Thursday, September 4, 2008

Are Principles of Justice from Reflective Equilibrium Coherent?

A philosophical essay by Courtney Flatt

In John Rawls’ “A Theory of Justice,” Rawls develops a process known as reflective equilibrium (RE), which is supposed to lead to the proper principles of justice. Rawls begins by setting up a purely hypothetical situation, in which a group of individuals in the original position (those who are the beginners of a society, or the original contract-makers) are placed under a veil of ignorance that is meant to deprive the individuals of “…knowledge of those contingencies which sets men at odds and allows them to be guided by their prejudices.” (Rawls, p. 19) These include their social standing, economic status, physical characteristics, and any other differentiating aspects of the individuals. The veil allows for all other knowledge, including sociological, psychological, economical, and historical, and any other knowledge which universally applies to all people at all times. These specific restrictions are meant to make the contract “…generally acceptable [by all who will be affected by said contract, and] that no one should be advantaged or disadvantaged by natural fortune or social circumstances…” (Rawls, p. 18)

As Rawls says, “There is, however, another side to justifying a particular description of the original position. This is to see if the principles which would be chosen match our considered convictions of justice or extend them in an acceptable way.” From this comes RE. While under the veil of ignorance, the people in the original position would chose principles of justice. Then they would be brought out from under the veil of ignorance, and would be able to match those principles with their considered convictions (which are our beliefs that have been consciously decided upon and scrutinized more closely than some sort of steady feeling that has not been scrutinized). If it turns out that the principles of justice chosen under the veil of ignorance do not match their considered convictions, then they can either revise their principles or revise their considered convictions, or even a little of both. “We can either modify the account of the initial situation or we can revise our existing judgments, for even the judgments we take provisionally as fixed points are liable to revision.” (Rawls, p. 20) Through the repetition of this process, reflecting on chosen principles and considered convictions and revision of one or the other, an equilibrium is to be achieved by having the correct balance between the principles and their considered convictions.

RE can be divided (roughly) into two sorts: wide and narrow. Narrow RE (nRE) is where only the principles that already match one’s intuitions and personal beliefs are considered, whereas in wide RE (wRE), all possible conceptions of justice that one might agree with are presented as options. “For the notion varies depending upon whether one is to be presented with only those descriptions which more or less match one’s existing judgments except for minor discrepancies (nRE), or whether one is to be presented with all possible descriptions to which one might plausibly conform one’s judgments together with all relevant philosophical arguments for them (wRE).” (Rawls, p. 49) In the original position, the contract-makers undergo wRE, because the hypothetical situation is supposed to give those principles of justice that do not simply conform to our already accepted notions of judgment, but to those that are chosen without bias and with all possible options.

In an article entitled “Nature and Soundness of the Contract and Coherence Arguments,” David Lyons argues that the principles decided upon through a process of RE (which from now on will refer to wide RE unless otherwise specified) rely on a coherentist theory of justification (which states that there are no non-inferentially justified beliefs; all justified beliefs are inferentially justified from other inferentially justified beliefs. In this view, you end up with a “web “of beliefs, in which all beliefs are connected to and justified by other beliefs in the web.). This would be great, claims Lyons, if coherentism was the true account of justification, but it is not. All beliefs in the coherentist theory are based on and inferentially justified by intuitions, and this, claims Lyons, is circular reasoning, and if that was not bad enough, all beliefs gained through intuition are arbitrary. He says that the higher-level intuitions (our considered judgments, those that we have seriously thought about and screened for natural error, and things such as prejudice) are used to set up the constraints of the original position, and then the principles that are the product of the original position are justified by our lower-level intuitions (our un-scrutinized beliefs). For the original position, our higher-level intuitions would be used to guide us in what restriction we would place on the contract makers, as to what self knowledge they would have, including personal preference, physical characteristics, and economic/social standing. These are considered as coming from our higher-level intuitions, because in order to determine what things people are prejudiced about, and why those prejudices are bad, we must make some sort of moral judgment, and these are the types of judgments that are based on closely scrutinized beliefs. It has even been suggested that these higher-level beliefs were arrived at through a process of RE, which is where the charge of circularity comes in (using RE to decide how the original position should be set up in order to attempt a RE to attain principles of justice). This process, according to Lyons, is a circle of inferential justification, where the constraints of the original position are inferred from our higher-level intuitions, and then our principles are justified by our inference of our lower-level intuitions. Since Lyons disagrees that coherentism is the true account of justification, the only result that can come of it is that the process of RE does not justify its principles.

In a more colloquial, and perhaps explicit way, the main concern of Lyons and others with coherentist theories is expressed by the “garbage in, garbage out” problem. The idea behind this is that if, in a theory where beliefs are inferentially justified from other beliefs, some belief P that is used to justify other beliefs is arbitrary or untrue (in other words “garbage”), then all beliefs inferred and therefore justified by that belief P will be arbitrary or untrue (they will also turn out to be “garbage“). If there is garbage going into a theory of justification, and further beliefs are said to be justified and inferred by said garbage, then those further beliefs will also be garbage. One solution offered to this problem is found in a foundational theory of justification.

