Meta-ethics: Railton’s Moral Reductionism (part 1)

by Luke Muehlhauser on April 20, 2011 in Ethics

For the purposes of later work on meta-ethics (a long series explaining my meta-ethical theory, pluralistic moral reductionism), it will be helpful for me to summarize Peter Railton’s moral reductionism for my audience. To do so, I’ll be summarizing some of chapter 9 in Miller’s Introduction to Contemporary Metaethics.

I previously explained the differences between cognitivism and non-cognitivism, naturalism and non-naturalism, and naturalistic reductionism vs. naturalistic non-reductionism in my post on Cornell realism. Now I discuss one kind of naturalistic moral reductionism, the kind developed by Peter Railton in the 1980s and 90s.

 

Railton’s approach to moral theory

Railton is a methodological naturalist in that he claims to do philosophy with methods that are continuous with the sciences. He is also a metaphysical naturalist in that he thinks only natural things exist – things made of quarks and so on.

His strategy is “to postulate a realm of facts in virtue of the contribution they would make to the a posteriori explanation of features of our experience.” Specifically, he plans to offer reforming definitions for moral terms. For example, one might define ‘morally right’ as standing in for some natural property N (say, ‘contributes to the well-being of conscious creatures’). Railton doesn’t mean to claim that when normal English speakers say ‘morally right’ they mean to call up property N. Rather, he seeks to find reductive definitions for moral terms that fit the world and explain certain features of our experience.

And because he offers a reductionistic view of morality, Railton does not need to argue (with the Cornell realists) that moral properties pull their own weight in explanations. Instead, the natural properties to which moral properties reduce already pull their own weight in scientific explanations.

When offering reductionistic accounts, one might offer a vindicative reduction or an eliminative reduction. Consider the case of ‘polywater’, which in the 60s was thought to be a special form of water. It turned out that polywater was just normal water that had been contaminated by poorly cleaned lab equipment. Here, the reduction of polywater to “water containing impurities from poorly cleaned lab equipment” was an eliminative reduction; it called for an abandonment of the concept of polywater.

Other times, a reduction leaves enough of our original concept in tact that we find it tolerable to keep the term as it was. The reduction of water to H2O is one example. We once thought that water was one of the fundamental elements of the world, but it turned out that water is itself made of more fundamental elements. Still, the reduction of water to H2O “reinforces, rather than impugning, our sense that there really is water.” This reduction merely showed us more clearly what water really is in the natural world. If we had discovered that water was merely air with a few impurities in it (say, fire), we may have found that reduction as being eliminative, and abandoned our use of the concept of ‘water.’

Railton’s project is to see whether his reduction of moral properties to natural properties can be vindicative rather than eliminative. His criteria for success are (1) that the natural properties to which moral terms are reduced can play a successful role in explaining things (an explanatory role), and (2) that these natural properties can engage people motivationally “in the ways characteristic of moral properties” (a normative role).

 

Non-moral value

Railton begins with a reductive account of non-moral value: that something can be “desirable for someone, or good for him.”

Hobbes can be thought of as offering a reforming definition for non-moral goodness such that “good for X” means “desired by X.” Sex is (non-morally) good for me if I desire sex. Etc.

But Railton resists such a crude reduction of non-moral goodness:

My [actual desires] frequently reflect ignorance, confusion, or lack of consideration, as hindsight attests. The fact that I am now constituted that I desire something which, had I better knowledge of it, I would wish I had never sought, does not seem to recommend it to me as part of my good.

This suggests a different reforming definition for non-moral goodness. Suppose we say that “G is good for X” means that X would desire G if X was fully informed and ideally rational.

But this doesn’t seem quite right, either. Consider Joe, a philosopher lost in Chicago. A map would be good for Joe, but Joe wouldn’t desire a map if he was fully informed and ideally rational because then he would already know everything about the geography of Chicago.

So let’s try again. Railton suggests that “G is good for X” means that a fully informed, perfectly rational version of X would want the non-ideal X to want G in his circumstances. Joe’s fully informed and perfectly rational self would have no need for a map of Chicago, but it seems a fully informed and perfectly rational Joe would want the non-ideal Joe to have a map if the non-ideal Joe were lost in Chicago.

