Evolutionary Naturalism and the Logical Structure of Valuation:
The Other Side of Error Theory

Richard A. Richards

Abstract: One standard way to conceive value is in terms of properties, as G.E. Moore does when he argues that goodness is a “simple and indefinable” property. Those who advocate this property conception of value often draw an analogy with color. The analogy with color is informative, but since these accounts typically get color wrong, they also get value wrong. Instead of being conceived as properties simply predicated of objects, colors should be conceived as relations between subjects and objects in viewing contexts. Similarly, as evolutionary naturalism makes clear, value is relational in that it is constituted by a relation between object, subject and context, and is grounded on “mattering”. This relational account of value has an advantage over the property conception of value because it better explains actual ethical behavior. But most importantly, it gets valuation right, revealing all the important factors, and making clear how values are subjective—dependent on the nature of the subject, but still objective—grounded on the facts about mattering.

Keywords: Meta-ethics, Values, Evolutionary Naturalism, Color, Properties, Relations, Moore


In the introduction to his Evolutionary Naturalism, Michael Ruse admitted that evolutionary ethics was commonly regarded as “positively unclean.” (Ruse:1995:9) This attitude toward evolutionary ethics is hardly surprising. Many evolutionists believe that ethical systems and moral sentiments are ultimately nothing more than evolutionary adaptations that help humans live together in relative harmony. Ruse explains:

We think that we ought to do certain things and that we ought not to do other things, because this is our biology’s way of making us break from our usual selfish or self-interested attitudes and to get on with the job of cooperating with others… Morality is no more—although certainly no less—than an adaptation, and as such has the same status as such things as teeth and eyes and noses. (Ruse:1995:241)

This implies, according to Ruse and others,[1] that there are no absolute moral values serving as foundation for our beliefs about what is morally right and wrong, and therefore, the human tendency to believe in moral absolutes can be explained away as an adaptive error foisted upon us by our evolutionary past. (Ruse:1995:248-9) I will not here challenge these “unclean” conclusions. I believe they are largely correct. And we should thank evolutionary theory for revealing these sorts of errors for what they are! But there is another more positive side to error theory: what we know about evolution can reveal a common logical error we make about valuation, and how to correct that error.

The now standard starting point for evolutionary ethicists is the position Ruse advocates and describes as “evolutionary naturalism.” This position has two main components. First, there is a substantive assumption that since humans are the products of evolution by natural selection, evolutionary theory can help us understand them and their behaviors. (Ruse:1995:7) Second, there is a methodological assumption—the “naturalism” in evolutionary naturalism. Ruse again: “Naturalism for me…means trying to understand through empirical law. This means that you have got to appeal to experience—you cannot just think things through a priori…” (Ruse:1995:2) Naturalism is, in this sense, a commitment to understanding human nature through scientific investigation.

For the evolutionary naturalist, an important source of insight comes from what we know about the evolutionary processes that produced modern humans. One approach looks for insights into human nature through particular adaptation hypotheses about social behavior. This is an appropriate and often promising strategy. But what I want to focus on here does not get grounded on any particular adaptation hypothesis. It is grounded on the most basic commitments of evolutionary theory. The concept of fitness, I will argue, is the fundamental valuational concept—for humans and all other creatures—and it implies that valuation is relational and subjective, yet still objective in important ways. In short, evolution tells us that values are complex relational facts.

To make my case, I will begin by identifying where standard conceptions of valuation go wrong—by trying to conceive value in terms of simple predication and properties. Second, I will explain how evolutionary fitness reveals the relational structure of valuation. Third will be a sketch of how ethical values might fit into this relational structure. Fourth, I will distinguish this approach from similar accounts of value and defend it from some obvious objections. Fifth, I will sketch out an error theory in terms of logical structure. Finally I will summarize the advantages of the relational view, and outline some possible philosophical implications. In the end, I hope to have shown that the relational conception of valuation not only gets the logical structure of valuation right, but it also provides the basis for a framework that can better explain our actual moral beliefs, practices and sentiments.

I. The Received View: Values as Properties

The problem with standard conceptions of valuation and values at issue here is at bottom logical, but it is manifested in many different ways of conceiving value. Standard conceptions sometimes go wrong by conceiving value propositions in terms of simple predications like “x is good”. One influential approach begins with this structure and conceives values in terms of properties. G. E. Moore does this in his Principia Ethica, where he argues that appeals to natural properties like pleasure and happiness in defining goodness commit the “naturalistic fallacy.” His basis for this conclusion is the “open question” argument: goodness cannot be defined in terms of natural properties because for any natural property or set of properties, it is an open question whether it is good in any particular case. (Moore:15) Moore concludes that goodness must be a simple, indefinable property.

