Saturday, October 29, 2011

Two Kinds of Intrinsic Value

There are at least two major, distinguishable meanings for “intrinsic value.” Unfortunately, these meanings often go unspecified, which results in a lot of unnecessary confusion. Christine Korsgaard covered this in her 1983 paper, “Two Distinctions in Goodness" from which I'll be borrowing.

The first kind of intrinsic value is contrasted with instrumental value. Instrumental value is the value something has because it's helpful or supportive of something else which has value; think “derivative value.” For example, an important point in how we treat (other) animals is whether there is any reason to consider factors beyond the value of animals to human well-being.

The second kind of intrinsic value is value that is located inside as opposed to outside the thing that is valuable. As Korsgaard put it, “It refers, one might say, to the location or source of the goodness rather than the way we value the thing.”1 This kind of intrinsic value might be, for example, a property of the valuable thing itself which does not depend on anyone in the world valuing it.

So if I claim old trees have intrinsic value, it's not clear whether I'm saying that old trees have non-derivative value or whether I'm saying old trees have value regardless of anyone valuing them. Suppose I personally and directly value old trees. Also suppose that old trees need someone to value them in order to have value. In this situation, old trees have the first kind of intrinsic value but lack the second kind of intrinsic value.

Korsgaard considers the first kind of intrinsic value to be something of a misnomer, since only the second kind of intrinsic value contrasts with extrinsic value. Whether value is derived or not is simply another issue.

1. Korsgaard, C.M. (1983). Two distinctions in goodness. The Philosophical Review. 92(2). [search link]

Tuesday, October 11, 2011


Expressivism is a type of moral theory which says that moral judgments are expressions of desire-like attitudes, rather than true/false assertions of fact. This straightforwardly explains why it would be odd for a person to judge something to be morally right or wrong and then be indifferent towards it.

For simple moral utterances like "Slapping children is wrong" or "You ought to watch out for pedestrians," it can make sense to understand these as expressions of attitudes. The big problem for expressivism is explaining what's going on in other sorts of moral utterances like:

  • Is it wrong to slap children?
  • If it's wrong to slap children, then it's wrong for my neighbor to slap her son.
  • Either it's wrong to slap children, or I've been misinformed.

These sure seem to involve true/false logic! While some expressivists have tried to show that appearances are deceiving and these sentences don't — after all — involve true/false logic, others have embraced a hybrid view which includes both the expression of attitudes and some true/false logic in the meaning of moral language.

Daniel Boisvert's Expressive-Assertivism is one of these hybrid forms of expressivism.1 It was inspired, of all things, by the way ethnic slurs work (which I'll explain as I go over the three core features of Expressive-Assertivism).

First Core Feature: Dual-Use Principle

In normal circumstances, a person who speaks a simple moral sentence like "Slapping children is wrong" is performing two distinct speech acts (two "direct illocutionary acts"). One speech act is expressive and the other is assertive.

Likewise, the simple use of an ethnic slur "John is chink" simultaneously asserts something about John (that he's of Chinese descent) and expresses contempt for people of Chinese descent.

Second Core Feature: Extensionality Principle

In normal circumstances, a person who uses moral terms in any so-called extensional context still performs the same kind of expressive speech act as in simple moral sentences. The three "problem" sentences above are examples of extensional contexts, but "wrong" in "Lisa thinks that slapping children is wrong" is in an intensional context which falls outside the scope of this core feature.

The hateful attitude expression is still present when people use slurs in sentences like "Is John a chink?" and "Either John is a chink, or I've been misinformed." And notice how it's the same sort of contempt for people of Chinese descent we saw in the simple case.

Third Core Feature: Generality Principle

In normal circumstances, the expressive speech act is not aimed directly at the object of moral judgment, but at everything in a broader category.

Using the word "chink" normally expresses a contempt for all people of Chinese descent. This is why the negative attitude expression is still at full force when John's membership in the hated category is uncertain.

Solving 'The Moral Problem'

Michael Smith famously characterizes the central problem of metaethics as finding a way to show that the following two propositions "are both consistent and true" with regard to each other and with a Humean theory of motivation:
1. Moral judgments of the form 'It is right that I φ' express a subject's beliefs about an objective matter of fact, a fact about what it is right for her to do.
2. If someone judges that it is right that she φs, then ceteris paribus, she is motivated to φ.2
Expressive-Assertivism provides an account for both elements. The assertive speech act concerns the "objective matter of fact" of whether the judged action, practice, etc. has a certain property. The expressive speech act directed at all things which have that property explains an individual's motivation.