Thus, Railton proposes:

An individual’s good consists in what he would want himself to want, or to pursue, were he to contemplate his present situation from a standpoint fully and vividly informed about himself and his circumstances, and entirely free of cognitive error or lapses of instrumental rationality.

This is Railton’s ‘full-information analysis’ of what is non-morally good for a person.

Note also that this account leads to some predictions. As Miller puts it:

Firstly, since a person’s good is determined by what her fully informed and ideally rational self would want her to want given her actual circumstances, we would expect that, ceteris paribus, an individual will generally be a better judge of what is good for her than other people; secondly, that one’s knowledge of one’s good should increase as one’s informational base increases with experience; thirdly, since an agent’s ideal self will take into account what her non-ideal self is like as well as her actual circumstances, we would expect that people with ‘similar personal and social characteristics’ will tend, in similar circumstances, to have similar values; and fourthly, what is good for one person will tend to approach what is good for another ‘in those areas of life where individuals are most alike in other regards…

We haven’t done enough science to know if these predictions are correct, but they are certainly widely-held views.

In ways like this, Railton’s ‘full-information’ reduction of non-moral goodness can play a genuine role in explaining features of our experience, satisfying the first criterion. What about the second criterion? Railton poses the issue:

It does seem to me to capture an important feature of the concept of intrinsic value to say that what is intrinsically valuable for a person must have a connection with what he would find in some degree compelling or attractive, at least if he were rational and aware. It would be an intolerably alienated conception of someone’s good to imagine that it might fail in any such way to engage him.

Railton argues that his account of non-moral goodness does motivate people in the required way. He gives the example of Beth, a successful and happy accountant who wants to become a writer. When she has saved enough money, she quits her job to be a writer. However, what she writes fails to impress editors and publishers. Beth did not know it, but she lacks the skills and temperament needed to be a writer. But if Beth were fully informed and ideally rational, this “Beth+” would want the non-ideal Beth to lose her desire to become a writer.

In general, we can think of it like this: If you discovered an advisor who had the same basic psychological makeup and values as you do, but was perfectly rational and fully informed, you would take their advice seriously. Think of how often we say “If I could speak now to my younger self…”

So Railton says his ‘full-information’ account of non-moral goodness plays both the explanatory and normative role required for his reduction of non-moral goodness to be vindicative rather than eliminative.

 

Railton’s account of moral rightness

Railton uses the same strategy to offer a reforming definition of moral rightness and then argue that his account plays a genuine explanatory and normative role. He begins by considering what seems to be distinctive about moral norms:

Moral evaluation seems to be concerned most centrally with the assessment of conduct or character where the interests of more than one individual are at stake… [it] assesses actions or outcomes in a peculiar way: the interests of the strongest or most prestigious party do not always prevail, purely prudential reasons may be subordinated, and so on. More generally, moral resolutions are thought to be determined by criteria of choice that are non-indexical and in some sense comprehensive. This has led a number of philosophers to seek to capture the special character of moral evaluation by identifying a moral point of view that is impartial, but equally concerned with all those potentially affected.

…moral norms reflect a certain kind of rationality, rationality not from the point of view of any particular individual, but from what might be called a social point of view.

Railton proposes an idealized concept of social rationality in terms of

what would be rationally approved of were the interests of all potentially affected individuals counted equally under circumstances of full and vivid information.

Now Railton is ready to offer his reforming definition of ‘moral rightness’:

x is morally right if and only if x would be approved of by an ideally instrumentally rational and fully informed agent considering the question ‘How best to maximize the amount of non-moral goodness?’ from a social point of view in which the interests of all potentially affected individuals were counted equally.

In other words, moral rightness is a matter of what is instrumentally rational from an idealized social point of view.

Next, Railton must show that this account plays a genuine explanatory role and a genuine normative role.

 

The explanatory role

Consider Lonnie, who feels miserable and lethargic in a foreign country. He comes to desire milk, but unbeknownst to Lonnie his lethargy is due to the fact that he is dehydrated, and drinking milk will only make the problem worse. Of course, Lonnie+ would want Lonnie to desire water rather than milk, and so water but not milk is non-morally good for Lonnie (on Railton’s account of non-moral goodness). If Lonnie goes ahead and drinks the milk, we can explain his resulting misery in terms of not doing what is non-morally good for him – in terms of not doing what he woud desire himself to do if he was ideally rational and fully informed.

