Corps de l’article

1. “GOOD” AS AN ATTRIBUTIVE ADJECTIVE

Grammarians distinguish between attributive and predicative adjectives. Geach (1956) drew philosophers’ attention to this distinction and argued that it could be extended and applied to our understanding of the logical grammar of evaluative terms. Consider a complex predicate of the form “is an AN” where A is an adjective and N is a noun. For example, “is a red book,” “is a sharp spade,” or “is a good move.” By looking at the way complex predicates behave, we can draw a distinction between two types of adjectives. In some instances, the predication “X is an AN” logically decomposes into the two predications “X is an N” and “X is A.” So, using Geach’s example, “X is a red book” logically decomposes into “X is a book” and “X is red.”[1] In these cases, the truth-value of the complex predication is simply the truth-value of the conjunction of its component predications—i.e., “X is an N” and “X is A.” If it is true that X is red and it is true that X is a book, then it is true that X is a red book. When complex predicates behave in this way, we can say that the adjective is predicative.

Significantly, however, not all adjectives behave like “red.” Some complex predicates are inferentially irregular in the sense that their truth-values are not given by the conjunction of their component predicates. In particular, Geach observed that the predicates “is good” and “is bad” are inferentially irregular. Thus, the predication “X is a good book” does not logically decompose into the predications “X is a book” and “X is good.” If the proposition “X is a good book” is true, we cannot infer that “X is good” simpliciter. For one thing, it is obscure what it could mean to say that “X is good” (if not as an ellipsis for X is a good book).[2] Or, consider the proposition “X is a good poison.” We cannot make two unqualified inferences that “X is a poison” and that “X is good.” Rather we mean something like “X is good as poison” or “X has all the properties poison should have” or “X is well-fitted to doing what poison is for.” In this sense “good” and “bad” are like “big” and “small.” We cannot infer from “Fred is a big flea” that “Fred is big.” Even big fleas are rather small creatures. Rather, we should understand something like “Fred is big for a flea.”

If Geach is right that the underlying logical form of “good” and “bad” suggests that they are attributive adjectives, then the predication “X is good” is incomplete. Geach put the point as follows:

Even when ‘good’ or ‘bad’ stands by itself as a predicate, and is thus grammatically predicative, some substantive has to be understood; there is no such thing as being just good or bad, there is only being a good or bad so-and-so.

Geach 1956, p. 34[3]

Although I am not persuaded that “good” is always used in this way—I discuss an important predicative use in §6—it seems correct that predicative uses of “good” are not normally truth-evaluable as they stand. In order to determine the truth-value of a proposition of the form “X is good,” we need to know what X is. Moreover, depending on what X is, it may be the case that we cannot coherently employ the predicate “is good” at all. Foot (2001, p. 2) describes how she sometimes secured recognition of this point by holding up a small piece of torn paper and asking whether or not it was good. Offering to pass it round so the audience could get a better look would provoke laughter in recognition of the grammatical-cum-logical absurdity of the question. We cannot talk meaningfully about scraps of paper being good in an unqualified sense.

In a similar vein, Thomson offers the following example:

You are standing in front of the array of melons at your grocer’s, feeling helpless. Your grocer notices. He points to one in particular and says, “That one’s good.” […] It would be utterly astonishing if when you asked, “Do you mean that that’s a good melon?” he replied, “Oh dear me no, I haven’t the faintest idea whether it’s a good melon, I meant only that it’s a good thing.

Thomson, 2008, p. 13

Again, this is comical and the comedy comes from the fact that it generally makes no sense to say that a thing is simply good as opposed to a good instance of its kind.

Thomson (1997, 2008) argues that there is a metaphysical explanation for this sense of absurdity. It is often thought that goodness is a property. Surface grammar suggests no difference between the predicate “is good” and predicates such as “is red” and “is spherical.” As it is normally the function of predicates to denote properties, it is natural to think that a sentence such as “X is good” has the function of attributing the property of goodness to X. With this picture in mind, it is then natural to think that the primary task of ethics is to identify whatever property the predicate “is good” denotes, or the universal that particular instances of good things instantiate. Once we have done this, we can ask where this property is instantiated and how this should guide our conduct.

