Judith Jarvis Thomson’s Normativity
- First Online:
- Cite this article as:
- Harman, G. Philos Stud (2011) 154: 435. doi:10.1007/s11098-011-9737-y
- 155 Views
Thomson’s (2008) Normativity is a careful, rigorous account of the meanings of basic normative terms like good, virtue, correct, ought, should, and must. Along the way Thomson refutes almost everything other philosophers have said about these topics. It is a very important book. 1
Since I am going first in this symposium, I am mainly going to summarize this excellent book. At the end, I will try to indicate briefly why I think it refutes the sort of theory I and others have previously favored.
Thomson begins by discussing evaluations using the word good. She notes as many others have that good is always used as an attributive adjective. Something may be a good K or good in a certain respect, but nothing is good period.
Thomson goes on to argue this means that emotivism, expressivism, prescriptivism and related accounts of the meaning of good cannot be generally correct. Nor does it make sense to suppose that there is no objective test for whether something is a good K or good for such and such a purpose.
She argues persuasively that there is such a property as being a good K if and only if K is a goodness fixing kind. So there is no such property as being a good pebble, good act, a good fact, a good state of affairs, a good possible world, and so on, unless what is meant is, for example, a good pebble to use as a paperweight, a morally good act, a state of affairs that is good for Jones, a possible world that is a good example in a certain discussion, and so on.
The word good means the same thing in moral and nonmoral contexts. It is not ambiguous. There is a difference between an act’s being morally good and it’s being good for the agent, but that difference is due to the fact that the phrases morally good and good for the agent have different meanings, not because good has different meanings.
You can believe that something is a good K without desiring it in any way. For example, you can believe that something is a good bird cage without in any way wanting to have it.
It is true that to say something is a good K is to commend it or praise it qua K. But that is not to desire it. And you can praise and “dispraise” someone at the same time, for example by saying that A is a good liar, or by saying in a recommendation for a philosophy job that B has good handwriting.
There is another way in which good is an attributive adjective—a way it shares with light in weight. Whether something counts as light in weight is relative to a comparison class. Similarly, whether something is a good typewriter is relative to a comparison class. Just as a typewriter may be light in weight for a typewriter made in 1900 but not for one made in 1980, it may be may be a good typewriter for a typewriter made in 1900 but not for one made in 1980.
This is not to say that something may be good, period, in relation to a comparison class. It can only be good in a certain way in relation to a comparison class.
Analogous remarks hold for better. One thing cannot be better, period, than another; it can only be better in this or that respect.
One implication is that consequentialism cannot be formulated as the view that one ought to act in such a way that the result of so acting is better than the result of any other way one could act. A result cannot be better, period, but only better in one or another way. (A possible world in which the agent does A cannot be a better possible world than one in which the agent does B, because “the kind possible world is not a goodness fixing kind” [p. 62].)
Thomson points out that it is counter-intuitive to say instead that you ought to act to bring about the most happy world. Suppose that you can save the lives of five strangers at the cost of your own. Suppose that people “at large” will be happier if you make this sacrifice. Does it follow that you ought to save their lives at the cost of your own life? “It is hard to see how anyone could sensibly claim that this does follow.” (p. 64)
K is a goodness-fixing kind, and
a K is as good a K as a K can be only if it has F, and
it is possible for there to be a K that lacks F, and
it is not nomologically impossible for there to be a K that has F.
One consequence of this is that, contrary to what is sometimes assumed, a property can be both an evaluative property and a descriptive property.
a K is morally as good a K as a K can be only if it has F, and
it is possible for there to be a K that lacks F, and
it is not nomologically impossible for there to be a K that has F.
Thomson offers similar definitions of intellectual virtues, physical virtues, social virtues, and aesthetic virtues.
Thomson turns next from goodness to correctness. A correctness fixing kind is a kind K that fixes the standards that a K has to meet in order to be correct quaK, as for example the kind map of England fixes the standards that have to be met for something to be a correct map of England, and the kind spelling of chiaroscuro sets the standard that a spelling of that word has to meet in order to be correct qua spelling of that word.
Correctness properties, like the property of being a correct map of England, are favorable evaluative properties. Being a correct map of England is a virtue in a map of England.