In the same way, Railton says we can explain things like social unrest in terms of a society not doing what is ‘morally right’ (that is, ‘instrumentally rational from an idealized social point of view’):

Just as an individual who significantly discounts some of his interests will be liable to certain sorts of dissatisfaction, so will a social arrangement – for example, a form of production, a social or political hierarchy, etc. – that departs from social rationality by significantly discounting the interests of a particular group that have a potential for dissatisfaction and unrest.

Of course, a society might believe it is doing what will best satisfy the interests of all those potentially affected, but be wrong about that. This is why the idealized (perfectly rational, fully informed) social point of view matters. Societies can cause social unrest even without meaning to do so.

 

The normative role

It must be noted that Railton rejects that the idea that moral obligation must always entail personal rational motivation:

On the present account rational motivation is not a precondition of moral obligation. For example, it could truthfully be said that I ought to be more generous even though greater generosity would not help me to promote my existing ends, or even to satisfy my objective interests. This could be so because what it would be morally right for me to do depends upon what is rational from a point of view that includes, but is not exhausted by, my own.

Railton illustrates this with an analogy to logical ‘oughts.’ Here is Miller:

Plausibly, I ought not to believe both a proposition P and some other proposition or propositions that entail its negation, not-P. But can it be said that simply in virtue of the truth of this ‘ought’ statement I have a reason to seek out and rectify all logical contradictions from my belief set? Think of the amount of intellectual effort this would require: I’d have to check my existing belief set for consistency (no easy matter) and ensure that every time I acquired a new belief it was consistent with my pre-existing belief set.

Logical oughtness does not entail that we always have a reason to adhere to logical oughts. Rather, it is enough that, as Railton says, “we are often concerned with whether our thinking is warranted in a sense that is more intimately connected with its truth-conduciveness than with its instrumentality to our peculiar personal goals.”

In the same way, we can account for moral facts without seeing morality as something it can’t be: “rationally compelling no matter what one’s ends.” Instead, perhaps we can “ask how we might change the ways we live so that moral conduct would more regularly be rational given the ends we actually will have.”

In the next post, I’ll examine some objections to Railton’s moral reductionism.

Previous post:

Next post:

{ 23 comments… read them below or add one }

Martin Freedman April 20, 2011 at 6:00 am

Pleased to see you writing this. If I had had a chance to restart my blog I would be focusing on ethics and not religion at all, a distraction partly driven by following your blog ;-)

Now I did write about Railton previously and criticised his idealisation of an agents desires given advice from agent+. I do not think his morality as social rationality needs it.

There are parallels between desirism and social rationality, for desirism one could say:
“x is morally right if and only if people generally have more all-things-considered and all-things-being-equal reasons to promote the desires that lead to x and inhibit the desires that lead to not x, over reasons to inhibit the desires that lead to x and promote the desires that lead to not x”
compared to
“x is morally right if and only if x would be approved of by an ideally instrumentally rational and fully informed agent considering the question ‘How best to maximize the amount of non-moral goodness?’ from a social point of view in which the interests of all potentially affected individuals were counted equally.”
The “people generally”, “all-things-considered” and “all-things-being-equal” conditions both implying Railton’s “ideally rational” and “full information” conditions and, also, Railton’s social point of view where “the interest of all potential affected individuals were counted equally”.

The key difference is in motivation, whilst both are still externalist and vindicatively reductionist, is that the desirism formations directly addresses the normative process whereas it is, as best, implied in Railton’s formation.

Another aborted project was to try and merge such formulations as the two above and also look at others such as Rawl’s reflective equilibirium and Scalon’s contractualism for example.

Anyway I am sure you are going in something like that direction but I will not anticipate where you end up.

Looking forward to future (and not too distant) posts.

  (Quote)

Luke Muehlhauser April 20, 2011 at 8:05 am

Cools. Sometime soon I gotta post an index of your recent posts on desirism.

  (Quote)

Martin Freedman April 20, 2011 at 10:26 am

Thanks but when I eventually restart my blog – no idea when – I will revisit all those posts and update them. Still go ahead if you think they are of any use as is.

