The Gettier Problem: A Study—Part 5




Definite Descriptions and the Gettier Example

Christoph Schmidt-Petri in his “Definite Descriptions and the Gettier Example”[1] provides a framework for the argument I outlined in an earlier post in terms of a distinction between the attributive use and the referential uses of definite descriptions drawing heavily from Donnellan’s analysis of definite descriptions. A fairly standard definition of a definite description is of a term that describes a particular thing and takes the form ‘the x such that φx’ where the definite description begins (in English) with the definite article, “the x,” and can be used in any sentence. Russell held that the truth value of the definite description in the sentence “the man who will get the job has ten coins in his pocket” does not change regardless of what stands for “the man who will get the job” as long as the attribution “has ten coins in his pocket” is true of him.[2] Using “Jones” or “Smith” in place of “the man who will get the job” does not change the truth value of the sentence as long as having ten coins in his pocket can be attributed to either man. What Kripke calls names are, on Russell’s account, disguised (or, perhaps better, abbreviated) definite descriptions. It is what is attributed that is important regardless of the term we use for the object that is the recipient of the attribution.

It is easy to see how this account supports Gettier’s examples. If it turns out to be Smith that actually gets the job and not Jones, the substitution of Smith for Jones or even “the man who has ten coins in his pocket” modifies the truth value in no way as long as the sentence expresses a true proposition by attributing something true of the subject regardless of what term is used for that subject. Thus, when Smith comes to believe “the man who will get the job has ten coins in his pocket,” the definite description, “the man who will get the job” can be replaced with a logical equivalent, “Smith” since he is the man who actually satisfies the predicate with no change in truth value.

Donnellan challenged Russell’s thesis and described two ways in which definite descriptions can be used. First, descriptions like P2 can be used attributively which is essentially Russell’s position. In attribution, we seek to assign an attribute to any object. The description in, “the businessman who started Microsoft is the richest man in the world” will yield a true proposition regardless of who that businessman is as long as that person satisfies the predicate of the sentence “is the richest man in the world.” It happens to be Bill Gates but could be anyone. Donnellan also describes a referential use of definite descriptions where the generalized term is intended to pick out a specific object and say something true about that object. In this usage, “the businessman” is intended to pick out Bill Gates specifically and describe him as the one who started Microsoft. On this usage, “the businessman who started Microsoft” refers to Bill Gates and is not satisfied by any other object.[3]

In the argument above, I outline three distinct beliefs Smith could have when attempting to understand Smith’s epistemic relation to P2 which are:

A. Jones will get the job and Jones has ten coins in his pocket

B. Smith will get the job and Smith has ten coins in his pocket

C. Some man will get the job and that man has ten coins in his pocket

Propositions A and B, I claim are specific beliefs about an individual and C is a general belief about any man whatever. Schmidt-Petri argues that Smith’s belief about P2 could be taken attributively but it could also be taken referentially but in either case, the reason Smith does not have knowledge is not due to a failure with JTB but due to Smith having a false or unjustified belief. Beliefs about referential descriptions should be taken as singular and the objects referred to in the sentence about the belief should be understood as a constituent. In attributive descriptions, the object being denoted does not contribute to the semantic content of the belief but merely to its quantificational structure. As I understand him, Schmidt-Petri is arguing that in existential propositions like ones used in attributive descriptions, the object referred to in the proposition is not related to the meaning; what the object actually refers to. Rather it is only related to the logical relation the object has to the rest of the sentence.

If taken referentially then, Smith’s beliefs would either be A or B. Smith would either believe that a particular person, Jones or Smith himself will get the job. If Smith believes A, he would not have a true belief and thus would not satisfy the requirements of JTB. If he believes B, he would have a true belief but would not be justified in believing it because he possesses no evidence that he Smith has ten coins in his pocket or that he would be getting the job and thus does not satisfy the requirements of JTB: he is not justified. Schmidt-Petri says it is certainly possible that Smith could believe C.[4] For example, he might have had evidence that every man that applied for the job was required to carry around ten coins in his pocket or that before the job is awarded, the committee deciding who gets the job gives the candidate ten coins and asks him to put them in his pocket. But Gettier in his case give us no reason to believe that Smith has any such evidence and thus most likely does not believe C. Even if he does in fact believe C, in this case he has not satisfied JTB because he lacks evidence for its truth and that is the reason Smith does not know C and, salva veritate, P2.

I would go a step further and rule out C on the grounds that Gettier tells us that Smith derived P2 from P1. We’re told that Smith believes P2 based on a deduction from P1. It seems to me that in order to derive P2 from P1, the semantic content of P1—specifically the referent of the name “Jones”–must be retained in P2—specifically the referent of the description “the man who has ten coins in his pocket”–for the entailment to be satisfied in the technical sense of that term. This would mean that Smith’s belief in P2 would contain the semantic content of P1 in some sense and gives us further reason to reject the idea that Smith believes C as an attributive belief.

Regardless, Schmidt-Petri has shown that whether we take Smith’s belief as referential (he believes A or B) or attributive (he believes C) in all three cases he does not know P2 because he has not satisfied the requirements of JTB not because JTB is inadequate in some way.


[1] Definite Descriptions and the Gettier Example, Christoph Schmidt-Petri, LSE , Centre for the Philosophy of the Natural and Social Science (DP 63/02). See also his “Is Gettier’s First Example Flawed?” in Knowledge and Belief. Papers of the 26th Int. Wittgenstein Symposium (Contributions of the Austrian Ludwig Wittgenstein Society, Vol XI), ed. W. Löffler and P. Weingartner, Kirchberg: The Austrian Ludwig Wittgenstein Society, 2003: pp. 317-319.

[2] Schmidt-Petri describes Russell’s view of definite descriptions in terms of the existential statement ∃x (Fx & ∀(Fy →y=x)&Gx)): There is at least one x such that, F is true of x and for every y, if F is true of y then y is identical to x, and G is true of x. There is at least one thing such that that thing is a man and for any other thing, if that thing is a man then that thing is identical to first thing and the first thing has 10 coins in his pocket. The variable ‘y’ establishes the identity of any individual for whom the attribution is true.

[3] Saul Kripke uses the term “referent of the description” to refer to the referential usage of definite descriptions though in a footnote, he introduces the term “semantic referent” as the referent of a name or description. So if one man says to another, “Jones is over there raking leaves” but it’s actually Smith, the semantic referent is Smith, the man raking leaves even though the name “Jones” was used for him. The referent of the name “Jones” is Smith if the speaker uses Jones to “label” Smith in referring to the object. In the case of a definite description, e.g. “the man” (Kripke uses the term “description” here but I take him to mean a definite description), whatever satisfies the predicate of the sentence “is raking leaves” is the semantic referent. Note that Kripke appears to say nothing here about how reference is transferred by logical deduction. He appears to be only concerned with direct reference. If I see two men under a tree and one is raking leaves and the other is napping, when I refer to “the man who is raking leaves,” the definite description picks out the man who is raking leaves not the napping man. See Saul A. Kripke, Naming and Necessity, (Cambridge, Massachusetts: Harvard University Press, 1980), 25,26, footnote 3

[4] Schmidt-Petri writes C as the existential statement: ∃x (Fx & ∀y (Fy → y=x) & Gx) and refers to this statement as the proper way to understand C as an attributive belief.



Blackberry Bushes

Blackberry blossoms, pink-white under bees— rabbits look up longingly. Blackberries ripe and sweet; my brother and I pick them— reaching...