Fregean Descriptivism
In class we discussed Fregeanism and some objections to Fregeanism from Priest and Kripke. In this post I will present a general version of Fregeanism as well as some further problems for the view. Feel free to defend Fregeanism from one of these objections for your comment paper for this week.
The following is a schema for Description Theories of the meaning of proper names (SDT):
Where P is a property, C is a condition, S is an agent, N is a proper name, O is an object, and L is S’s language:
i. P satisfies condition C.
ii. S believes that there is exactly one thing that is P.
iii. S authoritatively associates P with N.
iv. N refers to O in L iff O is the one and only thing that is P.
v. There is a predicate, F, and a language L’, such that:
a. L’ is an extension of L, and
b. F expresses P in L’, and
c. N in L is synonymous with “the F” in L’.
Clause (i) is designed to place substantive conditions on the relevant property.
Clause (i) is schematic because different theories provide different constraints, although all theories must provide some such constraint.
Minimal constraint on P: the constraints must allow descriptivists to block the relevant instances of Frege’s Puzzle (and the Hooded Man paradox) and must not be circular (i.e., 'the referent of 'N'').
Fregean constraint on P: P must be a purely general property. (Frege held that only purely general properties are constituents of Thoughts; objects themselves are never constituents of Thoughts.)
(As a historical note, Bertrand Russell also at one point held a view that amounted to a version of (SDT). Russell offered a different constraint on P from Frege, however: Russellian constraint on P: P may be a relational property, restricted to items with which one is directly acquainted; i.e., one’s self and one’s current experiences, and properties exemplified in one’s current experiences. Note that the arguments from Kripke equally threaten Russell's version of (SDT). Russell held this view in no small part because of considerations like the Hooded Man.)
Clause (iii) claims that there is some property that S authoritatively associates with N. Most people associate some property with (e.g.) ‘Twain’, i.e., that he is a human, a writer, an American, etc. Roughly, S authoritatively associates P with N if S holds that in worlds in which nothing is P, the referent of N does not exist. (Note: it need not be the case that everyone that is competent with N associates the same property with N, although all of those speakers might share a single public language. On such a view, ‘L’ picks out a speaker’s idiolect rather than any public language.)
Clause (v) is intended to relax the constraint that the definite description that is synonymous with a given proper name in L is expressible in L.
Finally, it’s worth mentioning that some more sophisticated descriptivists deny that there is a single property that speakers authoritatively associate with a proper name. Rather, the occurrences of ‘P’ in (i-v) should be replaced with ‘P1-Pn’, to reflect their conviction that the referent of N in L is the object that satisfies a weighted majority of P1-Pn. (John Searle held (holds?) this sort of view, for instance.)
Descriptivism and Frege’s Puzzle (also applicable to the Hooded Man, as we saw):
Suppose that S authoritatively associates being-the-author-of-HF with ‘Twain’ and being-the-most-famous-person-from-Hannibal-Missouri with ‘Sam Clemens’. Consider (3) and (4):
(3) Kant believes that Twain is the author of HF.
(4) Kant believes that Clemens is the author of HF.
The descriptivist can readily explain the difference in truth value between (3) and (4); they express different propositions. According to descriptivism, (3) and (4) are synonymous with (5) and (6), in S’s language:
(5) Kant believes that the author of HF is the author of HF.
(6) Kant believes that the most famous person from Hannibal, MO is the author of HF.
Since (5) and (6) express different propositions, they can differ in truth value. Since the complement clauses of (5) and (6) express different propositions, they can differ in informativeness, a prioricity, analyticity, and rational speakers could reasonably think that they differ in truth value.
Donnellan, Kaplan, and Kripke have offered modal, epistemic, and semantic objections to descriptivism.
Suppose S authoritatively associates being-the-author-of-HF with ‘Twain’. Consider (11) and (12):
(11) If Twain exists, then Twain is the author of HF.
(12) If the author of HF exists, then the author of HF is the author of HF.
Modal Objection (discussed in class)
13. a. If descriptivism is correct, then (11) is synonymous with (expresses the same proposition as) (12).
b. If (11) is synonymous with (12), then (11) is necessarily true iff (12) is.
c. It’s not the case that (11) is necessarily true iff (12) is.
d. Therefore, descriptivism is not correct.
This objection denies (v).