  (Quote)

anti_supernaturalist April 20, 2011 at 1:11 pm

Scientific naturalism does not exist

Science offers no certainty about any of its views.

Fundies seem to need high levels of reassurance that what they read in texts is “for certain” the word of some suprahuman guarantor of certainty.

Certainty is a chimera — an illusion chased by Greek philosophy, and injected into xianity from middle neo-Platonic thought, early 3rd century CE. Ultimately however xian ‘certainty’ always resorts to irrational fideism.

Just so, philosophers and philosophically inclined scientists (e.g. E. O. Wilson) vainly aim at metaphysical and ethical certainty based on the dogma of reductionism fostered by Plato’s antipodes, Democritus and Epicurus

If ‘naturalism’ implies that theoretical entities in science provide an ever abiding ontology — “Science” is the “God” equivalent serving — then, naturalism simply can not exist. The current state of science never leads to any final state. It is not in the nature of things.

Einstein put the matter clearly: “As far as the laws of mathematics refer to reality, they are not certain; and as far as they are certain, they do not refer to reality.”

Like language, mathematics manipulates abstract concepts through manipulating signs. Neither language nor mathematics uses signs (or expressions or statements) found “in the world,” since there are no concepts in the world to find.

No god taught a primal adamic language. No god excogitated the universe through mathematics. No book of Nature lies open to be read. All concepts are cultural artifacts.

“Naturalism” seems best described as a heuristic

As a principle which subsumes Occam’s Razor — it would be a touchstone for theorizing generally — do not introduce novel theoretical entities nor invoke new forces without well-attested observable changes in behavior or relationships among processes in the world which seem by current theory to be inexplicable.

The whole alleged metaphysical realm of beings, objects, processes, and their relationships (could it exist), has no standing in science because ‘gods’ and ‘mathematical entities’ do not leave empirical footprints.

“Naturalism” understood this way is not hostile to metaphysics

— just indifferent to it. Absolutely the correct attitude towards any metaphysical claim found in a text, claimed to be inspired by “God” or by “Nature”.

True believers across the US demand that scientific knowledge and ethics should be dictated by political ideologies rooted in 16th century Protestantism, or 13th century Catholicism, or 12th century Islam.

True believers in metaphysical naturalism in the 19th century, whose reduction bases included newtonian particle mechanics, maxwell’s electrodynamics, and evolutionary theory, experienced culture shock as their reductionist programs deflated.

Ethics and social ethical theory founded on the “facts” of the old naturalism, ran out of hot air alongside scientistic philosophizing. Easily included on the list of failures: marxism and social darwinism.

Simultaneously, the post-Einstein high style of armchair scientism indulged in by the “radicals” Russell, Whitehead, young Wittgenstein, Carnap vainly sought a promised land, The Logical Structure of the (empirical) World. Their by now decadent descendants still do.

The jargon infested haunts of philosophy (nicely exemplified by this site) still reek of the seminary — There are altogether no metaphysical phenomena, only metaphysical interpretations of phenomena.

the anti_supernaturalist

  (Quote)

Garren April 20, 2011 at 2:03 pm

I found this an excellent refresher on Railton and don’t think I could top it as an introduction.

The whole step of considering what an ideal self would do seems unnecessary. We can skip it and talk directly about which available action would — in fact — best fulfill a person’s fundamental desires. This avoids the problem of Joe and his map, since we never need to consider the desires of an omniscient person.

Also, the focus on fundamental desires automatically covers non-fundamental desires like Beth’s desire to be a writer because she (mistakenly) thinks it will help her feel fulfilled or something similar. (If simply being a writer were one of her fundamental desires, her knowing she would suck at it would not be a reason for her to avoid it.)

Idealized people might be an interesting metaphor, but I think we can do the same work without the complications introduced by taking the metaphor too seriously.

  (Quote)

Martin April 20, 2011 at 3:33 pm

Hey Mr Muelhauser, do you have any recommendations for Internet sources/articles that clearly lay out the taxonomy of meta-ethics? There seems to be a good article on it at Wikipedia, but it conflicts with some things I read elsewhere: http://en.wikipedia.org/wiki/Meta-ethics

Why, for instance, is ethical subjectivism categorized under cognitivism?