It also helps that the way Expressive-Assertivism goes about solving the moral problem is so similar to the way another kind of value judgement — the ethnic slur — plausibly works.

Answering the Objection from Missing Expressives

These next three sections break what is commonly known as the Frege-Geach problem (aka the embedding problem) into three more specific objections.

The 'missing expressives' objection concerns sentences like:
If it's wrong to slap children, then it's wrong for my neighbor to slap her son.
where it may seem like the speaker can't be expressing his own attitude at that moment. At least not like someone who says "Slapping children is wrong." But Expressive-Assertivism explains both cases as expressions of attitude toward things that are wrong in general (whatever 'wrong' means). This is the Generality Principle at work. Just because a speaker isn't sure whether an action or a person belongs to a despised category, doesn't mean his use of moral language (or the language of ethnic slurs) is expressive free.

In short, the expressives are there. They're just aimed more broadly.

Answering the Objection from Incomplete Semantics

Simple versions of expressivism run into trouble when they claim sentences like "Slapping children is wrong" only expresses an attitude, because this leaves at least some of the meaning of "slapping children is wrong" unexplained in sentences that start "If slapping children is wrong, then ...."

[expression of attitude].
If [expression of attitude], then ....
What the heck is the second kind of sentence supposed to mean? Happily, this isn't such a problem for Expressive-Assertivism since the true/false 'assertive' component makes normal sense in an 'if..then' context. A full reading with a reduction of 'wrong' might go something like this:
If slapping children is harmful on balance [expression of negative attitude toward things that are harmful on balance], then ...
I should be clear that Expressive-Assertivism as a theory doesn't have anything to say about whether wrongness is a property that can be reduced like this. A philosopher who agrees with G.E. Moore about moral properties being indefinable can adopt Expressive-Assertivism. She would just have the usual issue of explaining why we have strong attitudes toward such properties.

Answering the Objection from the Ambiguity of Attitude-Attribution Verbs

Consider these two sentences:
Sarah believes that her brother pickpockets.
Sarah believes that pickpocketing is wrong.
The first sentence attributes a true/false belief to Sarah. Under simple expressivism, the second sentence could only attribute an attitude to Sarah. But it would be odd if the phrase "believes that" is associated with true/false belief unless it's followed by ethical vocabulary.

As a hybrid theory, Expressive-Assertivism's account of the two sentences isn't quite so disparate, but the objection could still be pressed by saying it would be odd for "believes that" to only attribute a true/false belief, unless followed by ethical vocabulary in which case an attitude attribution is suddenly tacked on.

Boisvert responds by saying "believes that" attributes a psychological state potentially composed of both true/false belief and attitude to the subject. The same psychological state, in fact, that a person normally possesses when they utter just the clause after "believes that."

For example, when Sarah says "Pickpocketing is wrong" she would normally hold an attitude toward things that are wrong and also hold a true/false belief that pickpocketing fits in that category. If Jack says "Sarah believes that pickpocketing is wrong" then he would be attributing both the belief and the attitude to Sarah as part of her overall psychological state (try removing either element and Jack's claim is weaker than we normally understand it). Here's the key point: if Jack says "Sarah believes that her brother pickpockets" then Jack is still attributing a psychological state to Sarah, even if the state happens to lack an attitude component.

In programming terms, Boisvert would be saying "believes that" is like a single function which takes a structure rather than a simple variable as an argument. It wouldn't be like multiple, overloaded functions as the objection alleges. (I hope at least one person finds this analogy helpful!)

Wrapping Up

Expressive-Assertivism has two more attractive features covered in Boisvert's paper which I won't elaborate on: "it holds that the descriptive content of moral sentences is non speaker-relative" and "it is consistent with, but is not forced to accept, minimalism about truth."

Overall, I think Expressive-Assertivism is on the right track. Not surprising since I intentionally set out looking for something like it because I've held a rough kind of hybrid expressivist view for a while and knew someone had to be advocating a theory in the neighborhood. Boisvert isn't the only one working on hybrid theories, so you can expect to see more comparing and contrasting from me in the near future.

Added: "Near" turned out not to be so near. I still plan on getting back to this topic eventually.

1. Boisvert, D.R. (2008). Expressive-assertivism. Pacific Philosophical Quarterly. 89(2). p. 169-203. [direct link]
2. Smith, M. (1994). The moral problem. Oxford: Blackwell. p. 184.