In a foundational theory of justification, there is a linear line of beliefs with a base belief(s) that is noninferentially justified. All further beliefs are based on the original one. Though the additional beliefs along the foundational chain can gain extra justification from coherence with other beliefs in the chain that are perhaps not the original noninferentially justified beliefs, they are all traceable back to and dependent for full justification on the noninferentially justified belief. This is said to solve the problem of “garbage” by not relying on intuitive beliefs (unless the “intuition” it is dependent on is the Kantian sort of space and time, but those are not the type of belief intuitions that are under consideration; those intuitions are epistemically necessary and are not “believed in” as some sort of moral precept, but apparently this is a digression, though I think these kind of intuitions could solve the entire dilemma and lend credence to certain forms of RE),which as Lyons argues, could be arbitrary. Noninferentially justified beliefs are said to be less likely arbitrary because they do not rely on inference from other beliefs.

One variant of foundationalism that gives an explanation as to why some noninferential beliefs are less likely to be arbitrary is called reliablism. In reliablism, beliefs are said to be justified because they are based on some sort of reliable process, one that has been performed previously and that has a tendency to lead to the correct results (or beliefs). Unfortunately, there are a few problems with foundationalism. One, called the epistemic ascent argument, may show that foundationalism is nothing more than coherentism in drag. This argument claims that, in such cases as reliablism, it is inferred that beliefs are likely true, so the foundational beliefs would actually be inferential, which is what foundationalism claims and wishes to avoid. The second problem is that foundationalism invites dogmatism. By this, I mean that since the base beliefs in foundationalism don’t require any inferential justification, and are sometimes claimed to be self-evident, and therefore exempt from the burden of proof, it is easy to accept certain beliefs as true without justification. The third problem, which goes along with the dogmatism charge, is that a noninferential and justified belief does not equal a noninferentially justified belief. This means that a belief could be noninferential, but have justification, say from other beliefs (as was said earlier, that beliefs in a foundational chain can gain some justification from other beliefs, but that they are only fully justified when they are derived from a noninferential belief). This is, I think, how dogmatism occurs. A belief is believed noninferentially without any justification, and is therefore unjustified, but a dogmatist may claim that since it is noninferential, that it is justified, maybe through something like self-evidence. With these comments about coherentism and foundationalism in mind, we can consider another argument concerning RE that supports the principles that come from it through a coherentist theory.

Norman Daniels disagrees with Lyons. His argument, in “Wide Reflective Equilibrium and Theory Acceptance in Ethics,” states that the inferential nature of theory acceptance for RE is similar to the process that we use for scientific theory justification, and if it is good enough for science, then it is good enough for ethics. In the article, Daniels says that “In science, we have evidence that we are not dealing with accidental generalizations if we can derive the purported laws from a body of interconnected theories, provided these theories each, in a diverse and interesting way, go beyond the “facts” that the principle generalizes.” In this statement, “accidental generalizations” are likened to generalizations in moral theory that are not based on scrutinized evidence, but on an accidental coincidence that leads to the appearance of a certain state of affairs that may not be the true case. Daniels suggests that there is a way that RE can avoid that sort of “accidental generalization.” If there is independent support for the back ground theories that the moral principles are derived from, then there is stronger evidence that the principles are not just accidental generalizations. This going-beyond of the generalized principles by seeking justification outside of the directly inferred beliefs is Daniels’ clever way of avoiding the garbage problem. Though he does not use any noninferential beliefs, he seeks to strengthen the principles and beliefs of RE by expanding the belief web. By adding more beliefs that correspond to and support every other belief, the beliefs become more evidence-based, and even though they still continue to be inferential, there is a larger group of beliefs to infer from, which, if it is granted that all beliefs in the system must be coherent with every other belief, gives less of a chance of keeping beliefs that are garbage. In other words, the more beliefs there are, the more coherent the web of beliefs must be. The hope is that this larger coherentism will weed-out any bad or garbage beliefs.

Daniels, by putting restrictions on the inferences and principles, such as “We should require that the background theories…be more than reformulations of the same set of considered moral judgments involved when the principles are matched to moral judgments.” avoids the simple inference that Lyons talks about. By developing a more complex system of inference, much like that which we use in scientific theorizing, Daniels gives more support to the idea that beliefs can be inferentially justified, and that they can lead to the proper principles of justice through RE. While there are problems with variations of RE as to whether or not the resulting principles are justified, Daniels give one of the more coherent accounts of how such justification can happen.


Lyons, David. “Nature and Soundness of the Contract and Coherence Arguments.”

Daniels, Norman. “Wide Reflective Equilibrium and Theory Acceptance in Ethics.”

Rawls, John. A Theory of Justice. Massachusetts: Harvard UP, 2005.


Everett Marx said...

I love the phrase,"foundationalism is nothing more than coherentism in drag."

My larger concern, though, is with the use of "arbitrary." I'm used to a distinction between necessary and contingent but not between necessary and arbitrary. Is the suggestion that anything which is *not* necessary is therefore arbitrary, and equally arbitrary as any other contingent possbility?

That would create problems for me. I don't think you meant that. But I'm not sure what you *do* mean by "arbitrary" here.

Candace said...

This is way over my head.