What Is Harman-Style Relativism?

I am intrigued by Gil’s suggestion that a significant number of people may be moral relativists. The claim is an empirical one, of course, and since Shaun Nichols, Jonathan Weinberg and I have recently been doing some cross-cultural studies of epistemic views – finding some interesting differences – I started to wonder whether there might not be interesting cross-cultural differences about in the extent to which people are relativists. But in discussing how we might go about testing that hypothesis, I quickly hit a roadblock, because I continue to have trouble understanding just what Harman means by relativism, and thus how we might determine that someone is or is not a relativist. In this note I’ll try to explain why I find Gil’s account of relativism hard to understand.

In the material we’ve read (including his web page note) and in his comments in the seminar, Gil has used two rather different strategies to explain what he means by relativism. Let me consider these one at a time.

I. The “assigning truth conditions” account: In the Harman/Thomson volume (H/T) and again in the web note, the locution “assigning truth conditions” plays a central role in explaining relativism. In H/T Gil says:

To a first approximation, moral relativism makes the following claim about moral judgments:

(2) For the purposes of assigning truth conditions, a judgement of the form, it would be morally wrong of P to D, has to be understood as elliptical for a judgment of the form, in relation to moral framework M, it would be morally wrong of P to D…. [p. 4. Italics in the original; boldface added.]

In the seminar, Gil said something about this formulation having been influenced (perhaps influenced too much) by Thomson, and offered another account which I’ll discuss below. However, in the web note, Gil used a very similar cluster of locutions. Here’s a quote:

Stich notes an analogy between moral intuitions and grammatical intuitions. "The person who takes his moral intuitions to be universal and objective is like the person who takes his grammatical intuitions to be universal and objective – and who thus views people who speak different dialects or languages as mistaken about something."

Would all of this person's grammatical claims be false because of this? No. Except where he is specifically making claims about the other people who speak different dialects or languages, the best way to assign truth conditions would seem to be in relation to that person's language. If the person's grammatical claims are true of his dialect (or idiolect) then they should be counted true.

Similarly, according to one form of moral relativism, even if a certain speaker's ordinary moral judgments falsely present themselves as objectively and universally true, etc., the best account of truth conditions of most of these judgments counts them true if they are true in relation to the speaker's moral framework. [Boldface added.]

The reason that I find these comments puzzling is that they invoke a locution that is no part of most people’s “ordinary language”. If you ask my Grandmother (or a philosophically naïve freshmen who has just signed up for his first philosophy course) “How does S have to be understood for the purpose of assigning truth conditions?” they would not have any clear idea what you are talking about. I would share their puzzlement. Assigning truth conditions is not an activity in which I participate with any frequency. And if you asked me to do it, I would want to know a lot more about the task. For what purpose are these truth conditions being assigned? How do we judge whether one assignment is better than another? Until these are spelled out, I simply don’t understand what is being claimed when it is said that this or that is “the best way to assign truth conditions”. And, of course, the problem is if anything more acute for the naïve subject whose relativistic proclivities we might want to assess. Does the freshman really think that this or that is the best way of assigning truth conditions? How could she, until you tell her a lot more about the goals and constraints that are in play in the assigning-truth-conditions-game.

Let me be clear about the nature of my skepticism here. I don’t claim that Gil could not provide answers to these questions about the goals and constraints of the assigning-truth-conditions project, perhaps even well motivated answers. Nor do I claim that, with these answers at hand, he might not turn out to be right about the best way of assigning truth conditions to moral judgments. What I do claim is that he has not told us enough about the practice of assigning truth conditions to know whether he is right or to judge whether a naïve subject agrees or disagrees with him. One way of summarizing all this is to say that Gil’s “assigning truth conditions” formulation of relativism presupposes (a fragment of something like) a semantic theory. And since he hasn’t told us what that semantic theory is, the view is not sufficiently well specified to assess. This, as it happens, is also what I propose to argue about the second formulation of Gil’s relativism.

