Opacity Revisited
by Dennis J. Darland
April 23, 2009
Revised April 25, 2009
Revised Again May 3, 2009
Last revised 03.05.2009 09.06 time
Copyright © 2009 Dennis J. Darland


I am going to try to restate my views on opacity which I have stated before, except without symbolic logic, that is using only English. First, what does opacity amount to? For Quine it means a position in a sentence lacks purely referential position. Purely referential position means that the position is subject to the substitutivity of identity, a term which which Quine attributes to Frege. (Quine, Word and Object, p. 142)

In, say, the sentence "Cicero denounced Cataline", 'Cicero' and 'Cataline' have purely referential position. Thus since Cicero = Tully, it follows, from that sentence, that "Tully denounced Cataline".

An example of a position which lacks purely referential position, (or is opaque) is "Tom believes that Cicero denounced Cataline". When Tom does not know Cicero = Tully, that sentence can be true without "Tom believes that Tully denounced Cataline" being true.(This example comes from Word and Object, p. 145)

First, let us acknowledge that from the fact that the words 'Cicero', 'Tully' and 'Cataline' exist, does not guarantee that there are objects corresponding to them. Tom believing that Tully denounced Cataline, does not guarantee that Tully exists. (forget for now the possibility of replacing names with definite descriptions, which I am aware of.) My analysis of this sentence is that Tom is related in a 'belief_that' relation to the words 'Cicero', 'denounced' and 'Cataline'. It is a further fact that for Tom, 'Cicero' is a name for Cicero, and 'Cataline' is a name for Cataline, etc.

Thus my analysis of "Tom believes that Tully denounced Cataline" is
Tom believes_that 'Cicero' 'denounced' 'Cataline' (The belief_that is in words)
and 'Cicero' is Tom's name for Cicero, and so forth.

Now, although Cicero = Tully, it is not true that 'Cicero' = 'Tully', so it does not follow: Tom believes_that 'Tully' 'denounced' 'Cataline'. (and this is not a violation of substitutivity of identicals - the names are not identical.)
The reason for the complex existential quantifiers (which I used before) was that no particular name or symbol must be used to have a belief. Thus let x be a variable. Then the analysis becomes:
There is an x such that Tom believes that x 'denounced' 'Cataline'
and x is a name of Cicero for Tom. (Note: 'Cicero' is no longer present - 'denounced' and 'Cataline' could be handled similarly.)
(Also if Cicero did not exist then 'Cicero' being a name of Cicero for Tom would be false. In such a case, Cicero might, be denoted by a definite description.)
Thus, using existential quantification we can define a belief relation between Tom, Cicero, etc.
We need not know what the symbol is, only that it occurs in the two relations.

Questions and Answers


Question:
If I understand you correctly, you claim that the positions in; sentences which appear not to be purely referential really are purely referential after all, once we take into account what the users take then words to mean.  Is that correct? I think it's awkward to cash out "believes that" sentences in terms of a relation between a person and words.  It seems possible for a dog to believe about a person dressed in black that it smells the same as a person dressed in white, or even believe that it is the same person again.  All without words.
Answer:
I was trying to take a simple case.  I would maintain that the symbols (which in my simple case I called names need not be words - using words just being a simple case to verbally explain.  Also there would probably be in the case of a person an experience corresponding to the word (in cases involving words) - which could vary from person to person (and which in some cases we might call the meaning of the word for that person). Likewise in the case of a dog, the dog would probably have some sort of experience which I would consider a sort of symbol. But perhaps nothing conscious need happen - just a kind of disposition to behaviour. These experiences themselves alone are not sufficient to constitute meaning. There also needs to be a practice associated with them - connecting them - to the object (if any) to which they name.    However it seems that apparent opacity (which Quine thinks is real opacity) is only possible with words and only occurs because of multiple people using the same words in differing ways - mainly due to incomplete knowledge. We know Cicero = Tully, but Tom does not. Perhaps it could also occur with only one person (with a language) with incomplete knowledge. He might come to realize, e.g. that two creeks he had given different names were, in fact, the same creek. 
Question:
You write: "Now, although Cicero = Tully, it is not true that 'Cicero' = 'Tully', so it does not follow: "Tom believes_that 'Tully' 'denounced' 'Cataline'. (and this is not a violation of substitutivity of identicals - the names are not identical.) Quine would agree with you that it does not follow.  Apparently what you really mean is that within a suitable referential view, it is no surprise that the sentence does not follow.
Answer:
I would maintain that, with the correct analysis of belief, there are no opaque contexts at all - this being a simple case to explain. Of course I am only considering words which stand for objects here. There are other sorts of words, but perhaps most other sorts of words do not matter here.  We don't apply identity to them anyway -substitutivity of identity only applying to names of objects. 
Question:
It seems to me that opacity in Quine's sense is possible even for creatures without language.    A dog can believe that the human who smells like *this* is his owner, but not believe that the person who smells like *that* is his owner, when the only difference is that on the second occasion the owner has eaten a lot of garlic. How would you handle that case?
Answer:
To try to cast it into my form:

Dog believes that 'person with smell A' is 'owner'
Dog believes that 'person with smell B' is 'not owner'

but owner = not owner (in this instance)  (making 'not owner' be 'some person other than owner')
But, I would say, the dog has different symbols 'owner' and 'not owner' - though they, in this case, stand for the same person. (though not really words) just as Tom does in 'Cicero' and 'Tully'.

Back to Top