Epistemic Objection (discussed in class)
14. a. If descriptivism is correct, then (11) is synonymous (expresses the same proposition as) (12).
b. If (11) is synonymous with (12), then (11) is a priori iff (12) is.
c. It’s not the case that (11) is a priori iff (12) is.
d. Therefore, descriptivism is not correct.
This objection denies (v).
Semantic Objections: Ignorance and Error
Ignorance
Suppose S only associates being-a-physicist with ‘Richard Feynman’. Then:
15. It’s not the case that S believes there is only one physicist, so (ii) is false.
16. It’s not the case that S associates a definite description with ‘RF’, so (iii) is false.
17. ‘RF’ refers to RF in S’s language, so (iv) is false.
Error
Suppose S authoritatively associates being-the-first-European-to-land-in-America with ‘Christopher Columbus’.
18. a If descriptivism is correct, then in S’s language, ‘Christopher Columbus’ refers to some Norse sailor.
b. It’s not the case that in S’s language, ‘Christopher Columbus’ refers to some Norse sailor.
c. Therefore, descriptivism is not correct.
This objection denies (iv).
In addition to modal, epistemic, and semantic objections, Salmon has identified problems Fregeans have with apparently de re beliefs. (This is not unrelated to Priest's objection that we considered.) So suppose the following is the case:
19. Locke is thinking of Leibniz that he is tall.
(19) is formalized in the variable-quantifier idiom as (20):
20. There is an x such that x = Leibniz and Locke believes that x is tall.
(20) is true only if (21) is true under an assignment of Leibniz to 'x':
21. Locke believes that x is tall.
But (21), in turn, is true only if Locke bears the belief relation to the semantic content of (22), under an assignment of Leibniz to 'x':
22. x is tall.
The sole semantic content of a variable under an assignment, however, is the object that satisfies the formula under that assignment (the variable's referent). So the semantic content of (22) under the relevant assignment is an object-involving proposition, one that we might represent as an ordered n-tuple of Leibniz himself and the property of being tall. If this is correct, then no purely general Thought involving only properties (concepts/senses) is such that believing it is sufficient to make Locke's de re belief true. So if any de re belief is true, Fregeanism is false.
The next problem is also due to Salmon. Consider (23) as uttered by you:
23. Aristotle believed that Plato was wise.
If Fregeanism is correct, (23) expresses a Thought that we may represent as follows:
24. [s Aristotle s, s believed s, ss Plato ss, ss was wise ss]
(23), in turn, is true iff Aristotle bore the belief relation to [s Plato s, s was wise s], on Frege's view.
Now consider which properties you authoritatively associate with 'Plato'. They almost surely differ from whatever properties Aristotle authoritatively associated with the Greek translation of 'Plato', if any. So in your mouth, 'Plato' expresses [s Plato s]. But in Aristotle's mouth, the Greek translation of 'Plato' expressed [s* Plato s*], where this indicates whatever sense it was that Aristotle managed to express.
As we noted, [s Plato s] is distinct from [s* Plato s*]. But since your utterance of 'Plato' expresses the former rather than the latter, your utterance of (23) is true iff Aristotle bore the belief relation to [s Plato s, s was wise s]. But surely he did not. If anything, he bore the belief relation to a Thought involving [s* Plato s*]. So your attribution of a belief to Aristotle that Plato was wise is false if Fregeanism is correct. But it's not false. So Fregeanism is incorrect.
Finally, Kripke pointed out in "A Puzzle About Belief" that things only get worse when we consider attitude ascriptions with quantified subjects. Consider (25):
25. Most late 19th century Germans thought that Hegel was smart.
Not only does your sense of 'Hegel' differ from theirs, but theirs differed from each other. If the Germans' thoughts about Hegel were adequately diverse, there may be no descriptive property common to all of their "Hegel Thoughts" at all--much less any common to yours. So even if the previous problem were somehow avoided so that (23) came out true, the truth of (25) on Frege's view requires attributing highly diverse beliefs to the 19th century Germans. But (25) does not attribute highly diverse beliefs to them; rather, it says that most of them had some belief in common. Fregeanism gets the wrong result, even assuming the previous difficulty is met. So the view is incorrect.
Oh yeah, and Augustine, Anselm, Aquinas, Descartes, Spinoza, Berkeley, Hume, Nietzsche, Wittgenstein, Quine. (That is not a separate objection to Frege.)
