Peter Geach has a well-known argument that there are no legitimate predicative uses of ‘is good,’ and that this suggests that there is no impersonal, perspective-neutral sort of value of the kind regarded as important by e.g. utilitarians and Moorean consequentialists. The argument is based on the premise that sentences of the form ‘x is good’ do not describe x. Geach claims that all uses of ‘good’ are attributive, as ‘small’ in ‘x is a small elephant,’ rather than predicative, as ‘red’ in ‘x is a red car.’ The use of ‘good’ in ‘x is a good car’ clearly conforms to this pattern, but, according to Geach, the pattern extends even to uses of ‘good’ that appear to involve predication, not attribution. Geach argues that all uses of ‘good’ whose surface grammar is predicative are either logically attributive uses in disguise, or illegitimate. Geach’s main reason for thinking so appears to be that ‘good’ cannot be used to describe the objects to which it is applied in a direct way.
Judith Jarvis Thomson makes the related point that all goodness is goodness “in a way.” Thomson argues that when we are asked, for example, whether it is good that Albert (e.g.) rings the doorbell, our information is insufficient to warrant a judgment. We don’t know what ‘is good’ means–we can’t determine what is being asked. Similarly, although St. Francis was good, and chocolate is good, there is no single scale of evaluation on which chocolate and St. Francis can both be placed in virtue of their respective goodnesses. Although they are both good, they are not good in the same way. St. Francis was morally good, that is, virtuous; chocolate is good tasting, which is an utterly different thing. According to Thomson there is no goodness simpliciter. Something might be good in one of various ways of being an instrument; good for making cheesecake; good at games such as chess or Boggle, or at sports such as baseball; good as a person–possessing some important virtues or lacking some important vices; but all goodness is relativized. The brandy might be good tasting but not a good remedy for the common cold. All goodness is goodness “in a way,” and there is no fact of the matter about whether the brandy is good, for that word does not express any single genuinely perspective-neutral evaluative property. This explains why the question about the goodness of Albert’s ringing the doorbell is impossible to answer: we need to know which particular way of being good is the way in question. We can’t know whether something is good without knowing which particular way of being good is the way in question.
There are, of course, lots of objections to these arguments in the literature. It has been pointed out, for example, that when God created heaven and earth and saw that it was good, or when Martha Stewart appraises a mouse carved out of a mango simmered in a raspberry reduction sauce as a good thing, these seem to be straightforwardly predicative uses of ‘good’ with no clear attributive paraphrase. But the mere existence of these counterexamples does not reveal the mistake the Geach/Thomson view makes, or explain why they might have been tempted to make it.
Here’s my diagnosis. There are many predicates that seem to satisfy the general pattern of use established by ‘good,’ but which are logically predicative, not attributive. The predicate ‘is impressive’ has a logically predicative use, but is not particularly descriptive. Some predicates are more specific than others, and the more general and less specific a predicate is, the less “descriptive” it is. It would be a mistake to conclude that an alleged predicate was merely logically attributive on the basis of its generality or non-specificity. Uses of ‘is good,’ whether logically attributive or predicative, are often not descriptive in themselves. If, for example, I say that this is a good car, it is not clear that I have described it in any way. I might think it’s good because it has good gas mileage, or because it has a powerful engine, or because it is fast, or because it is reliable, or because it has a lot of passenger space, or because it has a high safety rating. Although any of these attributes might serve as the basis for and entail the truth of my evaluative judgment, each of them is descriptive in a way the evaluative judgment as such is not. This is, in part, because there are myriad kinds of goodness (including goodness as an instrument, goodness as a sign, logical goodness, aesthetic goodness, good-tastingness, intrinsic ethical goodness if there is any such thing, etc), and for each kind of goodness, there are myriad possible supervenience bases for that kind and level of goodness. Knowing that something is “good,” even knowing what particular sort of goodness the speaker has in mind, does not carry much in the way of descriptive content. But this is because it is not specific; it is because of its logical properties, and not specifically because ‘good’ is always logically attributive and is never logically predicative. So part of the problem is that Geach has conflated descriptiveness with logical predication.
This sheds light on Thomson’s complaint that we don’t know what sentences such as ‘Albert’s ringing the doorbell was good’ are supposed to assert or express. There are, as Thomson points out, a variety of ways of being good, and (arguably) no single way in which both chocolate and St. Francis are good–since the primary manner in which chocolate is good has to do with flavor and the primary manner in which St. Francis was good has to do with virtue–goodness as a person. (Note also that the manner in which St. Francis was good is distinctively ethical in character, whereas the manner in which chocolate is good is not.) That we have trouble knowing whether Albert’s ringing the doorbell was good, given just that information, is not evidence that all “goodness” is “goodness in a way” and that perspective-neutral intrinsic ethical value is not among the ways; it is evidence that the term ‘good’ can be use to express a variety of intrinsic and relational evaluative properties, and without further context, it is impossible to know which property is the one expressed and why the relevant alleged object of value has been so evaluated. If the goodness in question is “goodness-to-eat,” as chocolate is good, then the answer is a clear “no.” Albert’s ringing the doorbell is not good in this way. If the goodness in question is ”goodness as a means to some further end,” then it depends on what the end is. If the end is to compel the occupant of the house to answer the door, then it is easy to imagine that Albert’s ringing the doorbell was good in this way. If the end was to compel the occupant to continue her nap, then it was probably not good in this way. If the goodness is non-derivative perspective neutral ethical value, then, again, Albert’s ringing the doorbell is probably not good in this way (though it may have been instrumentally good with respect to this kind of value). In any case, Thomson’s examples do not provide evidence against perspective-neutral value.