Friday, January 11, 2019

Metaethics 0005: Mark Schroeder - the Frege-Geach Problem

What follows is, in my view, one of the most powerful arguments against the claim that moral statements are not capable of being true or false.

The argument is that they certainly act like statements that are capable of being true or false.

Peter Geach (and, independently and less famously, John Searle) presented this argument in the 1960s against non-cognitivist moral theories like that of R.M. Hare. As discussed in Metaethics 003: R.M. Hare's Prescriptivism, Hare held that moral statements are commands much like the instructions in a cookbook or an order to "close the door." These types of statements do not have a truth value, and (according to Hare) neither do moral statements. Emotivism and expressivism similarly assert that moral statements are not truth-bearing. They are expressions of emotion (e.g., slamming one's fist on the table) or of other states (expressing a preference), none of which have a truth value.

Against the idea that moral claims are not statements, Geach - drawing from Gottlob Frege's distinction between what a statement means and its "assertoric force" (or force as an assertion) - provided an argument that they certainly act like regular statements. Indeed, they do not act like the types of claims (expressions) that the expressivists said they are.

Note that, in the false dichotomy that currently rules moral philosophy, if moral statements are not expressions then they report facts, and they cannot report facts unless intrinsic values exist. Intrinsic values certainly must exist. Consequently, we need to go back to these expressivist theories and figure out a way to fix the Frege-Geach Problem. Indeed, if I thought that the only alternative to expressivism was intrinsic value theory, I would be frantically searching for a way to fix expressivism as well. Fortunately, I do not have to do this.

Let's look at this problem in a bit more detail. The objection notes:

[T]here is no linguistic evidence whatsoever that the meaning of moral terms works differently than that of ordinary descriptive terms. On the contrary, everything that you can do syntactically with a descriptive predicate like ‘green’, you can do with a moral predicate like ‘wrong’, and when you do those things, they have the same semantic effects.

What the heck does this mean?

One example: you can turn a moral statement into a question. We can take the statement "lying is wrong" and turn it into a question: "Is lying wrong?" We cannot do this so easily with a command. "Do not lie" does not translate well into "Is do not lie?" We could, perhaps, turn the statement, "This is good" into "Is this good?" However, asking "Is this good?" is not the same as asking, "Do I commend this"? Since we can turn evaluative statements such as "lying is wrong" into a question but not imperatives like "do not lie", this suggests that moral statements actually are not imperatives. If they were, then they would behave the same way.

As an aside, let me quickly say that we can take the statement, "Lying is an act type that people generally have many and strong reasons to promote an aversion to using praise and condemnation" and turn it into the question, "Is lying an act type that people generally have many and strong reasons to promote an aversion to using praise and condemnation?" So, desirism is immune to the Frege-Geech problem. It only applies to desirism's non-cognitivist rivals.

Anyway, back to the argument:

The statement "This is green" has truth conditions. It can be true, or it can be false. If it is true, then "this is not green" is false. If it is false, then "this is not green" is true.

The statement "this is good" according to expressivism does not have any truth conditions. It is neither true nor false. However, it certainly looks as if it is the case that "this is good" behaves the same way as "this is green". That is to say, we certainly treat "this is good" as if it is capable of being true or false. If it is true then "this is not good' must be false. And if it is false, then "this is not good" can be true. How can it be the case that "this is good" works the same way as "this is green" if the latter can actually be true or false and the former (according to the theory being challenged) has no truth value?

As Schroeder reports, there is no argument that the non-cognitivist cannot answer the challenge of providing a semantic that handles these types of cases in the way a descriptive statement handles them.

This is true not only for questions, negations, and conditionals, but also for quantifiers, modals, tense, attitude-verbs, generics, adverbs of quantification, intensifying adverbs like ‘very’, and so on. Noncognitivists believe that moral terms have a different kind of semantics than ordinary descriptive terms, but somehow every complex-sentence forming construction manages to do exactly the same sort of things with them that it does with ordinary descriptive terms.

The challenge is that it would be a complex and difficult job - towards which I would argue, "Why go through all that effort if you do not have to?" Following the principle that the easiest course of action is the best, we should go for the easiest option of taking moral statements to be truth-bearing statements. Once again, those who do not think that this is the easiest option tend to think that the only other option is an unreasonable form form of realism.

The Higher-Order Solution

Simon Blackburn tried to solve this problem using higher-order disapprovals. He sought to make sense of the following argument form:

(1) Stealing is wrong.
(2) If stealing is wrong, then murder is wrong.
(3) Therefore, murder is wrong.

Remember, validity has nothing to do with the truth of the premises. In logic, this argument forms the pattern that logicians call "modus ponens" and is recognized as a valid argument type. Generalized, it takes the form, "P", "P -> Q", "Therefore, Q". The problem for the expressivists as discussed above is that "if stealing is wrong" in (2) does not express an anti-attitude towards stealing. So, it does not have the same meaning as "stealing is wrong" in (1). Blackburn wants to provide a way of expressing (2) that preserves this con-attitude.

Blackburn's answer is to claim that (2) expresses a con-attitude towards condemning stealing but not murder.

However, Mark van Roojen pointed out that this is not the only way to express a con-attitude towards both stealing and murder. He replaced (2) with

(2') It is wrong to both disapprove of stealing and not disapprove of murder.

However, (1), when combined with (2'), does not in fact yield (3). One cannot infer that murder is wrong in this case. The problem is that Blackburn did not provide the type of incoherence that we find in believing that P and believing that not-P. His incoherence is the type found in believing that P and believing that one does not believe that P. All higher-order solutions will have this problem and can be rejected for the same reasons.


I am going to end this here. The problem that the non-cognitivist are trying to handle does not exist. Since moral statements are descriptive statements, we do not need to worry about why moral statements act like descriptive statements.

No comments: