By Jeff Steele

Winner of the Outstanding Graduate Paper Award at the 58th Annual Meeting of the Florida Philosophical Association

Jeff Steele, University of South Florida


Moral and evaluative properties – like goodness – are odd sorts of things, especially on a naturalistic worldview. What do we make of such a moral realm? Given the truth of realism, we have two options: We can either posit a sui generis realm of moral and evaluative properties or we can identify moral and evaluative properties with natural or descriptive properties; the former view is called “ethical non-naturalism”; the latter view, “ethical naturalism.”1 After a brief account of ethical non- naturalism, I explore the standard arguments in favor of it with respect to goodness. Then, I explore Philippa Foot’s criticism of the open-question argument and her subsequent neo-Aristotelian theory of goodness. I argue that Foot’s account is vulnerable to a revised version of the open-question argument. Finally, I suggest two ways that Foot can escape the revised open-question argument: either via a classical naturalist account inspired by Thomas Aquinas that identifies goodness with the fulfillment of a thing’s function, or by characterizing her account as a version of non-reductive naturalism akin to non-reductive physicalism in the philosophy of mind. In either case, I argue that Foot’s neo-Aristotelian naturalism is preferable to ethical non-naturalism concerning the property of goodness. If it reduces and identifies goodness with some set of descriptive facts, it does so without succumbing to Moore’s open-question argument; if it resists the reduction of classical naturalism, it does so without having to postulate a sui generis epiphenomenal category of values.

A Non-Naturalist Theory of the Good

A non-naturalist theory of the good is a species of ethical non-naturalism; specifically, non- naturalism about the nature of goodness. Ethical non-naturalism is a realist theory about moral and evaluative properties: they exist independently of our conception of them. Since ethical non- naturalism is a realist theory pertaining to goodness, it is instructive to look at realism and its rivals generally with respect to the moral domain. This will allow us to situate properly ethical non- naturalism generally and a non-naturalist theory of the good, specifically.

With respect to any domain, we can divide views about the reality of the domain into three categories: nihilism, constructivism, and realism.2 A nihilist denies the reality of the domain itself. Consequently, a moral nihilist denies the reality of moral and evaluative facts and properties. A constructivist grants the reality of the domain’s existence, but claims that the reason for its existence depends to some degree upon some “constructive function out of which reality is created.”3 Moral constructivists construe the output of the function as the moral realm, but differ insofar as they offer diverse inputs upon which the moral reality is established. Subjectivists claim that subjective tastes of the individual that provide the proper input. For Kantians, it will be the rational will; for cultural relativists, the input will be various social agreements. Whatever the constructivist view offered, they agree that the moral domain exists and that its reality is dependent upon – even created by – some preferred standpoint.4

In contrast to nihilists, realists accept the reality of the domain; in contrast to constructivists, realists deny that the reality of the domain is dependent upon some preferred standpoint. On a realist account of some domain, the domain exists independently of our conception of it. A realist theory about a domain claims that the domain is a part of the fabric of the universe – something discovered, not created. Thus, a moral realist believes that moral facts and values exist apart from any human conception of them. In other words, there exists a stance-independent moral realm.5

If moral realism is true, then it faces the following problem: moral and evaluative properties are odd sorts of things, especially on a naturalistic worldview. They do not seem to be discoverable in the same way that scientific or empirical facts can be, but they pull upon us to do and act in ways that empirical facts do not. What do we make of such a moral realm? Given the truth of realism,6 we have two options: We can either posit a sui generis realm of moral and evaluative properties or we can identify moral and evaluative properties with natural or descriptive properties;7 the former view is called “ethical non-naturalism”; the latter view, “ethical naturalism.”8

A naturalist theory of the good will thus identify the property of goodness with some underlying natural, physical, or descriptive property or set of properties. In contrast, a non-naturalist theory of the good will argue that the property of goodness is a sui generis evaluative property that cannot be reduced to natural or descriptive facts, nor can it be described in terms of some underlying physical or natural properties. G.E. Moore, for example, claims that goodness is a simple property that is not analyzable in terms of non-moral properties.9 Let’s consider some reasons for this view.