(1) The peculiarity of Ethics is not that it investigates assertions about human conduct, but that it investigates assertions about that property of things which is denoted by the term ‘good,’ and the converse property denoted by the term ‘bad.’… (2) This property, by reference to which the subject-matter of Ethics must be defined, is itself simple and indefinable. (Moore:36)

Moore is not alone in thinking about value in property terms. Many of those engaged in the debates about naturalism and non-naturalism, intuitionism and realism adopt similar stances. In one recent ethics textbook, for instance, the author claims that: “Moral realism commits itself to the existence of external, independent moral properties.” (DeMarco:258) And in criticising Ruse, Rottschaefer and Martinsen give an evolutionary account that also conceives value in terms of properties.[2]

We have adopted a non-reductionist account of the moral realm, according to which the moral properties of persons and things, moral rightness and goodness, are distinct from natural non-moral properties because they supervene on the latter. (Rottschaefer and Martinsen:398)

Similarly, John McDowell, in his influential “Values and Secondary Properties,” argues that moral properties are analogous to “secondary properties” like colors. (McDowell) In doing so, McDowell is following the lead of Moore, who had drawn a similar analogy to color perception:

There is… no intrinsic difficulty in the contention that ‘good’ denotes a simple and indefinable quality. There are many other instances of such qualities. Consider yellow, for example. We may try to define it, by describing its physical equivalent; we may state what kind of light-vibrations must stimulate the normal eye, in order that we may perceive it. But a moment’s reflection is sufficient to shew that those light vibrations are not themselves what we mean by yellow. They are not what we perceive... The most that we can be entitled to say of those vibrations is that they are what corresponds in space to the yellow which we actually perceive. Yet a mistake of this simple kind has commonly been made about ‘good.’ It may be true that all things which are good are also something else, just as it is true that all things which are yellow produce a certain kind of vibration in the light. And it is a fact, that Ethics aims at discovering what are those other properties belonging to all things which are good. (Moore:10)

The color analogy, drawn also by others,[3] is right in an important respect. But since Moore gets color perception wrong, he also gets valuation wrong. How he gets color perception wrong is instructive in understanding how he, and others who adopt this property approach, generally get valuation wrong.

Ever since Galileo and Locke, colors have typically been conceived as secondary properties, and distinguished from primary properties such as shape and size. Usually, it is claimed that primary properties are independent of the subject, while secondary properties are not. (Hilbert:3-5) The shape of an object, for instance, does not depend on any facts about the observer. Colors, on the other hand, depend on the observer having the right constitution and being in the right environment—nothing is really yellow in absolute darkness or to a blind man. What is right about this conception of color that color perception is subjective—dependent on facts about the perceiver.[4] What is wrong is the attempt to conceive of color as nonetheless still a property of the object. This implies that the property is somehow both in the object and the subject. The object is yellow, but the yellowness is in the subject. Barry Maund begins his book on color with this ‘duality’ problem:

The status of colour has for long been puzzling. Almost everyone agrees that physical objects have colours: that sunsets are golden or red; that bananas are yellow; that claret is purple; and so on. Everyone agrees that objects are perceived as coloured. Where there is disagreement is over the nature of the colour physical objects have and the nature of the colour we perceive objects as having, and indeed whether the two are the same. (Maund:195:xiii)

There have been many diverse attempts to make sense out of this picture. Maund believes he has a solution to this puzzle in his (perhaps even more puzzling) conception of color as a “virtual property”: “an intrinsic, nonrelational property of physical objects, but one which no object in the actual world has.” (Maund:1995:xiii) Most of the other attempts to make sense out of this ‘duality’ of color—as being in both the object and subject—similarly conceive colors as properties, in various ways as: objective, irreducible and supervenient; objective and reducible; dispositional relative to light; dispositional relative to physical response; dispositional relative to appearance; or dispositional relative to functional role. (Maund:2002:4)

I can hardly evaluate all these property theories here. But a simple solution to the duality problem is to be found in the insight that color perception is relational: an object has a particular color only relative to certain perceiver under certain conditions. In absolute darkness nothing is really yellow. In red light everything is red. And for those who lack all color vision entirely, nothing is ever a color in the ordinary sense. Evan Thompson’s “ecological” approach is relational in this sense. According to Thompson, color is always relative to a perceiver and a “photic environment,”[5] and there are therefore no “absolute”, non-relational facts about color. (Thompson:2000:179-180) What this implies is that colors are not properties simply predicated of objects. In other words, “that banana is yellow” does not have a logical structure that accurately reflects color perception. A color predication that more accurately reflects perception, would involve a relation among an object, a subject that perceives the object, and the conditions under which the subject perceives the object. We can schematize this relational logical structure in the following schema:

Relational Color Schema: “w is color x for y in context(s) z.”

