Meta-ethics: Cornell Realism

by Luke Muehlhauser on April 18, 2011 in Ethics

For the purposes of later work on meta-ethics, it will be helpful for me to summarize Cornell Realism for my audience. To do so, I’ll be summarizing some of chapter 8 in Miller’s Introduction to Contemporary Metaethics.

A cognitivist theory of meta-ethics holds that moral judgments are truth-apt – they are the kinds of things that can be true or false. That is, when I say “Murder is wrong,” I’m not just expression an emotion (as many non-cognitivists would say), I’m trying to make a claim that I think is true.

One cognitivist theory called error theory holds that moral judgments are truth-apt but always false. This is analogous to atheism, which is an ‘error theory’ about God. Atheists usually think that people are making claims about God, but that these claims are systematically untrue because God doesn’t exist. Likewise, the moral error theorist says that people make claims about moral facts, but these claims are systematically untrue because moral facts don’t exist.

Another cognitivist theory called Moorean intuitionism says that moral judgments are truth-apt and that moral facts do exist but are not part of the natural world.

Other cognitivist theories hold that moral judgments are truth-apt and that moral facts exist in the natural world. Among ethical naturalists, there are reductionists and non-reductionists. Reductionists think that moral properties reduce to non-moral properties. Non-reductionists think that moral properties are irreducible natural properties in their own right.

Today we look at a cognitivist, naturalistic, non-reductionist theory of meta-ethics: Cornell realism. (It is named this because it was developed by professors with close connections to Cornell University.)

But first: what would it mean for moral properties to be natural properties if they are not reducible to non-moral properties? According to non-reductionists, moral properties ‘are constituted by’, or ‘are multiply realized by’, or ‘supervene upon’ non-moral properties, but they do not reduce to non-moral properties. Miller illustrates the difference by asking us to consider the moral property of rightness:

We can imagine an indefinite number of ways in which actions can be morally right. [Non-reductionists] think that, in any one example of moral rightness, the rightness can be identified with non-moral properties (e.g. the handing over of money, the opening of a door for someone else, etc.). But they claim that, across all morally right actions, there is no one non-moral property or set of non-moral properties that all such situations have in common and to which moral rightness can be reduced.

For example, one might argue that certain natural kinds like ‘catalyst’ or ‘organism’ are not obviously reducible to natural kinds in physics, and that mental types like being in pain is not necessarily reducible to neurological types like being in a state of C-fibre stimulation, and yet these things play a role in successful scientific explanations. Likewise, it might be that moral properties like ‘rightness’ are not clearly reducible to natural kinds in physics, and yet they are natural properties that can play a successful role in explanations.

 

Harman’s objection

But do moral properties play a role in successful explanations? Gilbert Harman thinks not – at least, not unless moral properties are reduced to natural properties (which is something that non-reductionists by definition don’t do).

Imagine some hoodlums pouring gasoline on a cat and setting the poor creature on fire. Does the wrongness of their action pull its weight in our explanation of the action? No. We can explain the action by appeal to ordinary psychological facts. And does the wrongness of this action pull its weight in our explanation of why witnesses came to believe that the action was wrong? No. Again, we need only appeal to ordinary psychological facts – for example those of evolutionary psychology and cultural influence.

We can explain everything about this story with ordinary non-moral facts. Moral facts are shaved off by Occam’s razor as unnecessary.

Nicholas Sturgeon replies that Harmon’s reductionist requirements are too strict, and then offers an example of an explanatorily efficacious moral fact:

…the fact that Hitler was morally depraved explains (at least in part) why he instigated and oversaw the death of millions of persons; the fact that Hitler instigated and oversaw the death of millions of persons explains why I believe that Hitler was morally depraved. So the fact that Hitler was morally depraved explains (in part) why I believe Hitler was morally depraved.

Harman replies that no mechanism has been offered for how the wrongness of Hitler’s acts is supposed to explain how we come to believe that Hitler is morally depraved. It is not as though the wrongness of Hitler’s act affects the quality of light reflected in the observer’s eye, or something.

Of course, the debate between Cornell realists and their critics is far more complex than this, so if you want more of the story you should read Miller’s chapter on it and continue from there.

Previous post:

Next post:

{ 16 comments… read them below or add one }

jimrandomh April 18, 2011 at 5:07 am

I think the “explaining vs explaining away” distinction is relevant here. The hoodlums setting fire to a cat can be explained by appeal to ordinary psychological facts, which in turn can (in principle) be explained by brain structures and biochemistry. But these explanations do not make the wrongness of their act disappear; we have explained it, but not explained it away.