The Case for a Non-Naturalist Theory of the Good

According to Moore, when one identifies the property of goodness with some natural property, as ethical naturalists do, one commits the “naturalistic fallacy.”10 In order to demonstrate this, Moore provides the “open question” argument. Suppose we take a classical ethical naturalist position, such as hedonism, that describes goodness in terms of pleasure. If good is identified with pleasure, then according to Moore, the two following claims will be identical:

(A) I know that X is good, but is it good?
(B) I know that X is pleasurable, but is it good?11

On Moore’s account (A) and (B) are not equivalent because (B) is an open question, whereas (A) is not. If I know that X is good, then I know that it is good. If I know that X is pleasurable, it is still an open question as to whether X is in fact good. Therefore, the two statements cannot be identical and goodness is not the same as “pleasurable.” This argument can be reformulated for any naturalistic theory that tries to identify goodness with some natural or descriptive property. Moore concludes from this that (1) goodness is not synonymous with any natural property and thus (2) cannot be identified with any natural property.12

Landau also offers arguments in favor of ethical non-naturalism, which he describes as the view that moral properties are not identical to scientific properties and are therefore sui generis. In order to defend this view, Landau believes that the non-naturalist must answer two questions. First, why should we posit moral facts and properties over and above scientific ones? Second, if non-naturalism is true, what relations obtain between the two realms (i.e., the moral and the natural)?13

In answering the first question, Landau canvases the metaphysical terrain with respect to moral properties and their relation to natural properties: one can be an eliminativist, an identity theorist, or a non-reductionist. As for eliminativism, Landau relies upon earlier arguments in the book against these non-cognitivist and error theorists.14 In the realist camp, moral properties can be construed either along reductionist lines – by reducing and identifying moral properties with natural properties – or along non-reductionist lines. Classical naturalists take the former route while Landau and non-naturalists take the latter.

Landau argues that the open-question argument provides two reasons for preferring the non-reductionist route concerning moral properties. First, after much investigation and many attempts at a successful reduction of moral properties to natural ones, Moore’s open-question argument still remains an open question.15 Second, the best attempts at a naturalistic reduction via a naturalistic moral semantics have not provided any reason for accepting a reductionist picture of moral properties. On the contrary, Landau argues that the best proposals by classical naturalists – Boyd, Brink, and Jackson and Pettit – if successful, actually depart from a reductionist view.16 Thus in answering the first question, the reason we should posit moral facts and properties over and above physical ones is that there has yet to be a plausible reductionist account that adequately produces the necessary reduction.

However, if non-naturalism cannot give a plausible account as to how moral properties relate to physical properties, then classical naturalism may in fact have the upper hand. This is why Landau believes that the non-naturalist, as a non-reductionist about moral properties, needs to explain how the moral and non-moral realms interact. On this score, Landau develops an account that is inspired by property dualism in philosophy of mind in three ways.

First, property dualism upholds our common sense belief that mental properties are not synonymous with, nor identical to, physical properties. One reason for this, maintains the property dualist, is that mental states such as pain seem to be multiply realizable and thus cannot be reduced to, or identified solely in terms of, some physical property. Pain, for example, cannot be reduced to C-fibers firing because it is possible that creatures without C-fibers can be in pain – other physical realizers could conceivably play an equivalent role for any mental state. In a similar fashion, the non- naturalist upholds our common sense belief that moral and physical properties are neither synonymous nor identical. And, one reason for this is that we want moral states to be multiply realizable (i.e., not dependent upon, say, only one descriptive set of circumstances that would make an action right). If reductionism is true, however, moral properties could not be multiply realized.

Second, property dualism – unlike substance dualism – is not “ontologically extravagant.”17 Property dualism is not extravagant, though it does multiple entities: there are two types of properties and two types of facts – mental and physical – and the mental is realized by the physical but not identical to it. In the same way, Landau characterizes his non-reductive non-naturalism as not ontologically profligate. It does not posit moral properties and facts as floating completely free from the physical or descriptive facts, as some characterizations of substance dualism in philosophy of mind have it. Rather, moral properties and facts are realized by descriptive facts and thus partially constituted by them. But, since this is a non-reductive account, these same moral facts and properties in theory could be realized by a different set of descriptive facts. As Landau explains

The difference between the non-naturalist and the reductive naturalist, on this view, is a modal one. Non-naturalists can, and reductionists can’t, allow for the possibility of a moral property’s exemplification by means of some natural property other than the one whose instantiation, at a time, has in fact subserved it.18

Finally, property dualism claims that mental properties supervene on physical properties, such that, necessarily, any change in physical properties results in a change in mental properties. Supervenience is a special kind of relation; namely, one of covariance. As a covariant relation, a supervenience thesis claims that any two things indiscernible in all physical properties are indiscernible in mental properties. Landau’s version of ethical non-naturalism is akin to property dualism in this respect too: moral properties and facts supervene upon descriptive properties and facts, such that, necessarily, any change in the descriptive realm makes a difference on the moral realm. For Landau, “the non-moral features of a situation fix its moral status.”19 This final point is crucial for Landau – as for any moral theory – because we want it to be the case that two situations that do not differ with respect to the relevant descriptive facts should not differ with respect to the moral facts. In other words, if murder is wrong under a certain set of relevant descriptive features, it should be wrong in every case where the underlying descriptive features are realized.

