Tim Crane has a great paper here explaining ‘propositionalism’. This the view (some call it “sententialism”) that all reports of intentional states can be analysed as propositional attitude reports. I alluded to this in my last post . If we can analyse ‘Tom is thinking about a unicorn’ as ‘Tom has a thought about a unicorn’ then we can set aside the problem of Tom’s relation to an exotic ‘intentional object’ and talk instead about his thoughts, and their purported relation to such objects.
How does that help? Modern logic already explains how the relational form of a proposition can be explained without invoking exotic objects. Consider
Jake believes there is a gold mine in Surrey
Jake wishes there were a gold mine in Surrey
We are not tempted to invoke such objects because the first sentence, for example, can be analysed as ‘Jake believes that for some x, x is a gold mine and x is in Surrey’. This does not entail the existence of any x. The fact that Jake believes p does not entail that p is true. Now imagine a language where the italicised words above are represented by a single word. For example, suppose that ‘to Ø’ is a verb meaning means ‘to believe there is’. Thus
Elizabeth Ø’s a man in the moon
John Ø’s a golden mountain
The Greeks Ø’d a winged horse
This verb takes a subject (a person) and an accusative or object – a thing whose existence is believed in. Grammatically it appears superficially to relate two objects. For example, the first sentence apparently relates Elizabeth to the man in the moon, a nonexistent object. But its semantics implies no such thing. “Elizabeth Ø’s a man in the moon” simply means that Elizabeth believes there is a man in the moon. Since her belief (as far as we know) is false, it follows that there is no man in the moon. Not even a non-existent man in the moon. We don’t need to invoke queer Intentional Objects to explain why each of the three sentences above is true, even though no man is in the moon, no mountain is golden, and no horse has wings.
Nor do we have to expend much thought to explain why the first inference below, but not the second, is invalid.
Jake Ø’s a gold mine in Surrey, therefore some gold mine is Surrey
Jake owns a gold mine in Surrey, therefore some gold mine is Surrey
The realist wants to explain the difference by invoking different sorts of relation. He may suppose that ‘owns’ is a genuine relation which requires the existence of its relata, whereas ‘Ø’s’ is a queer, intentional relation which, while still a relation, does not require the existence of all of its relata. Nominalists know better. For the inference
Jake believes there is a gold mine in Surrey, therefore some gold mine is Surrey
is clearly invalid. We cannot infer the existence of something from mere belief in its existence. This is the evidence, then, of a deep logical structure for verbs like ‘think’, ‘desire’, ‘know’. They are syntactically simple. But they are (probably) logically complex. This is merely a hypothesis. Possibly there may be some other explanation. But the alternatives (nonexistent objects) are both implausible and illogical.
There are difficulties with propositionalism, as Crane notes in the paper. I will discuss these later.