Here w is an object, x is a color term, y is a person or persons, and z is a description of, or reference to, a particular context.

There are two advantages to thinking about color in this relational way. First, it gets color and color perception right. Colors exist because of how the visual systems of subjects (organisms) interact with objects in the world. This fact is universally recognized and is the basis for the duality problem described by Maund above. Thinking of colors as properties of objects, rather than relations, simply misrepresents colors and color perception. Second, the relational schema provides a framework that allows us to actually say something about actual patterns of color perception. Depending on our purposes, we can specify human color perception relative to ideal viewing conditions and ideal viewers, or, to the many ways that “normal” human vision varies. We can also adopt an evolutionary perspective and understand color perception in other species and how it functions relative to survival and reproduction. If we think of colors simply as properties of objects, we can say little about the many differences in how subjects perceive color.

What psychophysicists have discovered about human color vision, is that there is more variability than we might suppose. Human color vision is normally trichromatic, based on the presence of three kinds of cones, each with a different absorption spectrum. Deviations from normal color vision include the total absence of color vision found in monochromats, or the limited color vision found in dichromats, who lack one of the three kinds of cones, and are therefore are unable to distinguish some colors distinguishable by trichromats. Dichromats who lack the “green” cone (deuteranops), for instance, cannot distinguish green from red, or green from blue. (Coren:102) On the other hand, tetrachromats, who have four kinds of cones, and consequently an additional absorption spectrum, can distinguish more colors than can trichromats. (Coren:105) When confronted with a visible color spectrum, trichromats can typically identify 7 or 8 distinct color bands, while dichromats will identify only 5 or 6, but tetrachromats will identify around 10. (Coren:108) But even among the “normal” trichromats, there is significant variability in terms of the respective absorption spectra, and abilities to distinguish color. And in “anomalous trichromats,” there is an even more extreme variability, resulting in significant differences in the ability to identify and distinguish colors. And in one striking case, a subject had one eye that was trichromatic, while the other was dichromatic. It was possible to determine, through color matching, that when the trichromatic eye saw green, yellow, orange or red, the dichromatic eye saw only yellow. And when the trichromatic eye saw violet or blue, the dichromatic eye saw only blue. (Coren:103)

What this all suggests is that among humans, and even “normal” humans, color perception is highly variable and dependent on the subject. But there is even more variability if we look to non-human subjects. Tetrachromacy is common among vertebrates, including birds, fishes, amphibians and reptiles. And there is sensitivity among vertebrates, especially birds, to ultraviolet wavelengths:

Avian color vision is more complex than our own because it is at least tetrachromatic, perhaps even pentachromatic. It uses more of the available spectrum because its visual sensitivity and wavelength discrimination extend considerably into the ultraviolet (UV) region (300-400nm). (Thompson:2000:175)

This implies, according to Thompson, that a thing can have different colors to different perceivers.

…Because colors are perceiver relative, the same thing can simultaneously have different colors. For example, the surface of an object can be color one color to me and a different color to a tetrachromatic pigeon. (Thompson:2000:180)

It is unclear how the property conception can make sense out of all these varying patterns of color perception among humans and other species. The bottom line is that on this relational conception of color, and unlike the property conception, we can actually say something about the circumstances under which something can be a particular color to someone. And we can explicitly recognize all of the factors operating in color perception. This is an important advantage of the relational conception of color. Conceiving of color as a property—even as a supervenient property—simply suppresses important factors in color perception.[6] Little is to be gained from this.

