To reiterate, the problem associated with the triadic analysis of representation is that once behavioural dispositions are invoked in order to explain the relational character of mental content, they threaten to overwhelm any other story about mental content determination. But if mental content is determined by such behavioural dispositions, it can’t play a robust causal role in their production. In short, the triadic account seems to suggest that there is no content causation problem because there is no content causation.
In this context, however, it is pertinent to note that, despite his insistence that representation is triadic, Peirce expends a good of effort investigating the relations between representing vehicles and represented objects. His analysis of public forms of representation famously yields three different kinds of vehicle–object relations—convention, causation, and resemblance—associated with symbols, indexes, and icons, respectively (see Hardwick 1977 and Von Eckardt 1993, Ch. 4). If content determination is ultimately just a matter of interpretation, why would Peirce have been so bothered about these vehicle–object relations?
The answer, of course, is that Peirce was concerned not just with the fact that public representing vehicles effect interpretations in cognitive subjects, but with how they do so. The point here is that interpretation isn’t magic—it requires explanation. Consider, for example, Leonardo da Vinci’s Mona Lisa. According to the triadic story, the painting that hangs in the Louvre is not about anything on its own. Its standing as a representing vehicle hinges on its capacity to trigger interpretations in cognitive subjects. When we look at this painting it causes us to think about a dark-haired woman with a famously enigmatic smile. But what is it about this painting that endows it with this capacity? Part of the explanation here invokes our recognition of the resemblances between the painting and a woman with a certain kind of physical appearance. The painting wouldn’t have the same impact on us if these resemblances didn’t obtain. So a complete account of the painting’s aboutness must go beyond the fact that it triggers certain thoughts in us to include an explanation of how it does so. And it is here that vehicle–object relations such as resemblance are compulsory.
The general lesson to take away from this (far too brief) analysis is that the interpretation of public forms of representation cannot be disconnected from the cognitive subject’s (conscious or unconscious) recognition of what are generally known as content grounding relations between vehicles and represented objects. And what goes for the interpretation of public representing vehicles also goes for the interpretation of mental vehicles. On the triadic story being entertained here, mental vehicles, just like the Mona Lisa, aren’t about anything considered in isolation. Their aboutness is a consequence of the multifarious behavioural dispositions they create in us towards selective features of the world—dispositions to physically interact with these features, for example, or to make utterances about them. Since this form of interpretation likewise isn’t magic, a complete account of mental representation must explain how mental vehicles establish these behavioural capacities. And just as with the case of public representing vehicles, it is impossible to do this without recourse to content-grounding relations (something that is demonstrated by even exceedingly simple representation-using devices such as the humble thermostat—see O’Brien this collection, pp. 7–9).
Precisely because content-grounding relations must be invoked to explain how the former endow cognitive systems with behavioural dispositions towards the latter, content causation is back in business. But what kind of vehicle–object relations can turn this trick? This, of course, was one of the central questions that animated much of my discussion in the target paper (O’Brien this collection). Of the three grounding relations that Peirce found to be implicated in public forms of representation—convention, causation, and resemblance—the first is widely assumed to be unavailable for mental representation since it violates the naturalism constraint.[1] Despite is popularity in contemporary philosophy, the second, I argued at some length, is actually powerless to explain how mental vehicles create the requisite behavioural dispositions (this collection, pp. 6–7). This just leaves us with resemblance. Fortunately, this third vehicle–object relation is up to the task, or at least so my argument went, since the structural properties of mental vehicles that ground second-order resemblance relations can be exploited to shape the behavioural dispositions of a cognitive system towards worldly objects (this collection, p. 8). This is where the rubber of resemblance meets the road of content causation. And it is why a resemblance theory of content determination is mandatory if we are to explain why mind matters.