This view is undoubtedly attractive and a great deal of moral theory has presupposed it. Influentially, G. E. Moore held this sort of view of ethics. When he wrote that ethics is “the general enquiry into what is good,” (Moore 1903, p. 2) he meant that ethics is the general enquiry into what the property of goodness is and what has that property. Similarly, when he wrote that “the only possible reason that can justify any action is that by it the greatest possibility of what is good absolutely should be realised” (ibid., p. 60), he meant that we should maximize the number of instances of the property of goodness. For this reason, Thomson (1997, p. 273) calls this sort of view “Moore’s story,” although it is far from being peculiar to Moore.[4] Thomson also suggests that Moore’s commitment to the view that goodness is a property had a baleful influence on subsequent metaethics. For it is very difficult to see what property could be exemplified by all of the things that are good or how we should go about identifying this property. Locating the property of goodness in the natural order is an especially daunting task—all the more so for those who think that recognizing the good necessarily motivates us. Consequently, Thomson (2008, p. 11) suggests that the property of goodness is epistemologically and metaphysically “dark”. This is reminiscent of Mackie’s complaint that moral properties would have to be “queer” for moral predicates to refer.

Moore’s own view, of course, was that goodness is an unanalyzable, non-natural property that is outside the causal order and revealed by a sui generis cognitive faculty of moral intuition. A more common response has been the non-cognitivist line that what is shared by all of those things we call “good” is not some mysterious property of goodness, but the fact that we approve of them in some way. Thus, normativity is built into judgment rather than the world. If, however, Geach was right that “good” and “bad” are attributive adjectives, then it appears that Moore’s story rests on a mistake. The fact that we are wont to say “X is good” disguises the fact that “is good” is an incomplete predicate. This sends us off in fruitless pursuit of the property denoted by the predicate “is good,” such as the universal goodness that all good things instantiate. However, it is no more plausible to think that there is a universal of goodness that all good things instantiate than to think that there is a universal of bigness instantiated by big fleas, big noses, and big attics.[5]

Saying that there is no property of goodness is neither to say that the term “good” is meaningless nor to say that there is no property, or set of properties, denoted by particular attributive uses of the term. Nor, again, is it to say that there is nothing in virtue of which propositions including the term “good” are true. Nor, finally, is it to concede that the term “good” involves us in a hopeless hotchpotch of equivocation. We are still free to investigate how attributive uses of “good” function and under what conditions it is true to say that something is good (just as there are truth conditions for attributing “bigness” to fleas, noses, and attics). But in order to identify truthmakers for propositions with evaluative content, we need to consider the kind of thing that is claimed to be good or bad. I approach this task below, starting with the (relatively) straightforward case of artifacts, before looking at natural functions and then human lives. I offer the qualification in advance, however, that it would be unwise to assume that there is single, legitimate way in which “good” functions in English, let alone a single, legitimate way in which cognate and near cognates of “good” function in other languages.[6] In the end, however, my concern is much less with logical grammar or with natural language than with hints about the metaphysics of value.

2. GOODNESS, GRADING, AND KINDS

One way in which we might understand how the term “good” functions is suggested by Urmson’s (1950, 1968) observation that “good” is used primarily as a term for grading. To say that something is good is normally to say that it measures up more or less well to some standard or set of criteria.[7] For instance, to say that an apple is good is to say, inter alia, that it has a pleasing taste, is pest-free, is relatively blemish-free, and stores well. To the extent that an apple meets these criteria, it is a good apple. To the extent that it falls short of these standards, it is a bad apple. Approximation to a standard also helps us to make comparative judgments. Thus, Urmson pays careful attention to the classificatory scheme of “super,” “extra fancy,” “fancy,” and “domestic” apple grades, detailing the particular characteristics of this ordered set of adjectives (Urmson 1950, p. 151-154). In this way, evaluative judgments appear to contrast with deontological judgments. Because “good” is a scalar adjective, we can say that one thing is better, more excellent, or closer to the ideal than another. By contrast, we do not normally say that something is “righter” or more obligatory than something else.

Because “good,” unlike “right,” has comparative and superlative forms, Evans objects that Urmson’s account cannot deal with a single specimen of a kind. Evans writes:

We could not compare it with anything and, therefore, we could not say it was good in the sense of being good of its kind. There can be no superlative judgments unless there can also be comparative judgments.