Thomson then makes the very important point that there are two ways in which assertions can be correct. The word assertion is ambiguous. It can be used to refer to the proposition asserted or to the act of asserting that proposition. The proposition asserted may be correct even if the asserting of it is incorrect because of mispronouncing a word or some minor ungrammaticality, as in “Jack gave a book to Sam and I.” Thomson calls the correctness or incorrectness of the act of asserting internal correctness or incorrectness. She calls the correctness or incorrectness of the proposition asserted external correctness.
The distinction is important. Reflecting on it shows, for example, that Timothy Williamson is mistaken in defending what he calls the knowledge rule, according to which one must: assert that p only if one knows that p. Williamson says this rule “expresses the kind of obligation characteristic of constituent rules,” such as the rules of a game like chess.
However, as Thomson observes, such rules “impose no obligation of any kind … They don’t tell you what you must or even what you ought to do.” They only tell you what counts as a legal or correct move in the game. They don’t tell you that you must or are obligated to make such a move. So Williamson’s argument could at best support the weaker knowledge rule that an act of asserting that p is a correct asserting that p only if the asserter knows that p. But even that is wrong.
To show this, Thomson considers Williamson’s example in which a speaker asserts something that turns out to be true although the speaker did not know that it was true. Williamson says you might have a complaint against the speaker. The speaker should not have asserted that p if the speaker did not know it was true. Thomson agrees but points out that that is not enough to show that the asserting of p was incorrect. It needn’t have been internally incorrect; everything might have been pronounced correctly with correct grammar. And the proposition asserted is certainly correct. So even the weaker knowledge rule must be rejected.
Thomson goes on to observe that being an internally correct describing, reporting, explaining, or answering a question is a virtue in describing, reporting, explaining, or answering, so these are virtue/kind properties. So is being an externally correct describing, reporting, explaining, or answering a question. It is a virtue in describing something that one’s description be true, and similarly for reporting, explaining, and answering a question.
Now recall that F is a virtue in a K only if a K is as good a K as it can be only if has F. Since one might assert a false proposition as well as it is possible to assert that proposition, it follows that it is not a virtue in asserting something that the proposition asserted be true. This is so, even though it is a virtue in answering a question that the proposition one asserts in answering that question be true.
Next Thomson observes that, although one can assert a proposition more or less well, one cannot in any similar way believe a proposition more or less well. So, although assertings can be internally correct or incorrect, believings cannot be internally correct or incorrect. And just as it is not a virtue in an instance of asserting that it is externally correct, it is also not a virtue in an instance of believing that it is externally correct. A believing is not better as a believing if it turns out to be correct. So, being a correct belief is not a favorable evaluative property.
In this connection Thomson also argues persuasively that slogans like “Truth is normative,” “meaning is normative,” and “belief is normative” should be rejected as confused.
Turning to reasons, Thomson proposes that an objective reason for something is always a fact that is evidence for a relevant belief (or makes that belief more probable). For example, something is an objective reason for S to do D if and only if it is a fact that is evidence for the belief that S should do D.
Furthermore, P is a subjective reason for S to do D if and only if S believes that P is a fact and is a reason to do D.
Suppose it is a fact that someone will give S a great deal of money to believe that P? Is that a practical reason for S to believe that P? Thomson says it is not a reason for S to believe that P because it is not evidence that P is true. And, although there can be practical reasons to do something, according to Thomson believing that P isn’t something one can do in the relevant sense. 2
Thomson proposes that for P to be S’s reason for ϕ-ing, where ϕ-ing is believing a certain thing or being in some other mental state, is for it to be the case that S believes that P is a fact that is an objective reason for S to ϕ, where S’s having that belief is the reason why S ϕs. 3
There can be a reason for S to do D even if doing D would not promote anything that S wants. There may be a moral reason for S to save V’s life, which S can easily do, even if S does not think there is such a moral reason to do so. (In this case it is a failing in S not to recognize that there is this reason.)
Similarly, S can want something that can be obtained only if S does D although the fact that S wants this is not a reason to think that S ought to D and so is not a reason for S to do D.
Perhaps the most important implication of Thomson’s discussion of reasons is that the “turn to reasons for action” in moral philosophy starting in the 1950s was a mistake. We cannot explain what people should, ought to, or must do by appealing to what there are reasons for them to do, because we must explain such reasons by appealing to what they should, ought to, or must do.
