worlds and the
stuff (XYZ) is not water. Putnam argues that the descriptive view does not
allow us to say that `water' is being used differently in each world, since the
descriptive properties that determine its use are the same. But this flies in
the face of the knowledge that `water' refers to one substance here and another
on Twin Earth. The theory which states that `water' rigidly designates H2O
in all worlds would seem to be the correct one, or at least the most plausible.
This is because the substance that `water' denotes on this World is different
from what `water' denotes on Twin Earth. `Water', then, would seem to be a
rigid designator since one means to signify by the word the liquid that he is
familiar with (H2O).
This signification is accomplished not by finding descriptive properties, or an
operational definition, but by pointing to an actual type of thing. Thereafter
it is a matter for empirical investigation to determine what its nature is.
We have seen how names and natural kind terms signify in a rigid manner. In
order to talk about a counterfactual person, say Nixon, we need to be able to
identify him in these other guisesthese contrary-to-fact stories. Of course,
there are only so many qualitative changes we can make before we are no longer
talking about the same person. (We would not want to hold Leibniz' reading of
the indiscernibility of identicals which seems to take every quantitative
change to be an essential change, or changing Nixon's eye colour would
constitute an essential change (cf. Hollinger, 1976). We find through
investigation that things have underlying structures which cannot be changed
without the thing going out of existence.[iv] Water's
structure of hydrogen and oxygen cannot be altered into XYZ without an
essential change occurring. Other superficial qualities can be altered, such as
the colour of water with food colouring, without affecting the essence. So, x
and y are essentially the same iff x and y have all the same essential properties.
But if x has only the same superficial properties as y, but a different
underlying structure, then x…y.
Besides the claim that certain internal properties are essential to a kind, we
could also claim that the causal origin imposes constraints upon how we can
think about a particular thing. So, now we will look at the essences of
individual things. For instance, if we did not know anything about the parents
of Queen Elizabeth II, we could suppose that her father was Harry S. Truman.
This is an hypothesis. But since we do know that she came from the Royal
Family, we can claim that this blue-blood origin is a necessary fact. Or, if a
table is made of a certain piece of wood, then it is necessary that it is made
of this piece of wood, and not of any other piece of wood, or ice. The critic
might want to claim that the facts of this thing's origin are contingent, and
not essential to it; the actual causal origin is merely a contingent fact. But,
according to Kripke, to imagine that this table is made of ice "is not to imagine this table as made of wood or ice, but rather it is to
imagine another table, resembling this
one in all external details, made of another block of wood, or even ice" (1980
p. 114). There just is no way to make sense of the claim that this
table had different origins than it did, without denying its identity. Simply the claim that it might
have turned out to be made of ice is not enough to weaken the necessary truth
that it is made out of a certain piece of wood and not any other
substance.
Putnam notes that this view of rigid designation originated by Kripke has
"startling consequences for the theory of necessary truth" (p. 232). Putnam
says that "once we have discovered the nature of water, nothing counts as a
possible world in which water doesn't have that nature" (p. 233). He explains
this necessity in terms of cross- world relations: if two persons who exist in
separate possible worlds are both five feet tall, then we can say that they
belong to the extension `same height as' (p. 232). Similarly, a liquid in W1
belongs to the `same -liquid relation' to a liquid in W2 if it is the same
substance as the liquid in W2. So,
the theory we have been presenting may be summarized by saying that an entity x, in an arbitrary possible world, is water [iff] it bears the [same- liquid relation] (construed as a cross-world relation) to the stuff we call `water' in the actual world (p. 232).
This relation is necessary because when we discover the essence of something we have discovered (part of) the identity of the thing. As Kripke states, scientific truths about the substance under consideration are not contingent but "necessary truths in the strictest possible sense" (1980 p.125). As we saw above, metaphysical necessity need not be equated with a priori necessity; we cannot just intuit, via our concepts alone, what is metaphysically necessary. Kripke and Putnam's theory allows us to have a strong sense of `necessity' that is based on a discovery view of essences. Putnam concludes:
Once we have discovered that water (in the actual world) is H2O, nothing counts as a possible world in which water isn't H2O. In particular, if a `logically possible' statement is one that holds in some `logically possible' world, it isn't logically possible that water isn't H2O (p. 233).
