The bulk of philosophical writing on representation in general and mental representation in particular assumes, either explicitly or implicitly, that representation is a dyadic relation between something that does the representing and something that is represented. The task for a theory of representation, from this perspective, is to explain the necessary and sufficient conditions under which this dyadic relation obtains (see e.g., Stich 1992). But such a dyadic conception provides very little guidance about the relational character of representational content. All we have to work with is a mysterious action-at-a-distance phenomenon, whereby one part of the world, in virtue of the obtaining of a certain relation, is about another part.
To fill this gap, philosophers have almost invariably modelled their understanding of content on the semantic properties of the elements that compose our natural languages. Given the towering influence of Tarskian truth-conditional semantics in this field, it is inevitable that the relational character of representational content is usually characterised in terms of reference (Kriegel 2008, p. 305). But such an approach, while perhaps appropriate for linguaform representation, sits awkwardly with all manner of the non-linguistic forms of representation with which we are familiar (Haugeland 1991; Fodor 2007; Cummins & Roth 2012). Moreover, it is not obvious we are more enlightened by replacing talk of aboutness with that of reference.
In this context, it is worth observing that over the years a minority of philosophers has expressed dissatisfaction with the dyadic conception of representation. The most salient complaint is that such an approach fails to take into consideration the role that “users” of representation play. The general thought here is that some parts of the world don’t represent other parts solely in virtue of some relationship between them; that the former represent the latter only when they are employed by some system to perform this representational function. According to Dennett, for example, physical entities “are by themselves quite inert as information bearers. […] They become information-bearers only when given roles in larger systems” (1982, p. 217). Likewise, Millikan has long observed that a biological approach to representation forces one to consider not just the “production” of representations, but also their “consumption” (1984). And, in a similar vein, Bechtel argues that that since whether something acts as a representation is ultimately determined by its function for some user, it follows that there are “three interrelated components in a representation story: what is represented, the representation, and the user of the representation” (1998, p. 299).
This triadic conception of representation is not new, of course, since it forms the basis of Charles Sanders Peirce’s theory of semiotics, which was developed in the latter part of the 19th century (Hardwick 1977). Indeed, Peirce’s (sometimes obscure) writings embody one of the most comprehensive analyses of representation in all of philosophical literature. Peirce approached this issue principally by investigating those public forms of representation with which we are all familiar—words, sentences, paintings, photographs, sculptures, maps, and so forth—but he also sought to apply his triadic analysis to the special case of mental representation. This suggests that Peirce’s writings might be an appropriate point of departure for exploring what the triadic conception entails about the relational character of representational content.
This strategy is very effectively adopted by von Eckardt when, following Peirce’s lead, she analyses representation as a triadic relation involving a “representing vehicle, a represented object, and an interpretation” (von Eckardt 1993, pp. 145-149).[4] As with dyadic stories, the representing vehicle is the physical object (e.g., a spoken or written word, painting, map, sculpture, etc.) that is about something, and the represented object is the object, property, event, relation, or state of affairs that the vehicle is about. It is the addition of the interpretative relatum that sets the triadic account apart:
A sign [i.e., a representing vehicle] […] is something which stands to somebody for something in some respect or capacity. It addresses somebody, that is, creates in the mind of that person an equivalent sign, or perhaps a more developed sign. That sign which is created I call the interpretant of the first sign. (von Eckardt 1993, p. 145)
Interpretation is thus understood as a cognitive effect in the subject for whom the vehicle operates as a representation. But as von Eckardt observes, not any kind of effect will do. This cognitive effect, presumably implicating the production of mental representing vehicles, must bring the subject into some appropriate relationship to the original vehicle’s represented object (von Eckardt 1993, p. 157). Given this constraint, it is natural to interpret this third relatum in terms of the subject’s thinking about the object in question. So (non-mental) representation, on the triadic story, is a functional kind: it is a process whereby a representing vehicle triggers a thought (or thoughts) in a subject about a represented object.
