Why might someone think that language is necessary for having propositional attitudes? A common reason for supposing that language is necessary for thought is that one is in the grip of a picture about the nature of thought—namely that thought is a type of language, or is linguistic or language-like. If propositions are linguistic entities, then creatures that lack the capacity for linguistic representation might well be unable to represent propositions and thus be unable to hold an attitude toward a proposition. However, since there are competing accounts of what propositions actually are, several of which see them as non-linguistic in nature, the intuitive language-like characteristics of propositions does not settle the question (Lewis 1979; Stalnaker 1984).
In several places Davidson gestures at related arguments for denying non-linguistic creatures propositional attitudes (1984, p. 156, 2001a, p. 98). These stem from an implicit commitment to propositional attitudes having certain characteristics that only languages possess. For instance, Davidson claims that propositional attitudes have definite content, and that only things expressed in language have definite content. Drawing on the discussion of Malcolm (1972) before him, he gives an example of a dog chasing a cat up a tree. Like Malcolm, he notes that we cannot attribute to the dog the thought that the cat ran up the maple, as opposed to that the cat ran up the tree. If there is no particular thought we can attribute to the dog, then the dog hasn’t had a thought with definite content, and so hasn’t had a propositional attitude. Davidson elsewhere claims that propositional attitudes are opaque[8], and that language accounts for their opacity (Davidson 2001a, p. 97). Although these claims can be combatted directly, I will not pursue those arguments here. Both the definite content claim and the opacity claim lose their teeth when it is recognized that they take the following form:
P1 Propositional attitudes have a property, p
C Therefore, language is necessary for propositional attitudes
This argument is fallacious—it would only be valid if nothing but language had property p. But no such argument is on offer. It is worth noting, moreover, that whether propositional attitudes have the property p in question is itself contentious—do all our beliefs have definite content? Finally, even if having property p were somehow constitutive of thought, and to have p thought had to be linguistic, this would still not entail that a creature with beliefs and desires must have language in Davidson’s sense. Fodor (see Fodor 1975), for instance, thinks that a creature must have a language of thought to have propositional attitudes, but he holds that it need not be able to speak or understand a public language to have a language of thought. Even if claims about definite content and opacity were true, that is, if Fodor is right, Davidson has erred in thinking that thought requires an external as opposed to an internal language. If animals have a language of thought, they are non-language-using believers.