Of course I will not be following Putnam in claiming that it is not logically possible that water cannot be any other substance; rather, on my account, it is essentially impossible that water be anything other than H2O. Essentially necessity, as we will see in the next chapter, will rule out the conceivability that water = XYZ. So Putnam is right that it is not conceivable that water is not H2O.
We will now look at some criticisms of the Kripke- Putnam view of essences and
essential necessity. The most often leveled critique centres around the claim
that we can get necessary truths from empirical discoveries. The other
criticism is a broadly anti-realist one concerning the world as it exists apart
from our conceptual activities; there is no mind- independent modality for us
to find in the world. Rather, the extramental world is modally innocent.
Avron Polakow claims that the theory of rigid designation of essences is ill
equipped to tell us how necessity arises from defeasible, scientific truths. He
notes that the new theory of reference is a reversal of the Fregean theory of
proper names as modeled on how one would analyze general terms; now it is
general terms, such as natural kind terms, which are modeled after proper names
(p. 79). He also outlines Putnam's critique of the received doctrine that
meaning determines extension and that the meaning of a term is just being in a
certain psychological state. He argues that Putnam is a metaphysical realist,
since the import of his theory is that it is not the meaning we attach to the
word `gold' that determines the extension of the term but, rather, it is the
nature of gold itself that determines whether the term has been used correctly
(p. 80). Polakow explains Putnam's theory as follows:
Had the meaning of the term been determined by the theory we use to define what it is for something to be called `gold' then we have to acknowledge that theories which define `gold' in different ways are referring to different things. If we bound `gold' to an operational definition in a given theory or language, then a meaning change of the term `gold' could require a change in the extension of the term. The nature of gold is language- independent, theory- independent and mind- independent (p. 80).
For Putnam, qua metaphysical realist, meaning exists outside the head. But, Polakow argues, there is a problem with this account when it is examined more closely.
Polakow argues that, while the account of rigid designation of natural kind terms
offered by Putnam and Kripke does try to deal with a kind of defeasibility, it
omits an essential condition of defeasibility. It is this important sort of
defeasibility that "is crucial to the whole theory" (p. 80). Furthermore, the
case that he makes against Kripke and Putnam will point out the troublesome
relationship between modal necessity and scientific necessity. The former is "a
metaphysical notion which takes as necessary that which is true in all possible
world, and which hence could not be otherwise" (ibid). Scientific necessity, on
the other hand, "takes as necessary that which could not be false without a
complete scientific theory being false" (ibid). There is a tenuous
relationship, according to Polakow, between which of the two types of necessity
has priority in the metaphysical realist account: if the metaphysical necessity
has priority, then the claim that water= H2O
is not defeasible; but Polakow argues that this goes against the very nature of
defeasible scientific knowledge.
As we have seen, Putnam does not deny that his account of rigid designation
disallows claims of defeasibility which is a sine qua non of scientific
theory at least before the consequent (~
x=y) kicks in. He admits that we might
point to a glass of gin and say `water'. We could be mistaken here. There is
more to rigid designation and finding the `same-liquid -relation' than guessing
and pointing. He is not just saying that everything that has a same- L
-relation to this stuff is to be defined as water; we must know that this stuff
actually is water. There is a social
component to this process, since the same-L-relation has meaning within a
language (viz. It's not private). But this is not a majoritarian account of
designation, since the persons in question must be the appropriate members of society: only
certain members of the community have the competence to investigate stuff
scientifically and tell us what the natures of such kinds of stuff are. Yet,
there is another constraint that the scientists must take into account— namely,
that there is a hidden microstructure
that things have which can be discovered. So, even the answers of experts is
defeasible, since they could be wrong about the microstructure. Polakow notes that this division of
linguistic labour encapsulates two principles:
(i). The extension of natural kind terms is to be determined by experts for those sort of terms.
(ii). The experts are constrained by the assumption that a natural kind has a hidden microstructure (p. 82).