Evans, 1962, p. 30

This shows an interesting misunderstanding. When we grade an example of a kind, the comparison is not with other members of the kind—although such comparisons can guide our judgment—but with a standard or set of criteria given by the nature of the kind.[8] Although we can grade apples, students, and many other things in order to rank them, grading is not essentially concerned with intraclass comparisons. It could be that all members of a kind are equally excellent, or far from the ideal. Perhaps Evans’s objection is due to the thought that we would struggle to evaluate a single instance of a kind we had not previously encountered. This is comparable to the difficulty we might face in evaluating an antique artifact with unknown origins and purpose. It would not help, however, to encounter a trove of antique artifacts with minor differences. We do not need a comparison class, but an understanding of what constitutes the relevant standard of evaluation—i.e., what kind of thing we have in front of us.

How, though, are we to determine the relevant standard or standards? The considerations of the previous section suggest that there will be no single answer. Rather, it will depend on the kind of thing we are evaluating. This was part of Geach’s point in saying that a substantive must be understood when we make attributions of goodness. The proposition “X is good” is typically elliptical for the proposition “X is a good K,” where K is the kind to which X belongs.[9] As the properties of good apples are quite different from those of good students, the two kinds are not evaluated against the same standards or criteria. This suggests a two-stage procedure for determining the truth-values of propositions of the form “X is good.” First, we identify the kind K to which X belongs. Second, we measure X against the standards that are appropriate for things of kind K.[10] We do not, as Moore’s story might suggest, independently determine whether X belongs to K and whether X instantiates the property of goodness.

Although these two steps of identifying and evaluating X are conceptually separable, they often rely on one another. In particular, identifying X as belonging to K can depend on identifying certain normative conditions that X satisfies. For instance, part of what it is to identify an object as a spade is to identify it as a digging implement. This is to say that it must have certain properties that fit it for its purpose.[11] “Spade” is a success term in the sense that an object that is not at all suitable for digging could not count as one. Although there are such things as rusty, blunt, and lightweight spades, there is no such thing as a spade made out of room-temperature tofu or mathematical formulae. Of course, the line between a bad spade and a non-spade may be vague or otherwise hard to determine. We can imagine, for example, a good spade that is left to rust and gradually deteriorate. Finally, it will become more rust than spade and so bad that it will no longer be a spade.[12] Sadly the same is true of us. Deteriorating function eventually undermines personhood. There are genuine difficulties here concerning vagueness. However, whatever line we take, we should not conclude that there are no persons or no spades.

3. FUNCTIONAL GOODNESS AND ARTIFACTS

I want to raise three potential concerns in this section. One concern facing kind-dependent standards of evaluation is that many kinds seem to have no plausible standards against which they can be evaluated. For instance, Raz argues:

[R]egarding many kinds of things it does not make sense to ask what is a good or well-functioning thing of that kind. There are no good or well-functioning stones, or pebbles, or streams, or hail, or snow, or mountains, or stars, or black holes, or electrons, or photons, and so on.

Raz 2003, p. 142

A second concern is that a kind-dependent account of goodness does not yet tell us how to determine the particular criteria against which to evaluate an instance of a kind. Thus, even if we can establish that there are standards for a kind, we still need to determine what those standards are. A third concern is that any such standards are not properly normative. Investigating the first concern will help us to make some progress with the second. I shall postpone discussion of the third concern until the final section of the paper.

Let us take mud as an example. Mud can be good for the purposes of, or good from the point of view of, farmers, brick-makers, and bathing hippopotamuses. It is hard, however, to see what it would mean to say that a particular clod is quite simply good.[13] Put another way, there is nothing that it is to be good qua mud, or for mud to have its goodness within itself, even though mud may be good qua something else (building material, cooling lubricant, etc.). So, it appears that some kinds, such as spades, come with evaluative standards built into their identity conditions. As Thomson (2008, p. 21) puts it, some kinds are “goodness-fixing”. Or, emphasizing the other side of the same relationship, Raz (2003, p. 39) talks of “kind-constituting values”. Other kinds, like mud, are neither goodness-fixing nor constituted by values. Ideally, we should like a principled way to distinguish the one from the other.

