The great insight of Charles Sanders Peirce’s analysis of representation is his claim that aboutness can’t be explained solely in terms of relations between representing vehicles and represented objects (Hardwick 1977). Instead, vehicles are about their objects in virtue of having a certain kind of effect on a cognitive subject—specifically, vehicles either trigger thoughts about objects (in cases of public representation) or they engender behavioural dispositions towards them (in cases of mental representation). According to Peirce, it is this additional relatum—known as interpretation—that renders representation triadic.
But once interpretation is added into the representational mix, it has the potential to overwhelm any content-grounding relations that may obtain between vehicles and objects. This is the thread that Koch astutely pulls on in her commentary. To the extent that one can appeal to the manner in which a representing vehicle modifies a subject’s behavioural dispositions in order to capture the relational character of content, it seems as though one can also appeal to these dispositions to fix the content of this vehicle. In short, the triadic account would appear to make those theories of content determination that appeal to vehicle-object relations—such as resemblance—redundant (or, at least, “only indirectly relevant”, as Koch charitably puts it; this collection, p. 8).
Koch is not alone in drawing out this consequence from the triadic nature of representation. It is precisely this idea about the role of behavioural dispositions in content fixation that forms the foundation of the instrumentalist approach to mental representation that Daniel Dennett has defended over many years (1978; 1987). Dennett was one of the early proponents of triadicity, insofar as he argued that it was only in virtue of their roles in cognitive systems that representing vehicles can be interpreted as bearers of information:
There is a strong by tacit undercurrent of conviction […] to the effect that only by being rendered explicit […] can an item of information play a role. The idea, apparently, is that in order to have an effect, in order to throw its weight around, as it were, an item of information must weigh something, must have a physical embodiment […]. I suspect, on the contrary, that this is almost backwards. [Representing vehicles]… are by themselves quite inert as information bearers […]. They become information-bearers only when given roles in larger systems. (Dennett 1982, p. 217)
Dennett has also famously argued that the consequence of taking the triadic account seriously is the rejection of any story that takes mental content to be determined independently of a cognitive creature’s patterns of behaviour.
Dennett’s instrumentalist approach to mental representation, however, has another famous consequence. If the full burden of content determination falls on the shoulders of interpretation—if, that is, it is a cognitive system’s behavioural dispositions ultimately fix the content of its representing vehicles—then content is a product of cognition, not an ingredient, and hence cannot be casually implicated in the production of behaviour.
This last point, of course, represents Dennett’s own solution to the content causation problem: he abandons the thesis that mental phenomena are causally efficacious of behaviour in virtue of their representational contents (see O’Brien this collection, fn. 1). This is also what Koch is hinting at when she indicates that my proposal to invoke the triadic account of representation might be better interpreted as rejecting the content causation problem rather than solving it (Koch this collection, p. 8). That is, far from showing that rehabilitation of the resemblance theory of content determination is mandatory, her (implicit) objection is that my proposed solution to the content causation really shows that there is no such thing as content causation in the first place.