Much of moral theory is just a matter of definitions – ie, how you define words like “should” and “moral”. Based on your description (I’d never encountered it previously), error theory is a sort of null definition – a way you can define those terms to make all moral statements false. And when you think of it that way, it’s a pretty stupid set of definitions to use, because it’s obviously not what people mean when they use those words.

Personally, I’m a cognitivist and a reductionist, and I hold that if you reduce the definitions of moral terms far enough you must necessarily find a reference to human brains making moral judgments somewhere along the way. But I’m a bit unclear on what the exact path of reductions should look like – I see several different ways to go about reducing from “should” to “brains” with no apparent way to choose between them. Fortunately, the different reductions agree with each other in broad qualitative terms in most cases.

  (Quote)

Luke Muehlhauser April 18, 2011 at 7:08 am

jimrandomh,

Yes, but reduction is not available to the Cornell realist, because Cornell realism is a non-reductive account of morality. That’s why the move you suggest is unavailable to them, but it is exactly what Railton and Jackson and others do.

You are correct about definitions and reduction and multiple reductions. You might be able to guess where I’m going if I tell you that the metaethical theory I’ll be defending in a Less Wrong sequence (shortly) is what I call Pluralistic Moral Reductionism.

  (Quote)

Garren April 18, 2011 at 8:11 am

I take it being a well-formed sentence is a non-natural property because there are no specific words in common with all well-formed sentences.

(Thanks for the Twitter hits. And you want ‘read’ not ‘reader’ in the last line of your post.)

  (Quote)

Citizen Ghost April 18, 2011 at 9:03 am

I have one question and one comment.

1. If “error theory” constitutes a a theory within cognitivism and proponents of “error theory” holds that moral facts don’t exist, in what way can they be said to be moral realists? How would this view differ from the view that objective morality does not exist and therefore, moral statements and judgments are expressions of either individual emotion, group consensus and/or subjective opinion?

2. I hope it’s not too presumptious of me to offer a suggestion. Not that you need any help. I’ve noticed that your posts about the Harris-Craig debate generated lots of comments and lively debate. These posts get considerably less feedback but also I know that you are now more interested in exploring other issues in philosophy, meta-ethics, etc. One way to take your “audience” with you on your journey might be introduce concepts such as this one through the prism of issues raised in the debate. For example, Harris offers a particular kind of moral theory in which morality is real and rooted in nature. In providing an analysis or criticism of Harris, you can open windows into these other moral theories and consider where his theory fits in and what other philosophers have to say. But maybe, in a sense, you are doing exactly that.

  (Quote)

lackofcheese April 18, 2011 at 9:54 am

Citizen Ghost,

If “error theory” constitutes a a theory within cognitivism and proponents of “error theory” holds that moral facts don’t exist, in what way can they be said to be moral realists?

Error theory is cognitivist, but it is anti-realist.

How would this view differ from the view that objective morality does not exist and therefore, moral statements and judgments are expressions of either individual emotion, group consensus and/or subjective opinion?

It’s unclear what is meant by “objective”, but the crucial claim of cognitivism is that moral judgements are statements of a kind that can be true or false. On the other hand, emotions are not true or false in this sense.

  (Quote)

Garren April 18, 2011 at 10:14 am

@Citizen Ghost
..”How would [error theory] differ from the view that objective morality does not exist and therefore, moral statements and judgments are expressions of either individual emotion, group consensus and/or subjective opinion?”

Compare it to religion. Most Atheists believe God-statements are intended to be taken as realist assertions about God by those who think God exists. This is different from taking God-statements as metaphors or expressions of emotions which aren’t meant to talk about a God person whose non-existence renders them false.

Some liberal theologians do go the second route, then act like Atheists are dense. Ugh.

  (Quote)

Garren April 18, 2011 at 10:28 am

“Pluralistic Moral Reductionism”

Since my own view is pluralist and reductionist, I’m looking forward to the more detailed description.

  (Quote)

Esteban R. (Formerly Steven R.) April 18, 2011 at 9:24 pm

Why aren’t there more responses? I read this in the morning and found it really interesting and was hoping to see more than the 7 responses that I saw.

So, to see if I understand, what Sturgeon is trying to do with his example of Hitler is to show that non-reducible moral properties can successfully explain why people disapprove of Hitler, right?