Philosophers attracted to a neo-Aristotelian naturalism about value may be inclined to approach this question via Aristotle’s function (ergon) argument. The Greek term “ergon” is ambiguous in a potentially suggestive and helpful way, having connotations of both “function” and “characteristic work.” As Aristotle writes, “What a thing is is always determined by its function: a thing really is itself when it can perform its function; an eye, for instance when it can see” (Meteorologica 390a10-2). So, the ideas of a thing’s function, and what it is that makes it the kind of thing that it is, are related to the idea of its ergon. To know, or to understand, what some kinds of things are, is, in part, to know what they are for; and this has significance when we come to evaluate instances of a kind. As I suggested above in the case of a spade, identifying something as an instance of a kind can depend on recognizing that it is at least minimally suited for performing its function. It is for this reason that the two steps of identifying the kind to which a thing belongs, and evaluating it, can be inseparable.

So, for instance, a spade, unlike the mud it disturbs, is for something. Because spades are for digging, a good spade has properties that fit it for that end. Because mud does not have a function, it does not make sense to think of mud as being good or bad except in relation to something else. Mud may be used for something, and may be good for that purpose in the sense of being useful, but it is not for anything. If this explanation of goodness-fixing is along the right lines, and if goodness is related to grading in something like the way I suggested in §2, then we should expect that the goodness of X will depend on the degree to which it fulfils the function of the kind K to which it belongs. This allows us to distinguish certain kinds that have built-in or intrinsic standards of evaluation. It also moves us in the direction of understanding what the standards of evaluation are for those kinds.

4. ARTIFACTS AND DESIGN

It might be objected that the example of a spade allows me to make the case too easily that the goodness of a thing depends on its function. Spades are unusual because they are artifacts, which have been designed for a purpose. Perhaps, then, if a spade has a function, and its function fixes its goodness as an instance of a kind, this is only because of its designer’s intentions. Following this, we can distinguish two types of concern: noting that a spade has the function of digging (i) fails to pick out any non-arbitrary metaphysical truths about goodness-fixing kinds, or (ii) fails to yield any significant insights concerning the evaluation of non-artifacts. I shall pick up the second concern in the next section. According to the first concern, even if we can talk of a spade as having the function of digging, we might just as easily assign it the function of being a doorstop, a weapon, or a work of art. Moreover, even if its designer intended the spade to function as a digging implement, why suppose that the designer’s intentions have any sort of priority over, say, its user’s intentions? Why suppose that either carves axiological reality at its joints? After all, a spade can be used for diverse purposes, including keeping doors open. It may even be good as a doorstop, or make a good doorstop. It might then seem that reflections concerning a spade’s function fail to settle questions concerning its goodness. The goodness of a spade depends entirely on its instrumental value with respect to the interests it serves. A heavy spade with a sharp edge is good because it enables us to do the things we want to do—plant potatoes, bury treasure, and the like—not because it measures up to an evaluative standard given by a goodness-fixing function that is independent of its user’s interests.

Some of this should be conceded. The function of a spade is evidently dependent in certain ways on human agency. It is not a coincidence that the goodness-making features of spades are, by and large, those features that give it instrumental value with respect to our needs and interests. A spade is an artifact designed with the satisfaction of our interests in mind and constructed so that it can be used by creatures like us in an environment like ours. A spade that is too heavy to lift, for example, is ipso facto a bad spade. However, it does not follow that we are unable to evaluate spades independently of either our present concerns or our personal concerns (which I suspect is what the worry largely amounts to). Nor does it follow that identifying an object’s function is entirely arbitrary or wholly dependent on the intentions, or interests, of its designer or user. We can, for example, recognize the excellence, or shoddy workmanship, of a piece of obsolete or alien technology. We can also evaluate artifacts that we do not want—to say “this is a good spade” is not necessarily to say “I want this spade,” even a little.[14] One might have absolutely no interest in planting potatoes or the like, and still correctly believe that a particular spade is a good one. So, even if artifacts are designed in response to human interests or concerns, it does not follow that evaluating the goodness or quality of the artifact must depend on taking into account the evaluator’s current wants or interests. Indeed, it is arguable that we are able to evaluate artifacts in light of their function even if that function serves no possible human interest. Some weapons may fall into this category.

