There seems to be an immunity relative to the first-person pronoun, which at least guarantees that you cannot have a belief like “I believe that I am in pain” and accidentally take someone else to have that belief. It probably also guarantees that in this case you cannot be wrong about who is in pain. We think that there are a few good theoretical candidates for explaining this kind of immunity. These candidates are:
Irrelevance of misidentification
Immunity to misguided reference
Reference magnetism
Since reference magnetism[10] is a highly controversial, metaphysical notion and it would take too much time to elaborate this view correctly (which would certainly include a refreshment of Lewis’ philosophy of reference), we will focus on the first two for the sake of this commentary.
1. Irrelevance of misidentification:
If you take judgments about yourself to be a) always starting with de re beliefs and b) single-predicative in form, it seems impossible to construe misidentification as being relevant to the truth-value of a sentence or proposition. This point has been made by Barz (2010). Barz takes the current discussion to assume that there are two fundamentally different kinds of errors that can occur: an error through misidentification and an error through mispredication. It should be clear what an error through mispredication is supposed to be: an error through mispredication occurrs when a person's judgment is wrong and is only wrong because the predicate she thinks applies to a particular object in fact does not apply to that object. Barz' definition of an error through misidentification (in general) is the following:
General error through misidentification (EM-G): A person S (i) believes (de re) of a certain thing that it is F, (ii) believes that thing to be identical with a, and (iii) thus judges that a is F. But (iv) a is not identical with the thing S believes to be F.
According to Barz this kind of error cannot happen at all, so the proponents of the IEM-theory are right—but in fact IEM is not an exclusive property of judgments concerning the first-person or involving the first-person pronoun, and is instead a property of any judgment. His argumentation can be summarised in one sentence: since there are examples of judgments involving misidentification that are nevertheless true, and since there cannot be judgments involving mispredication that are true, there are no errors through misidentification. A judgment is right or wrong solely depending on whether the predicate applies to the object.
Imagine the following situation that is usually used to distinguish between notional and referential use of singular terms: Peter is a detective, investigating the case of Smith's murder. Participating in the judicial proceedings, a man, accused of having murdered Smith, behaves so strangely that Peter, the detective, judges: Smith's murderer is a maniac. He is using the term “Smith's murderer” to refer to the person that is accused of having murdered Smith, and according to most theories of reference he does in fact refer to that person with that term. But what if that person is not the one who murdered Smith, but is nevertheless still a maniac? Thus a misidentification has occurred, but no error. On the other hand, if the person were Smith's murderer but not a maniac (maybe his weird behaviour was the result of pharmaceutical treatment)—Peter's judgment would be wrong.
The same goes for the traditional wrestler example. Imagine that wrestler A and wrestler B are in a close wrestling fight and wrestler A does not misidentify her arm with the arm of wrestler B but still, for some strange reason—maybe there are blood smears caused by a bleeding bird that flew over the two wrestlers—comes to judge “My arm is bleeding” (although wrestler B's arm is actually bleeding). She would be wrong, but her error would not be one of misidentification but of mispredication. Thus, as Barz believes, there are no errors through misidentification, because the only thing that necessarily suffices for the falsity of a judgment is mispredication.
As one can guess, Barz' theory does not completely fit with our theory of judgments. While we take judgments to be processes of inference, thus involving several propositions, Barz seems to take judgments to be relations to single, structured propositions. We can agree with Barz if he can explain how the identification component in the judgment—which would, in our terms, be one of the premises used during the inference—is in fact a kind of predication.
2. Immunity to misguided reference:
Howell (2007) wants to distinguish between two kinds of immunity: immunity to error through misidentification and immunity to misguided reference:
IEM is often confused with what I call Immunity to Misguided Reference (IMR). A judgment that x is F has IMR if it is impossible for someone to make that judgment while being mistaken about the reference of x. All I-judgments have IMR, while not all I-judgments are IEM. (Howell 2007, p. 584)
To say that there is something like immunity to misguided reference (IMR) does not mean that one can never be wrong about the reference of any term one uses. It just means that whenever you want to refer to yourself using the term “I” you cannot fail to do so.
We think that a majority of the proponents of IEM are in fact proponents of IMR. And because IEM is thought to be an immunity relative to the first-person pronoun (what we have termed IEM-FP), it makes sense to say that this immunity is in fact an immunity of referring acts in general and not of judgments exclusively. Talking about IMR can be helpful in two ways: first, it can be helpful in stressing the fact that IEM is not a theory about the self or about subjectivity but simply a theory about linguistic rules and reference. Thus IEM-FP is a trivial property that can be explained by the semantic rules of usage of the word “I”.
Second, it can be helpful for explaining our intuitions in complicated cases of self-reference and by determining the objects of beliefs. Think of the two wrestlers again. When one of the wrestlers states “I am bleeding” or “My arm is bleeding”, she is wrong, but it seems as if she is not necessarily wrong because of a misidentification. Let's have a look:
(1) Wrestler A correctly describes her belief, intending to refer to herself using the first-person pronoun.
(2) One cannot fail to refer to oneself when using the first-person pronoun. (IMR-rule)
(3) Wrestler A has a belief about herself (granted by accepting 1 and 2).
So far the argument is trivial—stating that Wrestler A has a belief about herself just means that she has any kind of belief. It does not show that Wrestler A has a de re belief about herself. This comes from the second part of the argument:
(4) A de re belief is a belief that holds if the believer is in a non-conceptual, contextual relation to the object the belief is about.[11]
(5) One is always in a non-conceptual, contextual relation to oneself.[12]
(C) Wrestler A has a de re belief about herself (granted by accepting 3, 4 and 5).
Opponents of the IEM-theory would have to state that wrestler A has no de re belief about herself, because the object her belief is really about is not herself, but wrestler B, misidentified with herself (thus creating a de dicto belief about herself and a de re belief about wrestler B). But by accepting IMR and certain accounts of de re attitudes we can see that wrestler A's attitude is a possible candidate for a de re belief about herself. Thus the only reason why she would be wrong is—as we have seen above—mispredication.