We see now that an attempt to solve the content causation problem must address the question of how the specific character of representational content can be analyzed in a way that invokes only the intrinsic properties of the brain (instead of being understood as “being a relational property and not an intrinsic property of the brain”). However, O’Brien advertises a theory of content determination that draws on second-order similarity between mental vehicles and the outside world as necessary for a solution to the content causation problem. In fact, he admits that “all theories of mental representation, in their efforts to explain the relational character of mental content, are forced to invoke world-mind relations of some kind”, where the latter term seemingly refers back to “relations that incorporate factors beyond the brain” (O’Brien this collection, p. 12). But how does this relate to the explicit goal of providing “an explanation of the relational character of mental content that invokes only the intrinsic properties of the brain”, which would only “prima facie [appear]” to be, but not—as the content causation problem is supposed to have a solution—actually be “impossible” (O’Brien this collection, p. 3)?
The answer to this might lie in a view which can be found in O’Brien & Opie:
Von Eckardt observes that the triadicity of representation in general, and mental representation in particular, can be analysed into two dyadic component relations: one between representing vehicle and represented object (which she calls the content grounding relation); the other between vehicle and interpretation […] This suggests that any theory of mental representation must be made up of (at least) two parts[7]: one that explains how the content of mental vehicles is grounded, and a second that explains how they are interpreted. (2004, p. 5)
When O’Brien writes that every theory of representational content, including his own, must make use of factors extrinsic to the brain, he most likely refers to the content grounding relation—in his case, second-order resemblance. Yet when he promises us “an explanation of the relational character of mental content that invokes only the intrinsic properties of the brain” (O’Brien this collection, p. 3), the scientific explanation mentioned most likely refers to the other part of the theory of mental representation: the internalist theory of interpretation. If O’Brien takes it that only this theory, which provides us with a reconstruction of the causal processes involved in mental representation, needs to be presented in terms of intrinsic properties of the brain, then he provides an account within the “narrow content program” (ibid, p. 3): this research program accepts the thesis that “[t]he representational contents of mental phenomena are not determined by the intrinsic properties of the brain” (ibid., p. 2) but— quite plausibly, I think—relaxes the metaphysical constraint made explicit in section 3 insofar as it only demands “an account of mental phenomena according to which (at least the causally relevant component of) their representational properties are determined by intrinsic properties of the brain” (ibid., p. 3, my emphasis).
If this is a correct reconstruction of O’Brien’s steps towards a solution to the content causation problem, then he has reached his goal if he:
has provided an account of the causally-relevant components of representation that makes use of only the intrinsic properties of the brain, and
can make sure that this account still deserves to be called an account of representation, i.e., captures the specific characteristics of representational content that we have so far called “relational”.
O’Brien claims that “[t]he insight offered by triadicity is that the relational character of mental content is to be discharged ultimately in terms of our behavioural dispositions towards features of the world” (this collection, p. 12). While it might not be clear at first sight why this provides a solution to the problem at hand, I am convinced that a view of dispositions as second-order properties—such as the property of having a property that becomes relevant for the causing of a certain manifestation once a certain stimulus is provided—helps us to see how O’Brien’s account provides a solution. I hope to have shown in section 2 how the adoption of the view that dispositions are second-order properties fits into his picture of representation. Thus I believe that it allows us to regard the first of the two requirements mentioned above as fulfilled: I see no reason why such a second-order property should not be understood as an intrinsic property of the brain. Furthermore, I hold that this view allows us to regard the second requirement as fulfilled, too: the dispositions in question seem to deserve the label “relational” insofar as, when combined with a certain stimulus, they are manifested in terms of overt, observable behaviour of a biological organism. When so manifested, they turn into concrete chains of events linked by causal relations. One can now argue that these potential relations are what let us intuitively characterize representation as relational. So understood, O’Brien’s explanation does justice to the project of providing an account that captures the specific character that makes representational content deserve the label “representational”, but without characterizing its causally efficacious components as being relational properties.