Perhaps, though, it will be suggested that it is the designer’s intentions that matter, and that this explains the weapons case. Given that a weapon has been designed to cause terrible casualties (however contrary to anyone’s actual interests), we can say that the weapon is good to the extent that it is well suited to satisfying its designer’s intentions. Furthermore, it might be argued on epistemological grounds that it would be very difficult to know the function of an artifact if we were ignorant of the designer’s intentions. The quality of an artifact—say, a carving—can be inscrutable when we are unaware of the role it was intended to serve, or the social norms, conventions, and connotations associated with the artifact. But it still does not follow that a designer can arbitrarily assign a function to an artifact. The reason, as I have already mentioned, is that terms that refer to functional kinds, such as “spade,” are success terms. This comes from the fact that the function of an object depends on the criteria of identity for objects of that kind. So, a spade could be entirely unfitted for use as a doorstop or a weapon and still be a spade (perhaps it is the wrong shape or too heavy to wield), but it could not be entirely unfitted for digging and still be a spade. Although a designer may well have a function in mind, that designer cannot decide at will that an artifact has a particular function. An artifact’s function depends on it possessing a set of properties, such as a suitable arrangement of its parts and materials, organized with respect to the achievement of some end.

5. NATURAL GOODNESS

The second type of objection I mentioned above allows that the function of artifacts such as spades provide us with legitimate examples of goodness, but claims that artifacts such as spades are atypical insofar as they are the products of rational agency. Therefore, they can hardly serve as paradigm cases of goodness.[15] One possible response would be to argue that the world is the product of a Creator and therefore shot through with rational purpose or design. A functional account of goodness seems to fit well with a theistic framework and it is probably not a coincidence that a number of prominent neo-Aristotelians are theists.[16] Given certain further assumptions concerning the attributes of God, this view would also seem to have just the right shape for grounding the to-be-pursued quality of goodness. Although this line of thought is attractive in the present context, it depends on theistic commitments that many philosophers, including myself, do not share. I propose, therefore, to set it aside and consider an alternative possibility.

It is often observed that even if the world is not the product of rational design, we are still able to understand much of the biological world in design terms. In particular, evolutionary theory provides an explanation for a large number of apparently teleological features in nature. While natural selection is unintentioned, it tends to produce organisms with parts and processes that are structured towards fitness-enhancing ends. The case looks particularly strong for parts or organs that contribute to complex biological systems. For instance, it seems natural to say that hearts are for pumping blood, that eyes are for seeing, that roots are for absorbing nutrients, etc. A plausible explanation for these biological functions can be given in terms of the evolutionary pressures bearing on an organism’s ancestors.[17] The reason that the function of the human heart is to pump blood and not, say, to make a thumping sound is that only the former made a contribution to the fitness of our ancestors.

These examples seem to fit well with a functional account of goodness. Given that the function of the eye is to see, it is very plausible that a good eye sees well. Perhaps more interesting, however, is the thought that evolution works at the level of behavioural traits. In particular, certain behavioural traits are adaptive for a species in the context of a distinctive way of life or “life-form” (Thompson 2008). For instance, climbing well is adaptive for squirrel monkeys because they primarily live on fruit that grows on trees. Or, again, signaling well is adaptive for meerkats because they live in social groups and are individually vulnerable to predators.

It looks attractive, then, to say that at least some traits can be evaluated in terms of their contribution to a species’ way of life. But, of course, we are also a species of animal with a distinctive way of life. Perhaps most significantly, we are, as MacIntyre (1999) calls us, “dependent rational animals.” We are social and linguistic animals, vulnerable to each other and to our environment, and capable of representing reasons to ourselves and to others. So, as with other animals, we might say that a trait is good when it is conducive to success with respect to our distinctive way of life. In the case of humans, stable traits of character, feeling, and thought are traditionally called virtues when they are good and vices when they are bad. As with other animals’ traits, these valuable traits depend on both nature and nurture. We become virtuous, if at all, through a combination of inheritance, instruction, luck, practice, and experience.

All of the above is quite programmatic. Nevertheless, considerations such as these have encouraged a number of philosophers to develop neo-Aristotelian forms of ethical naturalism.[18] They mean “naturalism” not in the sense of deferring ethics to the natural sciences, but in the sense of attempting to ground ethics in human nature. Perhaps Foot (2001) did most towards developing an account of natural goodness along these lines. The idea of natural goodness attractively connects the evaluation of human lives with our natural history and distinctive way of life, hence Geach’s well-known comment that “men need virtues as bees need stings” (Geach 1977, p. 17). Presumably, we could have evolved quite differently, but we are as we are with our special strengths and vulnerabilities, and these facts matter when we come to reflect upon what counts as living well for beings like us.