In summary, Landau argues that (1) all of the reductive attempts to identify moral facts and properties with physical ones either (a) leaves the matter an open question or (b) ends up giving a non-reductive account that is compatible with non-naturalism; and (2) ethical non-naturalism can explain the relationship between moral and physical properties in terms of a non-reductivism that parallels the sort of non-reductivism made popular by property dualists in philosophy of mind. Hence, ethical non-naturalism is preferred over reductive naturalism.

Since such an explanation is intended for any type of moral or evaluative fact or property, Landau’s version of ethical non-naturalism supports – indeed it undergirds – a non-naturalist theory of goodness. On such a view, goodness is not identical with any natural or descriptive property or set of properties, nor is goodness synonymous with any natural or descriptive property or set of properties, nor can goodness be reduced to any natural or descriptive property or set of properties. Furthermore, goodness, like other evaluative properties, will be multiply realizable. However, its realizability in some sense depends upon natural or descriptive facts and properties and supervenes upon them, such that, any difference in descriptive realm will result in a difference in the supervening property of goodness.

Foot’s Criticism of Moore’s Non-Naturalism

In Natural Goodness, Philippa Foot rejects Moore’s insistence that goodness is a non-natural property.20 In doing so, Foot exploits a distinction developed by Peter Geach between predicate adjectives and attributive adjectives.21 On Moore’s account, goodness functions as a predicate adjective in the same way that colors do: “x is good” has a similar logical structure to “x is red.”22 According to Foot, this twists the real logical grammar in predications of the sort “x is good”; for “x is good” really means “x is a good F.”

When “x is good” is used as a predicate adjective in the same way that “x is red” is used on Moore’s account, it masks a real logical difference between the two. When we use color words as predicates, the adjective that is predicated functions independently of the noun that we attach to it – redness is redness irrespective of which object we predicate it.23 In the case of goodness, however, whether “x is good” depends necessarily on that to which goodness is attributed. Hence Foot claims that goodness is an attributive adjective, in the same class as adjectives like “large” and “small” – adjectives whose truth of the predication depends essentially upon that to which it is predicated. As Geach puts it, “There is no such thing as being just good or bad, there is only being a good or bad so-and-so.”24 Thus, contra Moore, Foot claims that goodness does not function like a predicate adjective, but an attributive one, such that “Whether a particular F is a good F depends radically on what we substitute for ‘F’.”25 In this regard, Foot offers a neo-Aristotelian naturalist account of goodness whereby the evaluation that some F is good depends upon the class, category, or type of thing F actually is.

On such an account, moral judgments function in ways similar to evaluations of other living things based upon their life-form (or Aristotelian category), the evaluation of which, Foot calls “natural norms.” Such norms are grounded in answering purposive questions about the role certain actions play in “the life cycle of the species,” which Foot equates with the thing’s function.26 Foot first demonstrates how this is accomplished with plants and animals, and then argues that the same logical structure of evaluation applies to humans. In the case of evaluating plants and animals as “good” and “bad,” Foot offers four necessary and sufficient conditions. First, there is the specification of the life cycle consisting of the “self-maintenance and reproduction” for the given species. Second, one establishes how this is achieved for the given species (i.e., what tasks the species performs for the goals of self-maintenance and reproduction). Third, based upon these results, norms are derived for the given species: examples of this would be swiftness in a deer and fireproof bark for a Sequoia tree, respectively. Finally, the norms are thus applied to the members of the species, such that one can evaluate the member according to its performance with respect to the norm.27 A good deer is one that performs well with respect to these norms (say, swiftness); a defective deer is one that does not.

