— I'm in the midst of writing a fairly difficult paper on ideal and non-ideal theory in Rawls's moral & political theory, my ideas are rather inchoate, and I want to think “out loud” in hope of clarifying my understanding and maybe getting some useful pointers by those who know this stuff better than me. If you're not one of those people, be warned. I give thus one a black diamond.
Ideal theory produces the principles of justice that the “artificial agents” occupying the “original position” would choose acting as trustees for the”citizens of a well-ordered society.” Now, there is quite a cast of characters in Rawlsian constructivism. The citizens of a well-ordered society are not you and me. They too, like the ciphers in the original position, are artificial agents, or “agents of construction,” which is to say, strictly delineated characters of imagination. The citizens of a well ordered society (let me just call them Citizens) are characterized by a conception of the “moral person,” which is different from a theory of human nature. Moral persons (and thus Citizens) are taken to be possessed of a “sense of justice” that disposes them to comply with the principles of justice, whatever those might be, and to have a conception of their own good. And they are conceived as “free and equal” (about which I won't say anything here).
Now, the structure of justification for Rawlsian theory is quite complicated, and I don't believe I've nailed it down to my satisfaction. For instance, the agents in the OP are subject to “reasonable” constraints on the choice of principles. These constraints are justified by reference to the imputed properties of Citizens. But why characterize the citizens like this. Because so characterized, the agents in the original position, subject to reasonable constraints, acting as representatives of the Citizens, choose principles of justice that produce a well-ordered society for the Citizens. And, so, why care about this? So far, we haven't said anything about the actual world. This is a toy world, a figment of the imagination. Why do these principles have a claim on us? Well, they do in case they are in “reflective equilibrium” with our “considered judgments.” Now these are our actual considered judgments–the judgments of you and me. OK, and why does this have any justificatory force?
Well, we are alleged to have a sense of justice, too. Now insofar as our sense of justice plays a justificatory role in Rawls's theory, it is that a sense of justice is a part of our best theory of human nature. You and I aren't part of a model, so we aren't subject to the stipulations of the model conception of the moral person. Indeed, Rawls's model conception of the moral person is an acceptable part of the larger modeling device only if it agrees with our real-world considered judgments about our moral personhood. And it can't be just that our actual conception of ourselves as persons is such that we conceive of ourselves as having a sense of justice. We must actually have one. So, again, why the judgments that express our putative sense of justice supposed to be authoritative?
Rawls suggests that, as a first approximation, the function of a theory of justice is to describe our native sense of justice. And he compares the sense of justice to the Chomskyan capacity of syntactic competence. According to Chomsky, we have a native capacity to judge the grammaticality of utterances in the natural languages in which we have been brought up. The aim of a theory of syntax is to systematize native speakers intuitive judgments of grammaticality through a (hopefully minimal) set of principles that would predict these judgments. Ideally, this systematization mirrors the set of principles that is psychologically realized in language users and generates their competence.
OK, sorry. That's tough going. Now, some philosophers, namely Susan Dwyer, have run with the analogy to Chomskyan linguistic competence, further developing Rawls's intriguing idea of a sense of justice as a part of a broader moral competence characteristic of human nature. Now Susan wrote a paper on this, which I read a few years back, but no longer have access to, so I'm not addressing it. [Whoops! Just found it! But I haven't re-read it yet.] There's a damn good chance she addresses/answers/refutes everything I say. But for now, I'mimaginingg it so I can explore the analogy to linguistic competence in connection to the function our sense of justice is supposed to play in the justification of principles of justice derived from the Rawlsian construction device.
This takes me back to the distinction betweenn ideal and non-ideal theory. In ideal theory, we stipulate “perfect compliance” with principles in order to visualize what a fictive, ideal well-ordered society looks like. Perfect compliance is ensured within the model by the sense of justice of Citizens, which disposes them to adhere to the principles, and common knowledge among Citizens that all possess such a sense of justice.
Now, the fact that we do not inhabit an ideal well-ordered society implies one or more of a few things. Either our sense of justice is not quite that of the Citizens, i.e., is non-ideal, or our sense of justice is not fully realized, or it is fully realized but we make constant performance errors, or we lack common knowledge of our sense of justice.
On either of the first two options, the analogy with linguistic competence fails. Take the first. It's puzzling how ideal linguistic competence could differ from our actual linguistic competence. What would ideal linguistic competence be, other than actual competence. Competence with a more computationally efficient UG? Maybe competence in a Begriffschrift, a logically perfect, non-ambiguous language? But that's more than anything about semantics, while linguistic competence is about grammar. Take the second. Every cognitively normal person above a certain age has full linguistic competence. We can all generate and comprehend grammatical sentences of our native languages without fail, barring performance errors (which we make when we are slip up when we are tired, drunk, get confused, or whatever.) No special environmental conditions are necessary to develop linguistic competence, and Rawls stipulates that the every person past a certain age under normal social circumstances develops a sense of justice.
How about the third option: systematic performance errors. Well, this is weird idea. How could you tell what is and isn't a performance error if performance errors are systematic, much less develop a theory that there is an underlying “moral grammar” at all.