It is important to recognize that although natural selection has partly made us what we are (and more fully made other animals what they are), natural goodness is not a concept drawn directly from evolutionary biology. What matters is the place of particular traits in the life of a species as they have become. What Thompson (2008, p. 20) calls “natural-historical judgments” concerns the present nature of a species. These have an evolutionary explanation, but the norms that have resulted from our natural history are not justified by evolutionary considerations.[19] Advocates of an ethics of natural goodness can avoid the charge that they are trafficking in explanatory reasons under the guise of justificatory reasons or crudely reducing value to reproductive fitness.[20] What it is to be a good human depends on what it is to be a human, but not directly on the causal explanation of the nature of our species.[21]

6. REASON AND VALUE

Neo-Aristotelian ethical naturalism builds on Geach’s insight that “good” is (typically) attributive. If we are interested in what it is to live well, we get off on the wrong foot by asking with Moore “What is good?” Instead, we should ask “What is it to be a good human being?” Moreover, our approach to answering this question shouldn’t be radically disjunct from the way in which we evaluate other living organisms. We should study the parts, processes, and traits that are fitted towards the ends appropriate for things of the relevant kind. To my mind, this is a promising approach to practical philosophy. Notwithstanding a number of prominent advocates, however, it remains a minority view. In this final section, I want to look at one possible source of resistance, which I shall relate back to the distinction between predicative and attributive adjectives. It is the concern that I briefly raised in §3—namely, that attributive goodness is not genuinely normative.

While advocates of neo-Aristotelian ethical naturalism emphasize the continuity between humans and other living organisms, it might seem that our differences matter more for the purposes of evaluative discourse and, especially, morality. In particular, it is distinctive of human rationality that we can represent putative sources of value to ourselves and submit them to critical scrutiny. As Foot observes, “while animals go for the good (thing) that they see, human beings go for what they see as good” (Foot 2001, p. 56). Irrespective of whether humans are unique in the way that Foot suggests, rational beings like us face a special sort of problem. Even if we agree that there are norms of natural goodness, it seems, on the face of it, to be an open question as to whether we should reflectively endorse these norms.[22] As Slote (2003) observes, it is hard to deny that we are a warlike species or that aggression between subgroups of humans has been a pervasive feature of human history. Nevertheless, most of us would reject the conclusion that warlike behaviour is good for human beings.[23] It might be replied that aggression and war are (normally) harmful for creatures like us. We do not flourish under such conditions. Although this is right, Slote objects that it is unclear how this response can be directly grounded in natural goodness. If warfare is part of our distinctive way of life, it appears that the response employs an independent criterion in order to evaluate norms of natural goodness.

We should not rest content, then, with the thought that some trait is good in relation to our distinctive way of life. Rather, as rational beings we have the ability to question whether we should reflectively endorse the pursuit of particular goods. Perhaps the goal of leading good human lives understood along neo-Aristotelian lines should be rejected. Arguably, only values that can survive this process of rational scrutiny are candidates for being truly normative—for being standards that make claims on the thought and action of rational animals like us.

Here is one way to make sense of this concern. I suggested in §1 that “good” is not always used in its attributive sense. Geach (1956, p. 36) dismisses predicative uses of “good” as “a peculiarly philosophical use of words” . However, this looks like ad hoc monster-barring. Expressions like “friendship is good,” “health is good,” “pleasure is good,” and “God saw that the light was good” are common enough. What, then, is the point of such expressions? One possibility is that people intend to express their ontological commitments with respect to the properties of friendship, health, pleasure, or light. But this strikes me as implausible. In fact, it is unclear that everyday discourse commits us to any particular view of the metaphysics of evaluative properties (see further Johnston 1993). Most of us are hard pressed to explain the metaphysical pictures or assumptions that lie behind our moral judgments.

I propose instead that the predicative form is normally used to express another idea. I have in mind the sort of usage Wiggins describes as follows:

What the philosophical lexicographer recapitulates under “good”, we might say, is the history of our constant interrogation of the life that we lead and the place where we lead it, our constant interrogation of the things that concern us or might concern us or ought to concern us.

Wiggins 2009, p. 198

Pace Geach, we sometimes use the word “good” in a predicative sense to mean the sort of thing that is intrinsically worthy of pursuit, or the sort of thing that finally stands up to interrogation. We do not mean that pleasure, friendship, and health are good of their kind, or that they are instrumentally valuable, but that they are good things and that it is, therefore, rational to pursue them.