If so, I’m extremely dissatisfied with it; I don’t think Hitler saw himself as morally depraved or even that his actions are properly explained as being morally depraved. It seems to me that his actions were the culmination of various attitudes in Europe towards nation, government, minorities and “traditional values” and that he saw himself as acting on behalf of such sentiments. In fact, the reason we condemn Hitler’s actions is because ultimately the goals of peace, stability and prosperity promised by Fascists was met and violated; we do not see people attacking these aims, but rather, the actions that didn’t lead to this actions. I don’t think “Hitler was evil” comes even close to explaining his actions, motivations or anything at all unless we reduce it to the cause and effect of his actions and then, to take something from Garren’s blog, how “helpful” it is to reaching goals we want for society.

—-

Garren, I know your comment about proper sentence structure wasn’t serious, but…what would a natural property sentence structure be (or are there none but you were just making a point about the type of reasoning used by non-reductionists)? I know I’m also being vague here, but care to elaborate on your objection? Thanks.

—-

Citizen Ghost: I get the suspicion that less comments are generated because this actually requires a lot of thought (I know it does for me!) and getting to know new terms. Many people who come to this blog seem to already be familiar with the basics of theological discussion and debate, but when it comes to other fields, they’re less knowledgeable. So, even if Luke did that, the lack of familiarity seems to discourage any type of debate or insights; after all, most of the debate in those sections revolved around the same tired squabbles of before and not so much on moral theory.

  (Quote)

Esteban R. (Formerly Steven R.) April 18, 2011 at 9:45 pm

In fact, the reason we condemn Hitler’s actions is because ultimately the goals of peace, stability and prosperity promised by Fascists was met and violated; we do not see people attacking these aims, but rather, the actions that didn’t lead to this actions.

*The actions that didn’t lead to the fulfillment of those promises that we view as positive.

  (Quote)

Garren April 18, 2011 at 11:04 pm

Esteban,
..”Garren, I know your comment about proper sentence structure wasn’t serious, but…what would a natural property sentence structure be […]”

Yeah, I was being facetious.

What I’m suggesting is that the property of being a well-formed sentence is a (natural) higher order syntactical pattern, not relying on a common substantial element. In other words, well-formed sentences don’t need to have a specific word or set of words in common.

Of course this is obvious with sentences, which is why I was using it as a metaphor for the moral evaluation of actions. Just as we don’t look for particular words in a sentence to count as well-formed, we aren’t looking for particular elements in an action to count as good or evil. That’s the dead end that led Moore to giving up and advocating mystical intuition. We might as well give up on grammar saying, “I know well-formed sentences when I hear them, and that’s the end of it!”

The solution in both cases is to investigate what process our minds are automatically running to evaluate sentences and situations. We’re looking for the higher-order ‘grammar’ of moral evaluation, not some spooky property we apprehend with a mystery faculty!

The worry is that — if we do come to understand how our moral judgments work — this might undermine all our motivation to act morally. I look at it as an opportunity to affirm moral intuitions based on what we truly value and reject evolved/encultured moral intuitions which go against what we truly value. Moral normativity is a product of our valuing and the way the world works so as to bring about what we value.

One might object, “But what should we most fundamentally value?” I consider this a bad question, based on the mistaken assumption that ‘oughtness’ is prior to valuing. It’s not. Only a more fundamental valuing can place an ‘ought’ on another valuing, but once we’re down to fundamental valuing, we have hit bedrock.

I’ll stop rambling now.

  (Quote)

James Gray April 19, 2011 at 12:24 am

Gilbert Harman thinks not – at least, not unless moral properties are reduced to natural properties (which is something that non-reductionists by definition don’t do).

I thought you said this is a naturalistic theory. Naturalism should not be equated with reductionism nor should natural facts be equated with reductionistic facts.

Harman wants to say that moral properties can’t explain anything unless they are reducible to nonmoral properties. That certainly isn’t obvious. Mental properties seem to explain nonmental properties, for example.

  (Quote)

Garren April 19, 2011 at 1:21 am

@James Gray,
..”Naturalism should not be equated with reductionism”

Eh, I’m with Glenn Peoples on this one:

“If you think that moral goodness is constituted or identical with (or caused by) other facts, then you’re an ethical naturalist. If however you believe in a free-standing property of moral goodness, then you’re an ethical non-naturalist.”