There are a couple of significant consequences of the triadicity of representation. The first is that, contrary to a dyadic story, representing vehicles aren’t about anything independent of interpretations. Words, sentences, paintings, photographs, sculptures, maps, and so forth, considered in isolation from the cognitive impact they have on us, don’t represent. This, of course, does some violence to the way that we talk about public representing vehicles—but it is far from catastrophic. The relevant revision is to think of these physical entities as possessing the capacity to trigger the necessary cognitive effects in us. The second (and, for our purposes, more important) consequence is that, unlike dyadic accounts in which content is unpacked solely in terms of relations between vehicles and represented objects, the triadic story entails that content is also conditioned by the interpretative relatum. This imposes an additional explanatory requirement on theories of content determination. It is not enough to merely explain how relations between vehicles and objects make it the case that the former are about the latter. These theories must also explain how it is in virtue of these relations that representing vehicles are capable of triggering thoughts in subjects about represented objects.
Once it has been suitably modified for the special, and presumably foundational, case of mental representation, the additional explanatory requirement that triadicity imposes on theories of content determination can form the basis of a richer account of the relational character of mental content. Such modification is necessary, of course, because treating the interpretation of mental vehicles solely in terms of a subject thinking about a represented object violates the naturalism constraint. This is the requirement that we explain mental representation without recourse to the antecendently representational (see e.g., Cummins 1989, pp. 127–129; Cummins 1996, pp. 3–4; Dretske 1981, p. xi; Fodor 1987, pp. 97–98; Millikan 1984, p. 87; von Eckardt 1993, pp. 234–239).
The relevant modification is fairly obvious, however, and represents a well-trodden path in philosophy. From the perspective of Peirce’s triadic analysis, the role of interpretation is to forge a psychologically efficacious connection between the user of a representing vehicle and the vehicle’s object. With public forms of representation it is perfectly acceptable to unpack this in terms of the (non-mental) vehicle activating thoughts directed at the object. But if we allow this story to run a little further it will point us in the right direction for the interpretation of mental vehicles too. Thoughts directed at objects modify our behavioural dispositions towards these same objects. This is why public forms of representation are so useful—they enable us to regulate our behaviour towards selective aspects of the world. But this story can be transported into the brain in order to account for the interpretation of mental representing vehicles. Instead of external vehicles triggering thoughts, and these in turn modifying behavioural dispositions, we simply suppose that mental vehicles have the same cognitive and ultimately behavioural effects. This acts to block the threatened regress since, presumably, it is possible to unpack behavioural dispositions without invoking further mental representation.
We are now in a position to deliver on one of the aims enumerated in the introductory section: that of fashioning a more illuminating account of the relational character of mental content. We saw earlier that Kriegel describes this character in terms of the “cognitive contact” between mental phenomena and the worldly aspects they represent, but admits that this notion isn’t particularly transparent. Happily, the triadic analysis of mental representation affords a means of explicating what this cognitive contact consists in. Rather than simply employing the vague language of aboutness, the triadic analysis encourages us to understand the relational character of mental content in terms of the capacity of mental phenomena to regulate the behaviour of subjects towards specific aspects of the world. Cognitive contact is thus a relatively straightforward causal capacity. It is the capacity of cognitive creatures, bestowed by their internal states, to respond selectively to elements of the environment in which they are embedded.
This is where things currently stand. A solution to the content causation problem requires something that prima facie appears impossible—namely, an explanation of the relational character of mental content that invokes only the intrinsic properties of the brain. But the paradoxical appearance of content causation, I have suggested, might be a legacy of the dyadic conception of representation that has conditioned philosophical intuitions about content determination, but which provides little guidance about the relational character of mental content. The triadic analysis of representation, I have argued, generates a more enlightening account of this relational character—one pitched in terms of the causal capacities of cognitive creatures to regulate their behaviour towards specific aspects of their environments. From the perspective of this analysis, therefore, a solution to the content causation problem requires a theory of content determination to explain how relations between mental vehicles and their represented objects can endow subjects with the capacity to respond selectively to those very features of the world.
Philosophers seeking to fashion theories of mental content determination over the centuries have famously focused on just two kinds of relations between mental vehicles and their represented objects: “causal” relations and “resemblance” relations (Fodor 1984, pp. 232–233). In the following section I shall engage in an all-too-brief appraisal of the prospects of these two world-mind relations to deliver a solution to the content causation problem.