It seems, then, that we can understand the kinds of worries expressed about the value of a life of natural goodness as reflecting the following question: Is a life that is good in the attributive sense a life that is worthy of pursuit? Thus, when we consider whether to reflectively endorse a way of life, our own or someone else’s, we can ask whether it is good in the no-holds-barred predicative sense. Such questions lead quickly on to general enquiries into the human condition and whether life is good or meaningful. I want to conclude with a few observations about the conditions under which they arise and how the relationship between attributive and predicative evaluations helps us to understand them.

In this context, consider Tolstoy’s account of his doubts concerning the meaning or value of his own life. In My Confession, he recalls that at the height of his literary success, financially secure and with a loving wife and family, the question kept pressing upon him “Why? Well, and then?” (Tolstoy 1987, p. 10) whenever he considered any prospective good. Repeatedly hitting upon this question, he reported:

I felt that what I was standing on had given way, that I had no foundation to stand on, that that which I had lived by no longer existed, and that I had nothing to live by.

Tolstoy 1987, p. 11

There was apparently little doubt in Tolstoy’s mind that he was leading a good human life in the sense of achieving the kinds of goods that are typically available to us. Yet, he was unsure whether or not to reflectively endorse a life characterized by these goods. These are uncertainties that press themselves upon many reflective people at some point or another.

I suggest that we can understand Tolstoy’s predicament as follows: the attributive goodness of his life left its predicative goodness an open question. Let’s call the question of the predicative goodness of an attributively good life “Tolstoy’s Question.” Tolstoy partly assuaged his concerns, and alleviated his depression, by finding value in God’s design. If we could be confident that our natures had been shaped by a benevolent and powerful Creator, then we could also be confident that our natures are basically good. If so, to do well in the attributive sense would be to do well in the predicative sense too. If there is no such Creator, however, we may have less reason to be sanguine. Perhaps our nature is more like that of an OncoMouse or a broiler chicken in the sense that our parts and processes are organized with respect to ends that we would not reflectively endorse.

How might we provide a secular answer to Tolstoy’s Question? One possibility is to look for an answer in terms of our interests or what we think of as being good for us. Thus, we might express the hope that evolutionary forces have molded us such that, unlike the constitution of a broiler chicken, our constitution is well suited to meeting our interests. Depending on one’s theory of interests, it is a short step to ask whether attributive goodness satisfies our desires or preferences. This pattern of dialectic is both ancient and common. In Plato’s Republic, Socrates attempts to persuade Thrasymachus that justice is part of flourishing. Philosophers in the social-contract tradition appeal to the idea of “enlightened self-interest.” And the view that (moral) goodness benefits its possessor has been defended by some contemporary virtue ethicists.[24] Perhaps, then, an attributively good life is worthy of pursuit because it satisfies our interests. Let’s call this the Socratic Answer.

The question of whether an attributively good life is good for us in the sense of satisfying our interests is certainly an important one. Although I am doubtful that we should accept the Socratic Answer, I cannot pursue that question here. Rather, I want to point out that the Socratic Answer leaves important elements of Tolstoy’s Question unanswered. In particular, its egocentric starting point fails to take account of the variety of perspectives from which an attributively good life could be evaluated. It is not that any defence of the Socratic Answer must be substantively egocentric in the sense that it only considers self-interest. As most of us have a mixture of self- and other-regarding interests, an egocentric question can receive a partly other-regarding reply. The problem is that the Socratic Answer is formally egocentric. Thus, we might be persuaded that we are adapted to satisfy our own interests, but remain concerned that satisfying those interests is not a good thing from some perspective that is not our own. For the fact that we have the interests that we do is no less a fact about our nature than any other fact. The normative significance of our interests is, therefore, just as much open to interrogation as anything else.

To make the point dramatically, perhaps our nature is like that of the smallpox or Zika virus. From the point of view of the universe, perhaps we should wish for our own demise. Some environmental ethicists would remind us that such thoughts are not entirely silly as we stand at the threshold of the Anthropocene. However, the point is not that we should (or should not) adopt this view of ourselves. Instead, the point is that the matter is not settled by reflecting on our interests or preferences. As inveterate interrogators of what concerns us, we are led to question each aspect of our nature and this includes our interests and preferences. We might express such thoughts by asking about the relationship between attributive goodness and predicative goodness, or what I have called Tolstoy’s Question. Neither an account grounded in attributive goodness nor the Socratic Answer fully addresses such doubts.