Next, Foot argues that natural normativity is explained in a similar fashion with respect to humans; however, she realizes that to a certain degree rationality complicates this picture. She concludes that given rationality, the human good is sui generis, and yet it nevertheless shares a similar logical structure to the evaluation of non-rational living things:

There is a “natural-history story” about how human beings achieve this good as there is about how plants and animals achieve theirs. There are truths such as “humans make clothes and build houses” that are to be compared with “birds grow feathers and build nests”; but also propositions such as “humans establish rules of conduct and recognize rights.” To determine what is goodness and what defect of character, disposition, and choice, we must consider what human good is and how human beings live: in other words, what kind of a living thing a human being is.28

Not all of the details of Foot’s characterization of the human good concern us here. What is important for our purposes is to see that she offers a naturalist theory of goodness intimately tied to the type of entity in question and the fulfillment of the thing’s function designated by its nature.

This is no doubt a naturalistic account of goodness. One interesting question about Foot’s version of naturalism is whether this is the same type of naturalism offered by the classical naturalists that critics like Moore and Landau target. In other words, does Foot identify the evaluative property of goodness with natural properties of the entity? Given her rejection of Moore’s characterization of goodness as a predicate adjective, Foot would probably insist that her account might sidestep the question entirely. Goodness is not some property that can be applied synonymously to whatever we wish to characterize as good. Instead, goodness, as attributive adjective, is indexed to the proper functioning of particular natural kinds. Fair enough, but we can reformulate the question in terms of her proposal: is goodness for any natural kind identical the set of descriptive facts concerning the function and the fulfillment thereof for that natural kind?

If the answer is “yes,” then we can reformulate the open-question argument in such a way that targets her account:

(A*) I know that X is a good F, but is it a good F?
(B*) I know that X fulfills its function as an F, but is it a good F?

On Foot’s account, it seems that (A*) and (B*) must be identical. Under this reconstruction of Moore’s open-question argument, clearly (A*) is not an open question. What’s disputed is whether or not (B*) is an open question. If it is an open question, then Foot’s account fairs no better than classical naturalist accounts that identify goodness with some set of natural or descriptive properties.

Whether this proposal succumbs to Moore’s argument is difficult to say. On the one hand, (B*) clearly seems different in kind than (B), possibly lending support for the fact that it is not an open question. On the other hand, (B*) relies upon purposive and teleological notions that are out of fashion in today’s world. And, if (B*) is not an open question, then it may in fact beg the question in favor of Foot’s proposal, since it assumes that fulfillment of one’s function just is what goodness means for the type of thing in question.

One way to escape this problem would be to provide some type of argument that goodness just is the fulfillment of one’s function. Presumably, this is what Foot attempts to do in Natural Goodness. However, it seems to me that she has simply offered a descriptive account of what goodness consist in, rather than providing an argument for the identity of the function of the thing and its goodness. We can strengthen her account by supplementing it with Aquinas’s claim that goodness and completion of being are the same in referent but different in sense. On Aquinas’s account, an F is good to the extent that it fulfills its function as the type of thing it is, a function that is demarcated by potentialities in its nature. Aquinas, in contrast to Foot, actually provides an argument for the identity.29 On such an account, goodness and actualization or fulfillment of one’s nature amount to the same thing. If we supplement Foot’s account with this proposal, then goodness and function will not be synonymous – which is why (B*) at first glance seems like an open question – but the proper functioning and the fulfillment thereof provides the basis for attributing the evaluative notion of goodness to an entity as a specimen of some natural kind. Thus, contra non-naturalism, Foot’s proposal would identify the goodness of a thing with some set of relevant descriptive properties, as well as escape the open-question argument.

On the other hand, if the above proposal fails in escaping the open-question argument aimed at identity theories, then maybe it is possible to characterize her account as non-reductive naturalism akin to non-reductive physicalism in philosophy of mind. On such a proposal, goodness can be a multiply realized and supervenient property dependent upon the relevant descriptive properties for its realization, while at the same not reducible to the descriptive or natural properties that characterize the thing in question.

The upshot of this approach is twofold. First, it can account for all of the relations between moral and non-moral facts and properties in ways similar to Landau’s non-naturalism (i.e., multiple realizability, ontological parsimony, and supervenience) without having to posit a sui generis category of values. Second, Landau’s characterization of non-naturalism runs aground precisely where its analogue in philosophy of mind (i.e., property dualism) does; namely, that it makes moral facts and properties epiphenomenal.30 This epiphenomenalism seems benign with respect to goodness; however, it appears that other types of moral and evaluative properties should be causally efficacious.31 If so, we have reason for rejecting Landau’s version of ethical non-naturalism. A non- reductive naturalist approach solves this very worry, for moral and evaluative properties will be causally efficacious.