  (Quote)

Luke Muehlhauser April 20, 2011 at 4:37 pm

Martin,

Chapter one of Miller’s Introduction to Contemporary Metaethics is good. Also, I’ll be doing a similar thing in a post for Less Wrong within the next few months.

  (Quote)

Garren April 20, 2011 at 6:53 pm

@Martin
..”Why, for instance, is ethical subjectivism categorized under cognitivism?”

If my ethical theory says that “I have positive vibes toward X” is the same as “X is morally good for me to do,” then a moral claim like “Feeding the homeless is morally good” can be true or false depending on whether I do, in fact, have positive vibes toward feeding the homeless.

Cognitivism only requires moral language be the sort of language that can be true/false. In non-moral terms, “I like dogs” would be cognitivist, but “Yay dogs!” would not be cognitivist.

  (Quote)

Martin Freedman April 21, 2011 at 12:24 am

Garren

It seems we agree on the point over the unnecessary step of idealised desires.

This is common to other desire-based reductive approaches, all writes such as Griffin, Boyd and Brink and Brandt (IIRC) feel the need to eliminate “mere desires” and solve the problem of prudence objectively (that is even if the agent is unable to see it) first – with “informed” or “rational” desires before getting to the issue of morality. Apart from anything else akrasia is a real problem for many if not all people anyway, and whilst that cannot count as a legitimate excuse it needs to be incorporated into any real-world ethical model. However morality as social rationality without idealised agents (and with recognising akrasia) still works as I said.

I am puzzled by what you meant by fundamental versus non-fundamental desires. Certainly there are some desires that are necessary as required in virtue of being a human being – the desires to eat, sleep, drink, sex and so on. Still I am not sure that is what you meant.

  (Quote)

CharlesR April 21, 2011 at 4:45 am

“But if Beth were fully informed and ideally rational, this “Beth+” would want the non-ideal Beth to lose her desire to become a writer.”

This doesn’t follow. She should have been writing while she was working so when she quit her job she had the skills and discipline. Neither of these are innate qualities. If she has the desire they are both things that she can learn.

  (Quote)

Martin April 21, 2011 at 8:49 am

Garren,

Interesting. I couldn’t see much difference between “I like dogs” and “Yay dogs!”. I still can’t, really. Aren’t they both subjective opinions? Aren’t these two statements synonymous?

  (Quote)

Garren April 21, 2011 at 8:58 am

@Martin,
..”I am puzzled by what you meant by fundamental versus non-fundamental desires. Certainly there are some desires that are necessary as required in virtue of being a human being – the desires to eat, sleep, drink, sex and so on. Still I am not sure that is what you meant.”

I don’t take any desire to be necessary, though some are obviously very frequently held by humans. What I mean by a ‘fundamental’ desire is one which is not held just because a person thinks it will help her fulfill other desires. It’s similar to the distinction made between instrumental and intrinsic value, without the projectivism.

fundamental desire + belief -> instrumental desire[s]

(And an instrumental desire can have subordinate instrumental desires of its own.)

By way of example, let’s say Cole has the following desire and belief structure:

..Desire to be with my family on Christmas (fundamental) + Belief that my family is in the next town -> Desire to travel to the next town

..Desire to travel to the next town + Belief that the main road with a big bridge is the most reliable way to get to the next town by Christmas -> Desire to take the main road

Cole desires to take the main road, unless he finds out the bridge is out. Cole desires to travel to the next town, unless he finds out his family is in another town. But if Cole’s desire to be with his family is truly fundamental, no additional information will nullify that desire.

  (Quote)

Garren April 21, 2011 at 9:06 am

@Martin,
..”Interesting. I couldn’t see much difference between “I like dogs” and “Yay dogs!”. I still can’t, really. Aren’t they both subjective opinions? Aren’t these two statements synonymous?”

Oops. I was writing the previous response when you posted this.

If it’s true that yay dogs!, then I’m probably not a neat freak.

If it’s true that I like dogs, then I’m probably not a neat freak.

See how the second one fits much better into a True/False logical form?