So, the experts must tell us ultimately what the microstructure is; such an account is tantamount to telling us what the natures and essences of things are. And once we have these microstructural identities we have knowledge of necessities about how things are. We discover what (e.g.) water really is, H2O. Of course we can conceive that water is really composed of something else; but to do this is to say that water has some other microstructure rather than H2O. But once we know that water is this other stuff, then we have knowledge of a necessary truth(which is an application of Kripke's principle: x=y e ~ x=y).
Polakow outlines Putnam's argument for the denial that water = XYZ to question
whether `water =H2O'
is defeasible. The first point is that water is found to be identical to H2O;
secondly, from Kripke's principle we get the claim that necessarily water is H2O.
>From the Twin Earth example, Polakow argues, it might seem that we could
add to the second claim that water = XYZ. And with Kripke's principle this
would give us necessarily water = XYZ. But it is this latter claim that seems
absurd, since it seems impossible that one thing could be two different
substances, or that one substance be another substance(p. 83). Putnam must
assert that XYZ could not be
identical with H2O,
which requires him to hold a general principle that one set of elements are not
another set (p. 84). Polakow calls this the principle
of structure identity:
Given two chemical structures A and B it is not possible that if they are indeed two structures that be they be identical (p. 84).
Polakow claims that this principle is itself defeasible. First, though, he argues that there are no grounds on which Kripke or Putnam can defend the principle as non-defeasible.
Such a principle cannot be defended by an appeal to the theory of rigid
designation of natural kind terms; such a tack would be the claim that what we
discover water to be in this world defines what other possible worlds are. Such
a claim would be about the meanings of `discover', `water' and `H2O'
(p. 86). But this would be an a priori (based only on the meanings of the terms
involved) claim which, Polakow argues, goes against the empirical spirit of
Putnam and Kripke's theory. Another claim is that it is sufficient to adopt
Kripke's "principle that actual identity implies necessary identity as reason
for accepting the principle of structure identity" (p. 86). But this will not
work either, Polakow argues, for it conflates epistemic and metaphysical
issues. On Kripke's principle we can only assert that if we know that water =H2O,
then it is necessarily so. We need to have epistemic grounds for asserting the
antecedent, hence from Kripke's principle alone we do not get the necessity
needed for shoring up the principle of structure identity.
Now Polakow offers a thought experiment designed to show how the principle is
defeasible. It is as follows:
Suppose that an exciting discovery is made by two ingenious experimenters, Morkelson and Miley, that under very unique conditions H2O and XYZ seem to react in exactly the same way, contrary to all the known laws of chemistry. All other differences usually encountered with these substances disappear in the unique circumstances under which these two substances are being observed. An experimental chemist, Rolentz, suggests that the identity H2O =XYZ is more apparent than real, and can be explained by the interaction of the measuring equipment with what is being observed. Only after further ingenious experiments where the equipment is varied in many ways is it suspected that the equipment is not responsible for the surprising findings. It is then that a theoretical chemist, Zweistein, offers a theory to account for the identity of H2O and XYZ. In his `Relativity Theory of the Elements' Zweistein claims that we no longer treat substances, including those which are pure elements, in the same way as we treated them before. Instead, given certain circumstances it turns out that unsuspected laws come into operation whereby certain substances become structurally indistinguishable. When we discover the apparent anomaly that H2O =XYZ, then we have a case in which those circumstances obtain (pp. 86-7).
So, in certain circumstances water (H2O) and XYZ are seemingly identical. If this experiment works, then Polakow has shown that the principle of structure identity is defeasible; from which it follows, he argues, that essentialism is weakened, because we cannot assume that the essence of a kind is its microstructure. So, if we cannot presuppose that essence is microstructure, we cannot assert that what science finds the structure to be is necessary unconditionally (p. 88).
Polakow's argument has been that the nature of defeasible, scientific knowledge
is such that it does not lend itself to the Putnam/ Kripke theory of rigid
designation. Contra Putnam and Kripke, it would seem that we cannot say that we
have discovered a necessary truth about microstructure, which can be taken to
be the essence of the thing. The identities which are scientifically
discovered, such as water =H2O,
are potentially false (or, as Polakow says, what is scientifically necessary is
what is improbably false). Polakow has attempted to show that we cannot move
from such defeasible truths, such as the principle of structure identity, to
the stronger claim that such identities are necessary. The point of the thought
experiment is to show that the principle of structure identity is defeasible.