Whether we characterize Foot’s account as a form of classical naturalism that identifies goodness with some set of natural or descriptive facts that successfully maneuvers around Moore’s open-question argument, or as a version of non-reductive naturalism, Foot’s proposal is preferable to the ethical non-naturalist characterization of goodness: if it reduces and identifies goodness with some set of descriptive facts, it does so without succumbing to Moore’s open-question argument (when supplemented with Aquinas); if it resists the reduction of classical naturalism, it does so without having to postulate a sui generis epiphenomenal category of values.



Aquinas, Thomas. Summa Theologiae. Rome: Marietti, 1950.

Brink, David Owen. Moral Realism and the Foundations of Ethics. Cambridge: Cambridge UP, 1989. Foot, Philippa. Natural Goodness. Oxford: Oxford UP, 2001.

Geach, Peter. “Good and Evil.” Analysis 17:2 (December: 1956): 33-42.

Kim, Jaegwon. Mind in a Physical World. Cambridge: MIT Press, 1998.

Moore, G.E. Principia Ethica. Cambridge: Cambridge UP, 1994.

Shafer-Landau, Russ. Moral Realism. Oxford: Clarendon Press, 2003.


  1. Russ Shafer-Landau, Moral Realism (Oxford: Clarendon Press, 2003) 55.
  2. I am generally following Russ Shafer-Landau’s characterization of realism and its rivals with respect to some domain of reality. See Shafer-Landau, Moral Realism, 13-18.
  3. Shafer-Landau, Moral Realism, 14.
  4. Shafer-Landau, Moral Realism, 14.
  5. Shafer-Landau, Moral Realism, 15.
  6. A third option would be to deny the reality of the realm entirely (i.e., take a non-cognitivist route and claim that there are no moral truths in the strict sense, and then explain our talk of moral and evaluative properties in terms of some function of language, such as Ayer’s emotivism or Hare’s prescriptivism). Since I am presupposing the truth of realism at this point, I shall set this option aside.
  7. This can be accomplished in at least two ways: (1) moral and evaluative properties are identical with natural properties, which is the classical naturalist position and the one that Moore’s naturalistic fallacy and open question argument aimed at criticizing; or (2) a non-reductive naturalism: while moral and evaluative properties depend for their realization on physical properties, they cannot be reduced to such physical properties (See Brink, Moral Realism and the Foundations of Ethics).
  8. Shafer-Landau, Moral Realism, 55.
  9. G.E. Moore, Principia Ethica (Cambridge: Cambridge UP, 1994) 58-62.
  10. Moore actually offers three different versions on the naturalistic fallacy: (1) identifying goodness with some predicate other than goodness, (2) identifying goodness with some analyzable predicate, and (3) identifying goodness with some naturalistic or metaphysical predicate. See Moore, “Preface to the Second Edition,” Principia Ethica, 17. I will focus on the third characterization, since this gets to the heart of the difference between ethical naturalists and non-naturalists concerning the good.
  11. See Moore, Principia Ethica, 95, 112-144 for various formulations, and Shafer-Landau, Moral Realism, 57.
  12. Shafer-Landau, Moral Realism, 57.
  13. Shafer-Landau, Moral Realism, 65-66.
  14. See Shafer-Landau, Moral Realism, 18-52.
  15. Shafer-Landau, Moral Realism, 67.
  16. Shafer-Landau, Moral Realism, 68-71.
  17. Shafer-Landau, Moral Realism, 74.
  18. Shafer-Landau, Moral Realism, 75.
  19. Shafer-Landau, Moral Realism, 77.
  20. Philippa Foot, Natural Goodness (Oxford: Oxford University Press, 2001) 2-3.
  21. Peter Geach, “Good and Evil,” Analysis 17.2 (December: 1956): 33-42.
  22. See Moore, Principia Ethica, 59.
  23. Foot, Natural Goodness, 3.
  24. Geach, “Good and Evil,” 34.
  25. Foot, Natural Goodness, 3.
  26. Foot, Natural Goodness, 32.
  27. Foot, Natural Goodness, 34.
  28. Foot, Natural Goodness, 51.
  29. See Summa Theologiae I, 5.
  30. See Jaegwon Kim’s causal exclusion argument in Mind in a Physical World (Cambridge: MIT Press, 1998), chapter 2.
  31. It is true that Landau responds to this type of argument. But, if one remains unconvinced of the success of property dualism, one should seek other alternatives in the moral realm as well.

Jeff Steele

Jeff Steele is a Ph.D. candidate at the University of South Florida. His dissertation explores the metaphysics of goodness in late medieval philosophy, focusing specifically on John Duns Scotus. [email protected]