It is worth noting that the relational nature of color perception cannot be avoided by appealing to causal dispositions. On this strategy, we would identify the color of an object with its causal disposition (perhaps by virtue of its primary properties) to produce the perception of a particular color in normally sighted humans under standard viewing conditions. Color would then be identified with a simple property—a causal disposition.[7] In this case we could translate “x is yellow” into something like “x has the causal disposition to produce a perception of yellow under normal conditions for normally sighted humans”. The causal disposition would then just be whatever properties caused this tendency to produce the relevant kind of color perception. But by now it should be apparent that causal dispositions are relational. A yellow flower may have the power to produce a perception of yellow in “normal” trichromatic humans. But it may have a different power or disposition relative to dichromatic or tetrochromatic humans. And since it reflects in the ultraviolet range, it will have yet a different disposition relative to the birds that see ultraviolet light. There will be as many causal dispositions as there are distinct visual systems! To say simply that it has a disposition leaves out an important fact—that dispositions are always relative to perceivers.

II. The Fitness Model: Evolutionary Values as Relations

The value of evolution in understanding human behavior lies partly in the fact that it provides a general account of organic phenomena that is not subject to the errors that may arise from a narrow focus on the human point of view, or an outdated and mistaken way of conceiving human nature.[8] What evolution tells us is that valuation is relational in a manner similar to color perception. This is apparent in Darwin’s formulation of his theory in 1859 (emphasis added).

It may be said that natural selection is daily and hourly scrutinizing, throughout the world, every variation, even the slightest; rejecting that which is bad, preserving and adding up all that is good; silently and insensibly working, whenever and wherever opportunity offers, at the improvement of each organic being in relation to its organic and inorganic conditions of life. (Darwin:84)

Here Darwin is laying out the basic idea behind natural selection—the concept of fitness. On this conception of evolutionary value, an organism is good—fit—insofar as it functions well with respect to survival and reproduction in a particular environment. And a trait makes an organism good or fit, insofar as it assists in the functioning that supports survival and reproduction in the relevant environments. Ernst Mayr echoes Darwin when he tell us that in selection,

…The “goodness” of the new individual is constantly tested, from the larval (or embryonic) stage until adulthood and its period of reproduction. Those individuals who are most efficient in coping with the challenges of the environment and in competing with other members of their population and with those of other species will have the best chance to survive until the age or reproduction and to reproduce successfully. Numerous experiments and observations have revealed that certain individuals with particular attributes are clearly superior to others during this process of elimination. They are the ones that are “fittest to survive.” (Mayr:119-120)

The widespread acceptance of this relational fitness conception of goodness is so complete that no competent biologist would deny it.

The logical structure of this fitness conception of goodness is apparent in the ways we talk about fitness. No evolutionist says that traits like feathers are simply good. Rather they are good in particular ways for particular organisms and in particular contexts or environments. We recognize, for instance, that while feathers may be good for penguins—who use them as insulation in cold climates, they would not be good for tapeworms in the tapeworm environment—the gastrointestinal tracts of other organisms. What this basic fitness concept implies is that a trait (morphological, behavioral, etc.) can be good for an organism with respect to some way of functioning for that organism in an environment. On this basis, we can schematize ‘goodness’ in the following ways:

Simple Valuation Schema: “w is good (or bad) with respect to x for y in z.”

Here w is a trait, x is a way of functioning (relative to factors relevant to survival and reproduction), y is an organism, or group or kind of organism, and z is a context/environment, type of environment, or range of environments. There is also a derivative, comparative schema:

Comparative Valuation Schema: “v is better (or worse) than w with respect to x for y in z.”

Here v and w are alternative traits, and judgment is relative to the comparative goodness of traits. This kind of goodness is most relevant to selection processes. Notice that we can use these schemata to evaluate more than just traits in the narrow sense—as features of organisms. We can also evaluate organisms in terms of overall phenotypic fitness by instantiating appropriately: “phenotype w is good with respect to survival and reproduction for y in z”; and “phenotype v is better/worse than phenotype w with respect to survival and reproduction for y in z”. And we can evaluate external objects and environments. A rock can be good with respect to cracking nuts for humans, a particular stick or gum wrapper can be good for male bower birds in enticing females, a shell can be good for a hermit crab with respect to its shape, and a sunken ship may be good for an eel with respect to all its hiding places. In the grand evolutionary scheme, objects and environments have value in the same way that morphological and behavioral traits have value.