If successful, Polakow's argument would throw into doubt the ability of rigid
designation to apply to natural kind terms and essences. This conclusion would
be troubling to the theory advanced here, since we will need the theory of
rigid designation of natural kind terms and essences in order to apply it to
logical possibility in the next chapter.
My own reply to all of this is to point out that we need not accept Polakow's
notion of radical defeasibility, but can accept instead a weaker notion of
defeasibility. Polakow argues that
scientific knowledge is radically defeasible. By this he means that such
knowledge is by its very nature possibly false; a scientifically necessary theory
is never more than improbably false (p. 80). It never extends as far as
impossibly false, which is the view he seems to attribute to Putnam and
Kripke's use of scientific knowledge. But, can we not be weak defeasibilists?
Such a position would be one that held that, while the principle of structure
identity is defeasible, it does not follow that all specific uses of it are.
For example, water= H2O,
or gold =atomic # 79. These findings are very certain. One would be hard
pressed to come up with good reasons for thinking that possibly water =XYZ
given what we know about it. The proponent of weak defeasibility might agree
that scientific knowledge and in particular the principle of structure
identity are, in general, defeasible, since we could be mistaken; but it does
not follow that water= H2O
is defeasible.
What the defender of Putnam can do is to ask for some reasons why the principle
of structure identity regarding (e.g.) water does not work. Simply imagining a
case where it does not, where water=XYZ, is not sufficient. We can ask for some
reasons why this particular scientific claim is radically defeasible, rather
than weakly defeasible. Such a question seeks to ask why we ought to doubt a
well- backed- up theory by simple conjecture.
Of course it could still be replied that I have not shown that the principle of
structure identity is a metaphysical necessary truth. I have only claimed that
we need accept weak defeasibility, a claim that could still shore up Polakow's
contention that the principle is defeasible; hence the necessity of Kripke's
consequent (~ x=y) does not
follow, because this necessity relies on the structure principle which does not
lend itself to metaphysical necessity. While there does seem to be a tenuous
relationship in Putnam's account between scientific and metaphysical necessity,
I think that one of Polakow's premises can be successfully challenged. His
argument is, in essence, as follows:
(1). Putnam's Twin Earth argument regarding necessary truths which are discovered by scientists, is based on a principle of structure identity (PSI).
(2). But the PSI can be shown to be defeasible-- viz., that the PSI is not itself a necessary truth.
(3). Hence, since PSI is defeasible, Putnam's Twin Earth experiment does not work and necessary truth claims cannot follow from this experiment.
Even if we grant the first premise, the second can be challenged by arguing
that Polakow's thought experiment is not enough to show that the PSI is
defeasible. Polakow claims that his thought experiment has shown that, under
special circumstances, H2O
is XYZ. But his claim seems to allow
for the possibility that the two substances are merely indistinguishable, not
the same substance. In order to show that the PSI is defeasible, Polakow must
argue that the two substances are really identical not merely indistinguishable. Recall that
Polakow claimed that Putnam must accept the PSI (Given any two chemical
structures A and B it is not possible that if they are indeed two structures
that they be identical). Polakow needs to produce an example to show that this
principle is false in certain circumstances, hence defeasible. But all that he
can show with his counterexample is that in certain circumstances the two
elements (XYZ and H2O)
are indistinguishable, just as, for example, a mixture of red, white and blue
light might be indistinguishable from white light in some circumstances, but
that does not mean that this mixture is identical to white light. So, Polakow
needs a stronger thought experiment to show that the PSI is defeasible, since
his experiment only shows that under special circumstances the elements are
merely indistinguishable. And the latter claim squares with the PSI, since we
do not have one element in the special circumstance but two, structurally
distinguishable ones.
