In our book, Ben Jarvis and I defend a Fregean theory of mental content. We hope that it does three things: it provides the best resolution to (the most interesting version of) Frege's puzzle; it has a plausible story to tell about the relationship between Fregean senses and the psychological states that constitute propositional attitudes; and it is able to underwrite the epistemology of the a priori. We came to our work on mental content via the epistemology, but we consider this latter project independently motivated and foundational. Our treatment of mental content comprises Part I of the book, and we hope that the book is as much a contribution to mental content as it is to epistemology.
Consider these two propositions:
- Some roses are red.
- Some roses have a colour.
Here's a very natural idea: it's part of the essence of these two propositions that (1) entail (2). There are lots of ways one might fill this out, but it's very natural to say that part of what makes proposition (1) the proposition that it is is that any time it is true, (2) is also true. We take this natural idea and carry it a step further. Not only do propositions have truth conditions necessarily and essentially; they also have rational acceptance conditions necessarily and essentially. Part of what makes (1) and (2) the propositions that they are is that they stand in a particular rational relationship to one another. In particular, (1) rationally entails (2), in addition to metaphysically entailing it.
We call these rational entailment relations Fregean senses. You can think of Fregean senses in our sense as a kind of truth conditions. If you're comfortable thinking this way, they're equivalent to sets of 'rationally possible worlds' (where there are some of the latter in which, e.g., Hesperus is not Phosphorus). Fregean senses encode what a content rationally commits one to.
Our unstructured Fregean senses constitute a departure from Fregean orthodoxy, which would have structured senses. This is motivated in significant part by the kinds of considerations I discussed in this post last week. We think there is an important theoretical role to be played by such unstructured entities, because the notion of rational commitment is fundamental to our story about mental content. (Of course, we also believe in more structured counterparts -- these, we call 'propositions'. Naturally, there are many ways to apply labels in this neighborhood; we try to justify our terminological choices, but the possibility for superficial disagreement here is significant.)
You need Fregean senses, we think, for basically the same reason Frege thought: to account for Frege cases. On our view, however, the most fundamental category of Frege cases isn't about the possibility of informativeness, or the explanation for certain kinds of behaviour. Frege's puzzle is ultimately a puzzle about rationality. If I believe that Hesperus is a star, and then I learn that Hesperus is a planet, I face rational pressure to revise my previous belief. This wouldn't be so if I learned instead that Phosphorus is a planet. None of the neo-Russelian views out there, we argue, can explain this fact. We explain it very straightforwardly: HESPERUS and PHOSPHORUS are different contents, which carry different rational relations.
(This is a view about the metaphysics of attitudes, not about the semantics of attitude reports. As we explain in the book, our view in consistent with a lot of views -- including neo-Russellian ones -- about the latter.)
Rational commitments, on our story, are primitive and fundamental. Chapter 5 of our book draws an analogy between our way of thinking about senses with Timothy Williamson's suggestion to put knowledge 'first'. We think it is a mistake to seek substantive explanations for why certain rational entailments obtain between certain contents. This move might motivate some to suspect us of shrugging off the most fundamental questions, but this isn't necessarily the case. True enough, calling senses fundamental is in some sense a way of moving the bump in the carpet somewhere else. But we have a lot to say about its new location: the psychological realisation of Fregean sense.
If you spot us the suggestion that there are some abstract entities called 'propositions' that have inherent and essential rational relations with one another, a major open question becomes: how is it that we humans manage to stand in any kind of significant relations to these obscure entities? This is among the most central questions in Part I of our book. A nice and convenient answer, were it true, would be the familiar conceptual role theorist's answer: contents can be characterised by particular inferential roles, and a subject thinks thoughts with those contents by virtue of dispositions to infer according to those special roles. (This should remind you of Christopher Peacocke.) Unfortunately, as people like Quine and Williamson have shown, this nice and convenient answer isn't true. We need a more complicated story.
Ben and I agree with Peacocke that there are certain privileged inferential roles that play a special, content-fixing role. The inference from "is red" to "is coloured" is special in a way that that from "is red" to "looks at least a bit like sriracha" is not. But we don't think that this special inference need be encoded at all directly in the dispositions of any subject who possesses the concept RED. Instead, we suggest that these special inferences have a privileged teleo-normative, rather than dispositional, status. Part of what it is to possess the concept RED is to be such that inference to COLOURED is proper or correct. Part of what makes a football player a goalie is that she is supposed to prevent the ball from going into the net; it is partly in virtue of her behaviour that she is subject to this norm. But it's not a requirement that she be very good at her job.
In a closely analogous way, we think that there are rules of thought. Part of what it is to think is to be subject to certain rational norms; for example, the norm that one should infer (2) from (1). Subjects constitute thinkers partly in virtue of their behaviour and dispositions, but in a way that doesn't guarantee a particularly high level of compliance. According to the story of the book, subscription to particular rules emerges in virtue of the best systematisation of the myriad first-order dispositions to apply concepts in various ways. I can't go into much more detail in this blog post, but a different kind of analogy might help get the approach into mind. Imagine a wooded area, with various significant locations along the perimeter. People need to get from place to place, via the woods, and at first, it's pretty arbitrary what route they take. They don't all just go in a straight line, because some parts of the woods are easier to walk through than others. Over time, paths emerge. Lots of factors influence with paths come to exist -- which destinations are most important, the natural lay of the land, which routes already exist, etc. But once there are paths, there are, in some sense, correct ways to get through the woods. This path is the way you're supposed to go. This, even though nobody ever laid down the law; the path emerged over time as the product of lots of other more arbitrary activity. There's lots more to say about how this could work -- and there are many respects in which the analogy is imperfect -- but I hope that this gives at least a rough idea of the teleo-normative inferential roles that we discuss in the book.
(It is worth noting that an implication of the approach is that we need not construe contents individualistically. We're entirely open to the idea that contents are public, and the best systematisation of first-order dispositions occurs at a broader social level. If this is right, our view implies that rationality, like meaning, ain't in the head. That's fine with us.)
I'll write one more post about the third hook into the book -- consideration of the role of intuitions in epistemology -- soon.