Next Thomson argues that there cannot be two speakers, Smith and Jones, such that without either of them misusing words, Smith truly says “A ought to D” and Jones truly says “A ought to D′, ” although it is impossible that A could do both D and D′. This rules out certain sorts of moral relativism 4 and certain moral dilemmas. If A promised one person to D and another to D′ but can’t do both and there is no way in which one promise takes precedence over the other, then A should flip a coin to decide which to do. It is not true both that A ought to keep the one promise and also that A ought to keep the other promise.
Thomson also argues against a Kantian distinction between a hypothetical and a categorical “ought” where the hypothetical “ought” expresses what one ought to do in order to attain one or another goal and the categorical “ought” expresses what one ought to do all things considered. The categorical “ought” is just the ordinary “ought” and there is no hypothetical “ought” of the relevant sort, since it would be misusing words or speaking falsely to say, “Well, anyway, if she intends to kill him, then she ought to poison his coffee.” It’s just not true that she ought to poison him.
Does Thomson also reject any distinction between a prima facie or pro tanto “ought” and an all things considered or all out or all in “ought”? Some such distinction might allow someone to say truly, “Given that A promised Morton to D, A ought to D and, given that A promised Norbert to D′, A ought to D′, but A can’t do both, so what A should do all things considered is to flip a coin.” 5
In a final chapter, Thomson discusses relations between evaluatives (e.g., “X is a good K”) and directives (“X ought to F”). A good clock accurately represents the time; a clock ought to represent the time accurately. A clock that does not do so, is defective and Thomson suggests that directive “oughts” are to be explained in terms of what is defective.
For things other than human beings, she proposes this:
If A is not a human being, then for it to be the case that A ought to V is for it to be the case that there is a directive kind K such that (α) A is a K, and (β) if a K doesn’t V, then it is a defective K, and (γ) there is no directive kind K+ such that K is a sub-kind of K+, and such that if a K+ does V, then it is a defective K+.
For physical and mental characteristics of human beings, she proposes this:
If A is a human being, and V-ing is a physical or mental norm for the species, then for it to be the case that A ought to B is for it to be the case that a human being is a physically or mentally defective human being if he or she does not V.
For cases in which V-ing is acting in a certain way, she proposes this:
If A is a human being and to V is to act in a certain way, then for it to be the case that A ought to V is for it to be the case that, if A knows at the time what will probably happen if he or she Vs and what will probably happen if he or she does not V, then A is a defective human being if he or she does not V.
For cases in which to V is to believe something P, she proposes this:
If A is a human being, then for it to be the case that A ought to believe P is for it to be the case that (α) P is true, and (β) there is something that A ought to do such that the fact that P is true is among the facts that make it the case that A ought to do it.
For it to be the case that A ought to trust B is for it to be the case that A ought to believe that B is trustworthy, and similarly for other mental states.
I want to end by sketching how this book appears to refute what Thomson calls “the turn to reasons,” a turn that I (Harman 1977, 1986 and my contribution to Harman and Thomson 1996) and many others have taken in the past. Thomson does this by providing a conceptually much simpler meta-ethics in which what people should, ought to, or must do is not explained by appealing to what there are reasons for them to do because such reasons are themselves explained by appealing to what people should, ought to, or must do.
Very roughly speaking, this conceptual simplicity arises from her being able ultimately to explain directives (ought) in terms of evaluatives (good)—not as in consequentialism, of course, but as in a virtue theory theory that appeals to a notion of defect. As far as I can tell, the resulting theory appears at least verbally to account for almost everything in my view that survives her criticism. This sort of conceptual simplicity is highly relevant to theory choice in ethics.
It might be objected that Thomson’s approach fails because reasons and reasoning really are more basic notions than her approach allows. But this is not a good objection. Something may be basic or primitive in one or another way of developing a theory, but there is no objective notion of what is basic prior to theory.
According to Thomson, it is a moral defect not to care about justice. It might be objected that someone might still ask, “Why should I care about justice?” We need to give that person a reason to care and it doesn’t seem to be enough to say that lacking justice is a moral defect. However, the same problem arises in any view. Why should I care about what’s wrong? Why should I care about what I ought to do? Why should I care about what I have most reasons to do? In fact, why should I care about what I should care about? No view seems immune from this sort of worry.
This book discusses meta-ethics. She also plans a companion work of normative theory.
It has seemed to me that one sometimes has to decide whether to believe something and that is something one can do (Harman 1999, p. 95). But perhaps this is a misleading way to describe such a case.
I expect that this can be reformulated so as not to imply that something can be S’s reasons for ϕ-ing state only if S has the concepts of a fact and an objective reason.