The following is a schema for Description Theories of the meaning of proper names (SDT):
Where P is a property, C is a condition, S is an agent, N is a proper name, O is an object, and L is S’s language:
i. P satisfies condition C.
ii. S believes that there is exactly one thing that is P.
iii. S authoritatively associates P with N.
iv. N refers to O in L iff O is the one and only thing that is P.
v. There is a predicate, F, and a language L’, such that:
a. L’ is an extension of L, and
b. F expresses P in L’, and
c. N in L is synonymous with “the F” in L’.
Clause (i) is designed to place substantive conditions on the relevant property.
Clause (i) is schematic because different theories provide different constraints, although all theories must provide some such constraint.
Minimal constraint on P: the constraints must allow descriptivists to block the relevant instances of Frege’s Puzzle (and the Hooded Man paradox) and must not be circular (i.e., 'the referent of 'N'').
Fregean constraint on P: P must be a purely general property. (Frege held that only purely general properties are constituents of Thoughts; objects themselves are never constituents of Thoughts.)
(As a historical note, Bertrand Russell also at one point held a view that amounted to a version of (SDT). Russell offered a different constraint on P from Frege, however: Russellian constraint on P: P may be a relational property, restricted to items with which one is directly acquainted; i.e., one’s self and one’s current experiences, and properties exemplified in one’s current experiences. Note that the arguments from Kripke equally threaten Russell's version of (SDT). Russell held this view in no small part because of considerations like the Hooded Man.)
Clause (iii) claims that there is some property that S authoritatively associates with N. Most people associate some property with (e.g.) ‘Twain’, i.e., that he is a human, a writer, an American, etc. Roughly, S authoritatively associates P with N if S holds that in worlds in which nothing is P, the referent of N does not exist. (Note: it need not be the case that everyone that is competent with N associates the same property with N, although all of those speakers might share a single public language. On such a view, ‘L’ picks out a speaker’s idiolect rather than any public language.)
Clause (v) is intended to relax the constraint that the definite description that is synonymous with a given proper name in L is expressible in L.
Finally, it’s worth mentioning that some more sophisticated descriptivists deny that there is a single property that speakers authoritatively associate with a proper name. Rather, the occurrences of ‘P’ in (i-v) should be replaced with ‘P1-Pn’, to reflect their conviction that the referent of N in L is the object that satisfies a weighted majority of P1-Pn. (John Searle held (holds?) this sort of view, for instance.)
Descriptivism and Frege’s Puzzle (also applicable to the Hooded Man, as we saw):
Suppose that S authoritatively associates being-the-author-of-HF with ‘Twain’ and being-the-most-famous-person-from-Hannibal-Missouri with ‘Sam Clemens’. Consider (3) and (4):
(3) Kant believes that Twain is the author of HF.
(4) Kant believes that Clemens is the author of HF.
The descriptivist can readily explain the difference in truth value between (3) and (4); they express different propositions. According to descriptivism, (3) and (4) are synonymous with (5) and (6), in S’s language:
(5) Kant believes that the author of HF is the author of HF.
(6) Kant believes that the most famous person from Hannibal, MO is the author of HF.
Since (5) and (6) express different propositions, they can differ in truth value. Since the complement clauses of (5) and (6) express different propositions, they can differ in informativeness, a prioricity, analyticity, and rational speakers could reasonably think that they differ in truth value.
Donnellan, Kaplan, and Kripke have offered modal, epistemic, and semantic objections to descriptivism.
Suppose S authoritatively associates being-the-author-of-HF with ‘Twain’. Consider (11) and (12):
(11) If Twain exists, then Twain is the author of HF.
(12) If the author of HF exists, then the author of HF is the author of HF.
Modal Objection (discussed in class)
13. a. If descriptivism is correct, then (11) is synonymous with (expresses the same proposition as) (12).
b. If (11) is synonymous with (12), then (11) is necessarily true iff (12) is.
c. It’s not the case that (11) is necessarily true iff (12) is.
d. Therefore, descriptivism is not correct.
This objection denies (v).
Epistemic Objection (discussed in class)
14. a. If descriptivism is correct, then (11) is synonymous (expresses the same proposition as) (12).
b. If (11) is synonymous with (12), then (11) is a priori iff (12) is.
c. It’s not the case that (11) is a priori iff (12) is.
d. Therefore, descriptivism is not correct.
This objection denies (v).
Semantic Objections: Ignorance and Error
Ignorance
Suppose S only associates being-a-physicist with ‘Richard Feynman’. Then:
15. It’s not the case that S believes there is only one physicist, so (ii) is false.
16. It’s not the case that S associates a definite description with ‘RF’, so (iii) is false.
17. ‘RF’ refers to RF in S’s language, so (iv) is false.
Error
Suppose S authoritatively associates being-the-first-European-to-land-in-America with ‘Christopher Columbus’.
18. a If descriptivism is correct, then in S’s language, ‘Christopher Columbus’ refers to some Norse sailor.
b. It’s not the case that in S’s language, ‘Christopher Columbus’ refers to some Norse sailor.
c. Therefore, descriptivism is not correct.
This objection denies (iv).
In addition to modal, epistemic, and semantic objections, Salmon has identified problems Fregeans have with apparently de re beliefs. (This is not unrelated to Priest's objection that we considered.) So suppose the following is the case:
19. Locke is thinking of Leibniz that he is tall.
(19) is formalized in the variable-quantifier idiom as (20):
20. There is an x such that x = Leibniz and Locke believes that x is tall.
(20) is true only if (21) is true under an assignment of Leibniz to 'x':
21. Locke believes that x is tall.
But (21), in turn, is true only if Locke bears the belief relation to the semantic content of (22), under an assignment of Leibniz to 'x':
22. x is tall.
The sole semantic content of a variable under an assignment, however, is the object that satisfies the formula under that assignment (the variable's referent). So the semantic content of (22) under the relevant assignment is an object-involving proposition, one that we might represent as an ordered n-tuple of Leibniz himself and the property of being tall. If this is correct, then no purely general Thought involving only properties (concepts/senses) is such that believing it is sufficient to make Locke's de re belief true. So if any de re belief is true, Fregeanism is false.
The next problem is also due to Salmon. Consider (23) as uttered by you:
23. Aristotle believed that Plato was wise.
If Fregeanism is correct, (23) expresses a Thought that we may represent as follows:
24. [s Aristotle s, s believed s, ss Plato ss, ss was wise ss]
(23), in turn, is true iff Aristotle bore the belief relation to [s Plato s, s was wise s], on Frege's view.
Now consider which properties you authoritatively associate with 'Plato'. They almost surely differ from whatever properties Aristotle authoritatively associated with the Greek translation of 'Plato', if any. So in your mouth, 'Plato' expresses [s Plato s]. But in Aristotle's mouth, the Greek translation of 'Plato' expressed [s* Plato s*], where this indicates whatever sense it was that Aristotle managed to express.
As we noted, [s Plato s] is distinct from [s* Plato s*]. But since your utterance of 'Plato' expresses the former rather than the latter, your utterance of (23) is true iff Aristotle bore the belief relation to [s Plato s, s was wise s]. But surely he did not. If anything, he bore the belief relation to a Thought involving [s* Plato s*]. So your attribution of a belief to Aristotle that Plato was wise is false if Fregeanism is correct. But it's not false. So Fregeanism is incorrect.
Finally, Kripke pointed out in "A Puzzle About Belief" that things only get worse when we consider attitude ascriptions with quantified subjects. Consider (25):
25. Most late 19th century Germans thought that Hegel was smart.
Not only does your sense of 'Hegel' differ from theirs, but theirs differed from each other. If the Germans' thoughts about Hegel were adequately diverse, there may be no descriptive property common to all of their "Hegel Thoughts" at all--much less any common to yours. So even if the previous problem were somehow avoided so that (23) came out true, the truth of (25) on Frege's view requires attributing highly diverse beliefs to the 19th century Germans. But (25) does not attribute highly diverse beliefs to them; rather, it says that most of them had some belief in common. Fregeanism gets the wrong result, even assuming the previous difficulty is met. So the view is incorrect.
Oh yeah, and Augustine, Anselm, Aquinas, Descartes, Spinoza, Berkeley, Hume, Nietzsche, Wittgenstein, Quine. (That is not a separate objection to Frege.)
posted by Chris Tillman at 8:24 PM
0 Comments:
Post a Comment
<< Home