So then, the fourth: no common knowledge. This analogy straightforwardly fails. Common knowledge of linguistic competence is what makes communication possible.
Thus, the analogy, if we take it fairly literally, is not so good. If it was REALLY good, we wouldn't NEED a theory of justice, because we would already be living it, which we manifestly aren't. Let us suppose that there is universal agreement on certain very general moral principles, and this agreement is an expression of our native moral competence. Still, the persistent fact of moral disagreements so deep that people will kill each other over them, and the pervasiveness of gross injustice, suggests that the natural expression of our moral competence leaves us far far short of well-ordered societies.
In any case, I think the first hypothesis is most likely to be true. We have some sort of moral competence, and some sort of sense of justice is part of it, but it is not ideal in the sense that it disposes us to perfectly comply with the principles of justice. OK! But then we are forced to worry how we can justify the elements of the construction procedure that produce Rawls'sprincipless. If the theory of justice is supposed to characterize OUR sense of justice, then why suppose that our considered judgments would endorse the principles when they are the product of a construction procedure in which the stimpulation of a sense of justice unlike our own plays a prominent role?
Rawls says in several places that coming up a theory of justice is not an epistemic problem, because there is no objective moral order out there to discover. It may be true that there is no independent moral order, but it there's still an epistemic problem in determining what our sense of justice actually amounts to, and what conception of the person, and so on, it would endorse. If the strict analogy to linguistic competence is tight, then the theorist can take her own intuitions as evidence of the structure of our underlying moral competence. But the linguistic analogy is not tight, if it applies at all. So the theorist cannot assume that the intuitions she shares with her colleagues and students in the seminar room are representative. We'll have to look out the window. But this makeseverythingg REAL hard! We can't build a construction device without model conceptions. And we can't justify using this model conception rather than that unless we can test it against real people's sense of justice. And then the back and forth between judgments of particular cases and judgments about the acceptability of general principles derived from the construction device is bound to take a damn long time, and a lot of NSF grants.
Now, the later “political” Rawls modified his model conception of the well-ordered society to include reasonable but ineradicable pluralism of “comprehensive conceptions.” A comprehensive conception is like, say, Catholicism, Islam, or Kantian secular liberalism, or any such big theory that tells you all about your place in the universe, your nature as the kind of being you are, and so forth. Rawls thinks he needs to do this in order to ensure a kind of stability for a well ordered society. Unless the principles are those that anyone with a reasonable comprehensive conception could endorse, you won't get compliance with the principles, and the institutions composing the basic structure of society will be unstable. So the principles can't be rooted in any one comprehensive conception, but must be found in a some “political conception” consistent with the various reasonable big conceptions. Now, this is INSIDE the model conception of the well-ordered society. Why is Rawls modifying the model conception this way?
Before, the stipulation that the sense of justice of Citizens ensured perfect compliance was more or less enough to get the required stability. I think that Rawls was seeing that acomprehensivee conception is to a sense of justice something like what a natural language is to linguistic competence. And he's trying to create Esperanto. Our sense of justice is expressed THROUGH our comprehensive conceptions, and this causes our patterns of moral activity to take various not always fully consistent forms. When I'm talking to a Catholic conservative or a socialist, I feel quite like we're speaking different languages. Rawls sees this as an intrinsic aspect of the human condition under fairly liberal conditions. Compliance and stability based on a single comprehensive conception can only be maintained by the forced imposition of that conception by the coercive power of the state, which is flatly incompatible with liberal aspirations. So this feature of the real world, reasonable pluralism, has to be moved inside the model conception in order to jibe adequately and gain the assent of real folks whose sense of justice is expressed through a comprehensive conception.
So Rawls's hope is that the various reasonable comprehensive conceptions overlap to enough to sustain agreement. This still assumes that there is a universal sense of justice that is expressed neutrally enough in our considered judgments to endorse model conceptions that produce these “political” principles of justice. Maybe. But one wonders to what extent Rawls's model conception of the “moral person,” say, can be justified by a neutrally expressed moral competence, as opposed to a comprehensive conception-laden competence (based in Kant, say).
Furthermore, the fact that late Rawls moves the pluralism inside the model conception of the well-ordered society shows that he takes problems of compliance and stability seriously. Rawls notes that the theory of human nature constrains the conception of the person used inside the construction device. I imagine this works in rather the way that permanent features of social life, such as pluralism, constrain the conception of the well-ordered society. Are there also aspects of the theory of human nature that need to be made internal to Rawls's conception of the moral person in order to take compliance and stability problems fully seriously? Given the difference between the sense of justice of fictive moral persons and the sense of justice of real people, won't we need to move more features of real moral competence inside the conception of the moral person in order to produce principles of justice that would be endorsed by real people? And if we do this, what's the princinpled basis for stopping at some point before ideal and non-ideal theory collapse into one?
OK. That was rambling. And a horrifying, cryptic piece of writing if you don't already know what I'm trying to talk about. Sorry. But I find it useful to brain dump into the blog. So now tell me where to look in the vast unexplored (by me) territory of Rawls secondary literature in order to correct my mistakes and sort all this. Or just give it to me straight out. Thanks!