VII—The Supervenience of Mental Content

Abstract

It seems pretty obvious to many people that mental content cannot be a relational property; still more if the relational property is some variety of causation, however hedged. This is put sometimes by claiming that mental content is an 'intrinsic' property. But expressing it in this way does not give us a clue about the arguments that are supposed to back the contention, for relational properties could be 'intrinsic' in the sense of being essential-at least in some deflationary senses of 'essential': thus, it could follow from the concept of mental content, as manifested by some explicit definition, that mental content is relational. To express more properly the intuition, it was formerly customary to resort to epistemic considerations: we could not be as certain as we are regarding our own instantiating content properties if those were relational. Nowadays, the conceptual basis for claims like this one has be rendered so shaky that even the philosophers closest to the views that used to be sustained by those claims would rather rely on more direct metaphysical considerations. Like this: mental content cannot be relational, for two individuals who share content properties might still differ in relational properties. Of course, unless supported by further considerations, an assertion like this would just be begging the question against, say, the defender of that variety of functionalism which holds that its being in certain causal relations is analytic for a state to instantiate content properties. Nowadays very popular considerations to back the claim that two individuals could share content properties while still differing in relational properties ('Twin Earth' considerations) are based on a famous thought-experiment. I will present them in a more detailed way later. The general idea goes like this. (i) Content properties are supposed to be causally efficacious; they play an