Quine has leveled a famous claim against essentialism: what if there is a
character named Jones who counts as his avocations both cycling and
mathematics? Such a person would have rationality as an essential trait, qua
mathematician, and two-leggedness as an essential trait qua cyclist. But, Quine
says, is "this concrete individual necessarily rational and contingently
two-legged, or vice versa?" (1960 p. 199). It would seem that essentialism is
problematic because it would make each trait both essential and accidental,
hence leading to absurdity. For Quine, the essentialist is committed to
claiming that statements about some objects must be analytic in form. He says
the essentialist falsely maintains that,
an object, of itself and by whatever name or none, must be seen as having some of its traits necessarily and others contingently, despite the fact that the latter traits follow just as analytically from some ways of specifying the object as the former do from other ways of specifying it (1953 p. 155).
Therefore, the essentialist is committed to the paradoxical claim that certain predicates hold both necessarily and contingently of a thing.
There is a problem with Quine's claim above: namely, that he treats as the same
essential traits and analytic ones; he does not also realize
that cyclist and mathematician are not natural kinds, upon
which essences proper are based.
First of all, Quine is famous for having argued that so-called analytic truths
are no more privileged than other sorts of truths. But, in this passage he presumes
that, if essential claims are necessary then they would seem to be analytic.
Analytic claims are based on our linguistic conventions, so essentialist claims
are too. What he is trying to show in the quote above is that if we equate essential with analytic truths
truths that are specified by some sort of convention— we would end up with
absurd claims, that properties are both necessary and contingent. There is no
way of specifying the real essential truth, or analytic truth, of something,
hence there is only our own decision to accentuate some properties over others;
and this leads to conflicting claims about the necessity and contingency of
properties. As Robert Hollinger sums up:
For how are we to tell which of the properties `necessarily rational' or `necessarily two-legged' really expresses Jones' `real essence'? On the assumption that analyticity is the only criterion, there is no way of telling, and indeed no sense to the claim, that this quest is valid. At best, we would be expressing an unjustified inegalitarian attitude to some descriptions or classifications of Jones... (p. 329).
On the present view of essences, though, essential properties are not
analytically specified, but are discovered through investigation. Kripke's
claim is that the necessity of essential properties is an a posteriori
necessity. The analyticity requirement makes the necessity a priori, since
analyticity is a requirement about how something is known to be true. Yet if Putnam and Kripke are correct, we can
come to know essential properties from empirical investigation. If this is
true, then there is a non-conventional way of specifying what is essential to a
thing and what is not. If gold has a certain chemical structure so that it
dissolves in aqua regia and melts at the same temperature (under the same
atmospheric conditions), then it is not by convention, hence a matter of
analyticity, that we distinguish gold from other kinds of substance.
Lastly, it should be noted that Quine's cycling mathematician example seems more
plausible than it is. This is because he takes cycling and mathematicians to be
natural kinds. But this is patently false. Rationality,
animality, and man are,
plausibly, to be found in rerum natura,
hence are all natural kinds (Jones' specific difference, genus, and species,
respectively). Jones is a man—
this is one natural kind that he is a part of (Hollinger p. 338). His essential
traits are the ones he shares with the rest of his species. Cyclist and
mathematician are classes that he can belong to: he can be both. But Jones
cannot be a man both essentially and accidentally. So, Quine's example of a
cycling mathematician to refute essentialism was ill chosen.
Of course, Quine's argument against our ability to find ways of specifying
necessity is more resilient than I have suggested. For instance, Carter and
Bahde (pp. 305-06) argue that Quine's rejection of de re modal claims is
indicative of modal antirealism. The
modal antirealist believes that agents impose modal constraints upon the world;
and apart from our own minds there only exists a formless, modally innocent
`stuff' that awaits our conceptual cookie cutters. The general argument runs as
follows. As Quine famously stated, there are `no entities without identity'
(which is, for him, a `materially adequate' criterion of identity with no modal
force). And the identity conditions of the things in question are said to be,
in effect, modal constraints. But modal constraints for the modal anti-realist
are products of our conceptual activities. So, in the end, the identities of
things are dependent on our conceptual activities. This would seem to be
Quine's general argument regarding necessity; take for instance his claim:
[N]ecessarily exceeding 7 is no trait of the neutral thing itself, the number, which is the number of planets as well as 9. And so it is nonsense to say neutrally that there is something, x, that necessarily exceeds 7 (quoted in Carter and Bahde p. 307).