— from http://www.beretta-online.com/wordpress/2009/nuts-and-bolts-005-ethical-naturalism/

P.S. — I quote you, then disagree with you in my most recent blog post. But I am quoting you in venerable company, if that helps!

  (Quote)

Citizen Ghost April 19, 2011 at 5:52 am

Lack of Cheese.

Error theory is cognitivist, but it is anti-realist

Thanks. I came away from the initial post thinking that cognitivism was being explained as a category of moral realism. My error. And I agree that this approach to moral theory is considerably more useful than Craig’s approach of arguing for “objective morality” a vague term, more useful for rhetoric than for serious philosophical inquiry.

Garren,

Thanks – that’s a helpful explanation.

Esteban,

You’re certainly correct that these kinds of posts require a lot more thought than simply weighing in on debates. Naturally, the debates provide more entertainment value and I confess that I am not above such cheap thrills. But we can also see that individuals are interested in “atheism” for different reasons. For some, like Luke, the philosophy itself is the thing of greatest interest. For others, the philosophical arguments are just a means to an end. For many individuals (both proponents and opponents of atheism), the “culture wars” are the thing and debates, such as the one between Craig and Harris, are an arena for checking out the intellectual and rhetorical ammunition.

  (Quote)

JoeK April 20, 2011 at 8:01 am

Forgive me, for my philosophical sophistication is limited, but this confused me:

“[Non-reductionists] think that, in any one example of moral rightness, the rightness can be identified with non-moral properties (e.g. the handing over of money, the opening of a door for someone else, etc.). But they claim that, across all morally right actions, there is no one non-moral property or set of non-moral properties that all such situations have in common and to which moral rightness can be reduced.”

This seems to me analogous to:

“Non-reductionists think that, in any one example of [computation], the [computation] can be identified with non-[computational] properties (e.g. the [movement of electrons], the [firing of synapses], etc.). But they claim that, across all [computational events], there is no one non-[computational] property or set of non-[computational] properties that all such situations have in common and to which [computation] can be reduced.”

Which seems both obviously true of computation, and also perfectly compatible with a reductionist view of computation: every computation can be reduced to a collection of physical events — one merely needs to understand the manner in which the physical system represents computation. I don’t understand why the original statement is framed as a NON-reductionist view of morality. What am I missing? Or, how is my analogy flawed?

  (Quote)

antiplastic April 20, 2011 at 3:22 pm

Forgive me, for my philosophical sophistication is limited, but this confused me:

“[Non-reductionists] think that, in any one example of moral rightness, the rightness can be identified with non-moral properties (e.g. the handing over of money, the opening of a door for someone else, etc.). But they claim that, across all morally right actions, there is no one non-moral property or set of non-moral properties that all such situations have in common and to which moral rightness can be reduced.”

This seems to me analogous to:

“Non-reductionists think that, in any one example of [computation], the [computation] can be identified with non-[computational] properties (e.g. the [movement of electrons], the [firing of synapses], etc.). But they claim that, across all [computational events], there is no one non-[computational] property or set of non-[computational] properties that all such situations have in common and to which [computation] can be reduced.”

Which seems both obviously true of computation, and also perfectly compatible with a reductionist view of computation: every computation can be reduced to a collection of physical events — one merely needs to understand the manner in which the physical system represents computation.I don’t understand why the original statement is framed as a NON-reductionist view of morality.What am I missing? Or, how is my analogy flawed?

Your analogy is precisely correct, but your analysis elides the distinction between type and token reductions.

Every time an object heats up, this type of phenomenon can be reduced to phenomena of kinetic energy of individual atoms. But computers can be digital, or analog, electrical or steam powered, made of silicon or made of tinkertoys. In each individual case, a computer does what it does in virtue of the properties of its parts. But there is no such thing as “what being a computer reduces to” that is the same in all cases — nothing analogous to what kinetic molecular energy is for heat, and hence being a computer is non=reducible.

Another way this is put is saying that computation is a functional property, or that it is multiply realizable. Functionalism in phil-mind is another such view.

  (Quote)

Martin Freedman April 21, 2011 at 12:56 am

Anitplastic

That is a very good observation, helped given my background in cognitive science so your points made immediate sense and helps me make much better sense of Cornell Realism as a kind of moral functionalism.

(It must be noted that moral functionalism is a distinct although related form of ethical naturalism (see SEP on Cornell Realism followed by Moral Functionalism) promoted by Jackson who is a reductionist on moral properties. That is an interesting approach and I hope Luke will blog on that.)

  (Quote)

Leave a Comment