II. The “no true framework” account: In the seminar and after, Gil offered a rather different formulation of relativism. Since I don’t have a text to work with, I’ll have to rely on my notes and memory to reconstruct this version. Here’s what I think he said:

(1) Different people (or groups) have different moral frameworks.

[The notion of a “moral framework” is not entirely clear to me. I am assuming that a moral framework consists in or can be captured by a set of moral claims – typically quite general ones like: It is wrong (ceteris paribus?) to kill animals for food. or It is wrong (cp) to kill members of one’s own group. Another point on which I am unclear is what the relationship is between a framework and a person who “has” it. Does the person believe the claims? Or does she stand in some other sort of cognitive(?) relation to them?]

(2) Moral claims can be true or false relative to a framework.

[So, for example, the claim that it is wrong to kill chickens for food would be true relative the framework that I began to sketch in (1).]

(3) There is no true moral framework. Frameworks themselves are neither true nor false.

(4) Some frameworks are better than others. However, this claim itself is relative to a framework. Relative to framework A, framework A may be better than framework B. But relative to framework B, framework B may be better than framework A. And perhaps in some cases framework A may be better than framework B relative to both A and B.

(5) It is a good thing to preserve moral discourse. We should not simply get rid of it as some of Gil’s colleagues have allegedly suggested.

[I assume that this claim, too, is true only relative to a framework – one which Gil presumably endorses.]

Gil devotes considerable effort to defending (1) by citing examples of moral frameworks that differ on just about everything. As I noted above, I am not entirely clear about what a moral framework is. But if it is something that can be largely captured by a set of general moral claims, and if the relation between a person and her framework is something like belief, then I have no quarrel at all with (1). Indeed, I would have thought it was so obvious that it hardly needed to be defended. But I gather that others don’t find it at all obvious. So Gil’s efforts at defending (1) are welcome.

I assume that Gil would unpack (2) along something like the following lines: A moral claim like

(c) It was wrong for Jenny to kill chickens for food.

is entailed by a framework claim (or cluster of framework claims) like

(a) It is wrong (ceteris paribus) to kill animals for food.

along with appropriate factual assumptions. If that is what he means, again I have no objection.

(3) is the claim that I find most puzzling. And the reason, quite simply, is that it invokes semantic terms, ‘true’ and ‘false’, but does not tell us how they are to be understood. To see why this is a problem, suppose that one is a deflationist about truth, adopting a theory along the lines of those proposed by Hartry Field or Paul Horwich. On deflationary accounts, predicates like ‘_ is true’ and ‘_ refers to _’ exist “solely for the sake of a certain logical need.” (Horwich (1990), p. 2), and the notion of truth is merely a device for semantic ascent. So for a deflationist, instances of a schema like:

(T) ‘p’ is true iff p

are analytic. OK, now suppose I have a moral framework that includes:

(a) It is wrong (ceteris paribus) to kill animals for food.

From (a) and (T) it follows that

(a-T) ‘It is wrong (ceteris paribus) to kill animals for food’ is true.

So if I believe (or am committed to or accept or …?) (a) then it looks like I should also believe (or be committed to or accept or …?) (a-T). And if that’s right, then it looks like I can’t possibly be a relativist because I can’t believe (3).

Now, of course, deflationism is not the only game in Semantics Town. There are lots of theories of truth on offer on which the claim that a sentence is true entails a lot more than it does for deflationists. On these theories, one might argue that (a) doesn’t have what it takes to be true. Perhaps this is the part of philosophical space that Gil would move into in defending (3) against the deflationist-inspired argument I’ve sketched. But until he tells us a lot more about the semantic theory he is presupposing, there is no way to judge whether (3) – and relativism – are right.

Ah, yes, bit of clarification on another matter. When Gil claimed he had never felt guilt, I mumbled something about The Lexicon. It emerged after the seminar that many of the philosophy graduate students did not know what The Lexicon is. This is a serious matter. No one should get a PhD in philosophy without spending a few hours studying The Lexicon. It is available on the web at:

http://www.blackwellpublishers.co.uk/lexicon/#LEXICON

I recommend that you start with the H’s.