What is necessary, for Quine, is relative to a certain way of looking at the world, and not a part of the `neutral thing in itself'. If the world exists as it does, full of cats and dogs, trees and persons, then this can only be because our conceptual faculties have created and fit these identities over the modally innocent stuff.
Carter and Bahde argue that there is a problem with taking the world to be a
formless stuff that is modally innocent. If we can show that some identity
conditions are independent of our conceptual activity, then some kind of
independence for modal properties follows. They argue that if we take an
ordinary housecat, Ben and "the parcel of matter that is presently located
where Ben is located" (p. 312), which we will call `Jerry', we can see that
there would seem to be some object where Jerry is located. If Jerry is an
object, as it would seem to be "if we are to allow that Ben is a coincident
entity"(ibid), then Jerry would seem to have some modal properties —since these
are in effect identity conditions that are not fixed by our conceptual
activities. How can Jerry exist as a modally innocent stuff? If it is an
entity, then, contrary to the denial of modality in the first premiss above (No
entity without identity), it would have to have some persistence conditions,
hence modal properties. Moreover, these properties exist as part of Jerry's
separate, actual existence some independent identity conditions exists simply
because Jerry is an independently existing thing.
In this chapter I have attempted to defend the Kripke-Putnam theory of the
referential rigidity of natural kind terms. These terms refer to essences or
persons in a way that allows us to capture the `intuitive' belief that they are
non-variable; if they signify anything at all, then they signify a person or
kind of thing necessarily, so that, for instance, Nixon would still be Nixon
even if he had never entered politics. But had a Nixon doppelganger been born
in the Royal Family, this material difference would make him essentially
different from the real Nixon.
We will see the import of this view of essences next. As we noted in the first
chapter, claims such as the following are logically possible:
A pig flew over Dublin yesterday.
James Connolly once swam from Galway to New York.
Socrates was once a lion (Dunlop p.76).
Dunlop continues:
It might be said that incredible or not, such feats are logically possible. It is surely possible to conceive a world in which they occur. In that world entities superficially similar to pigs and men and lions,.... have a material constitution which does not impose the same limits on their possibilities as does the material constitution of pigs and men and lions in this world (p. 77).
As we will see, to say that such a world is logically possible is not to say that there is a world where our sort of pigs fly; but it is to say that there is a world very much like this one where something very much like our pigs fly. The burden of the next chapter will be to show that, while worlds where pig-like things fly are logically possible, worlds where pigs fly are not essentially possible. Indeed, there is no conceivable world where pigs can fly.
NOTES
[i] The phrase de re signifies that something is being predicated of a subject's res. So, in `Socrates is essentially a man' we are saying that the actual person, Socrates, is a man essentially. When we say that a certain proposition is necessarily true, we are making a de dicto claim; we are only talking about the dictum or proposition involved.
[ii] The problem here is that, as Russell saw, if (1) `The King of France is bald' is false, then, by the law of excluded middle which states that the contradictory of a proposition has the opposite truth value, (2) `The King of France is not bald' must be true. But this cannot be correct since there is no King of France. Rather than hold that we have a counter example to the law, Russell argued that we can distinguish a scope difference in the denial of `The King of France is bald'. We can, when negating, put the negation sign in front of the whole statement or within the statement, viz. In front of another quantifier. So, (2) can be shown as follows (where Fx= `is the King of France' and Gx = `x is bald'):
(2') (›x) ((Fx & ¬Gx) & (œy) (Fy e x=y)).
(2'') ¬(›x) (( Fx & Gx) & (œy) (Fy e x=y)).
[iii] By psychological state Putnam means any sort of mental state: concepts, or ideas or whatever you want to call representational content.
[iv] Perhaps there is not always an underlying physical structure as with substances; the essence of man could be rationality which is not exactly an underlying structure. In such a case an essence is just that which accounts for most of the differences (using tools, art, etc) between man and other things.