To: Pearl K.
From: Geoffrey Klempner
Subject: Substituting 'Cicero' for 'Tully' in statements of belief
Date: 27 March 2007 09:58

Dear Sachiko,

Thank you for your email of 16 March, with your University of London essay in response to the question, 'Given that Cicero is Tully, is it possible for Tom to believe that Cicero was a Roman Orator but not believe that Tully was a Roman Orator? Discuss.'

I came back from Prague at the end of last week but it's take a while for me to get back into the working mood. There was a piece on the CSR Conference on the Czech Radio International Service which includes a short except from my presentation:

http://www.radio.cz/en/article/89393

You have managed to get to grips with the basics of this problem, and present them with a reasonable degree of lucidity.

The only part where I put a question mark was where you said, 'However, it is possible to reduce the syntactically de re belief that 'Someone is such that Tom believes that he was a Roman Orator' into the semantically de dicto belief that 'Tom believes that someone was a Roman Orator', as both report essentially the very same belief and therefore statements such as the de re 6. above do not refute the Fregean claim that names have both sense and reference.'

On the face of it, 'Tom believes that someone was a Roman orator' merely states that Tom believes that there was at least one Roman orator, whereas, 'Someone is such that Tom believes that he was a Roman orator' implies, in addition, that there is a particular individual that Tom's belief is about. So I'm not quite sure what you were saying here.

Let me start by massaging your intuitions a little. Imagine the following conversation:

Bill: 'Did you know that Siss-roe was a Roman orator?'

Tom: 'Who was Siss-roe, I've never heard of him! Can you write the name down?'

Bill: 'There...'

Tom: 'Oh, you mean Siss-uh-roe...'

It seems plain daft to say (out loud) that Tom believes that Siss-uh-roe was a Roman orator but Tom does not believe that Siss-roe was a Roman orator.

Frege's solution is not meant to be applied to every possible case where there is potentially a break down in substitution in opaque contexts. That's the first thing. So now the question is, what is special about the cases where it does apply?

There seem to be two kinds of case, which are quite different in character. Frege's examples, of Hesperus and Phosphorus - or the mountain known from different sides as 'Afla' and 'Ateb' - are not at all common, in fact quite contrived. It is not often that we do find clearly distinct 'modes of presentation' where there is a clear physical or geographical difference between two different 'routes' to an object of reference.

Much more common is the kind of case where we use names like cards in a card index system. Tom has two mental cards marked 'Cicero' and 'Tully'. Information is collated on these two cards in a variety of ways. By some unfortunate accident, peculiar to the idiosyncratic way Tom's beliefs have developed, Tom has somehow failed to spot the duplication. By contrast with Frege's favoured examples, there is very little temptation in this case to posit a 'sense' for the names 'Cicero' and 'Tully'.

However, this would not bother Frege. His solution in this case would be to move from names to descriptions. The best representation of Tom's belief that Cicero was a Roman orator substitutes the description that Tom would most likely use for 'Cicero', whatever that may be. The content of Tom's belief about Cicero can thus be more idiosyncratic than, say, Tom's belief about Hesperus.

We've seen that it would be daft to say that Tom believes that Siss-uh-roe was a Roman orator but does not believe that Siss-roe was a Roman orator. So we are prepared to move a way from the criterion of what sentences Tom would assent to, when asked. The question is how much further we can, or should move.

The two cases above, the Frege case and the card index case suggest the principle that there has to be some substantial breakdown in the knowledge gathering process to justify the refusal to make the substitution. There are many other cases where we don't feel the slightest problem with reporting someone's belief about a person using a name which they themself would not recognize. In other words, in practice, much of our talk of belief is de re rather than de dicto. If one were looking for some sort of justification for this it would be that we are in the knowledge gathering game together, it is a co-operative enterprise.

However, this point focuses our attention on the difference between knowledge and belief. The case can be made (this is for epistemology) that belief is normally knowledge - unless something goes wrong. We are naturally constructed as knowledge gatherers. When the focus is specifically on an idiosyncratic belief, it is because our interest is in explaining the behaviour of an agent. For example, Alice is not afraid of Mack the Knife, the notorious serial murderer because she knows him as Mike the friendly neighbour who helps her carry out the garbage.

So, is it possible for Tom to believe that Cicero was a Roman Orator but not believe that Tully was a Roman Orator? My answer would be, maybe, if there is a sufficiently interesting/ relevant explanation of the breakdown. Otherwise, in normal speech we would not hesitate to make the substitution, in other words, to construe the belief as de re.

All the best,

Geoffrey