  (Quote)

Martin April 21, 2011 at 9:23 am

Hmm. I see how the difference there. But isn’t that just the language. Underlying the language, isn’t saying “I like dogs” the same emotional response as “yay dogs!”? Stripping out the language gives you the same thing, doesn’t it?

  (Quote)

martin April 21, 2011 at 9:44 am

Luke,

Thanks for the Miller recommendation. Do you have a similar recommendation for normative ethics as well?

  (Quote)

Garren April 21, 2011 at 12:05 pm

@Martin,

Yeah, the closeness is probably why individually subjective, cognitivist theories aren’t much more esteemed than non-cognitivist theories. Besides divine attitude theories.

  (Quote)

Luke Muehlhauser April 22, 2011 at 10:32 am

martin,

Alas, no. Kagan’s Normative Ethics and Singer’s Practical Ethics are commonly used, though.

  (Quote)

antiplastic April 22, 2011 at 11:04 am

Hmm. I see how the difference there. But isn’t that just the language. Underlying the language, isn’t saying “I like dogs” the same emotional response as “yay dogs!”? Stripping out the language gives you the same thing, doesn’t it?

Whether or not you like dogs can be true or false, so when you say “I like dogs”, you are either telling the truth or telling a falsehood.

But if you see a dog and you smile, applaud, and say “yeah!”, it makes no sense to say that smiling and applauding are true or false statements.

  (Quote)

stag April 22, 2011 at 9:50 pm

martin,

as antiplastic, above. Susceptibility to formalization within a system of logic is a property of the statement “I like x”, but not of “x – yeah!”. Likewise for the statement “x is morally wrong” and “x – booo!”. So if there is a property that pertains to moral statements which does not belong to expressions of emotion, we are not at liberty simply to ignore that property: it might be the key to the difference between morality and emotive responses.

And as it happens, it is the key: morality has a strong logical component, whereas emotional responses don’t. Right and wrong are at least partly about what we think and understand; whereas emotions are exclusively about what we feel.

Note that I don’t deny that emotions have any part in moral life. They have a strong part. But so does our intelligence.

  (Quote)

stag April 22, 2011 at 10:33 pm

garren

The recession of instrumental desires in the direction of fundamental desires: I think you do arrive at one that is necessary, namely, the desire for happiness/well-being in the broad sense. Every desire anyone has always includes the desire to be happy in its structure. Why does x want to be at home with his family? Because he desires to be happy, and he is happy with his family. That’s not the only reason, but it is necessarily involved. The medieval philosophers used to say, “bonus habet rationem finis”, or The good has the nature of an end. The terminus of a desire is uncontroversially an end. It is also necessarily, from a certain point of view, a good: I can’t desire anything unless I perceive, at some level, some good in it. But the good-for-one-more-of-my-appetites in general, whether rational or not, is just happiness/well-being in the broad sense. For this reason, I think that it is a necessary fundamental desire.

  (Quote)

stag April 22, 2011 at 10:35 pm

I meant to say the-good-for-one-or-more-of-my-appetites

  (Quote)

stag April 22, 2011 at 10:50 pm

anti-supernaturalist:

Your thinly erudite rant impresses no-one.

  (Quote)

cl April 24, 2011 at 3:23 pm

But Railton resists such a crude reduction of non-moral goodness:

My [actual desires] frequently reflect ignorance, confusion, or lack of consideration, as hindsight attests. The fact that I am now constituted that I desire something which, had I better knowledge of it, I would wish I had never sought, does not seem to recommend it to me as part of my good.

This suggests a different reforming definition for non-moral goodness. Suppose we say that “G is good for X” means that X would desire G if X was fully informed and ideally rational.

But this doesn’t seem quite right, either. Consider Joe, a philosopher lost in Chicago. A map would be good for Joe, but Joe wouldn’t desire a map if he was fully informed and ideally rational because then he would already know everything about the geography of Chicago.

I think you can resolve your trouble here by applying a more focused definition of fully informed. You seem to be using fully informed as a synonym for omniscient there, and of course, they are synonymous, but I suspect Railton might have been going for something more along the lines of, “fully informed of the repercussions of X fulfilling the desire for G.” May someone correct me if I’m wrong.

  (Quote)

Leave a Comment