What has value—is good or bad—in all these cases will be determined by what “matters”, or could matter, to a subject.[9] Since there is an indefinite number of ways that something might matter, it is hard to be very precise here, but the most obvious way something might matter, and therefore have value or disvalue, is by direct implication for survival and reproduction. This is how feathers matter to birds. But mattering is not limited to that which has direct implications for survival and reproduction. Something has value or disvalue if it might matter—can potentially affect survival and reproduction in some possible context. So what is usually regarded as having non-instrumental value such as pleasure, will matter or could matter, because feelings of pleasure might have implications for the reinforcement of certain behaviors,[10] that might in turn have implications for survival and reproduction. Consequently, mattering and value can get grounded on things like the pleasure social creatures get from the company of others, the enjoyment carnivores get from play, and the satisfaction philosophers get from solving philosophical puzzles. In generally, anything that has some potential effect on the subject can have value or disvalue to that subject. In the end, what matters to a subject, and therefore has value, is an open question, one answerable by empirical investigation into the nature of subjects and the environments in which they live.

What is important here is that first, these valuation schemata have relational structures that correctly represent all that is important for valuation in nature. Whether long necks are really good for giraffes, depends on how they function relative to the nature of giraffes, and the relevant environments—and for the functioning of selection, the presence of alternative traits. Second, the semantics of valuation—how we instantiate the variables—allows us to make the required distinctions. A trait may be good with respect to one kind of functioning, but not good with respect to another. Long necks may be good for browsing, but not for cardiovascular functioning. (It is not easy pumping blood to such heights.) Furthermore, long necks will be good in this way only relative to some environmental conditions—the presence of trees of a certain kind, height and density.

Third, it is also important to see that valuation will have different degrees of generality relative to each of the variables. Relative to the x’s: some kinds of mattering are more general in terms of overall survival capabilities, while others are more narrow. The highly specialized gastrointestinal system of the Panda, for instance, functions well in a narrow way, facilitating the digestion of some species of bamboo. But it functions poorly in a broad way in that it functions well only relative to the digestion of this bamboo. Relative to the y’s: some kinds of mattering may be good for a single individual organism or for a group of organisms, or multiple life forms.[11] A trait or behavior could be good, for instance, for a single ant, an ant colony, an ant species or an entire ecosystem. Relative to the z’s: mattering can be with respect to narrowly prescribed environments or a wide range of environments. The digestive apparatus of mice, for instance, allows them to eat a wide variety of foods from a wide variety of sources. They are therefore good (function well) in this respect in a wide variety of environmental contexts.

Fourth, we should recognize that while goodness (value) is subjective—it relates mattering to subjects—and something is therefore good only relative to a subject, goodness is also objective: whether a particular trait is good or bad in some particular way, is a factual and empirical matter. Something is not good or bad just because someone thinks it is. The objectivity of valuation is supported in the strongest possible way—by consequences for life, reproduction and death. Just as no evolutionary biologist would deny the objective facts of fitness, no philosopher should deny the objective facts of valuation. And we need look only to our own conditions of life to discover this goodness or badness relative to our own functioning. Evidence is found in the advertised remedies for back pain, sinus headaches, myopia and obesity related problems. And we can observe the valuational facts in our pets, their joint problems, diabetes, and all the breeding related flaws.[12] What is important is that value judgments in the form of these valuation schemata will have truth values in a straightforward way, and with no less objective certainty than the facts of life and death.

Finally, and most significantly, there will be regularities in mattering relations and valuation. Feathers function similarly in similar species—whether to assist in locomotion through gaseous environments or as insulation against the cold of polar environments. We can therefore formulate value generalizations of varying degrees. This fact is, of course, something we already recognize. We are all aware that long fur and blubber function well in cold environments. The relational conception of valuation gives us a framework within which we can formulate these generalizations, and see what they involve—reference to ways of mattering, the nature of the subjects, and environments.

In the introduction, I claimed that values are complex relational facts. We are now in a position to clarify this. I have so far been concerned primarily with valuation and not values per se. This is because valuation is fundamental, and whatever values are, they are derivative of, and based on valuation. This is a consequence of the naturalistic starting point. But since the term ‘value’ gets used in many ways—sometimes just to refer to beliefs or attitudes about value, it is worth specifying what values are on this relational approach. As I see it, values are best understood as the patterns of valuation that are brought out in the relational valuation propositions. Whatever I may believe about the value of long necks, opposable thumbs, or trichromatic color vision for various subjects, there are facts about how these traits function for their possessors and in various environments. These facts are values, as are the patterns of functioning. Consequently, there may be individual, group, species and even supra-specific values. The well functioning of bipedalism for primates in certain environments, for instance, constitutes a supra-specific value.

The implication of this concept of value is that there is no simple fact/value distinction. Rather there are value facts and non-value facts. Value facts are about things that matter to subjects—and consequently have the relational logical structure I have schematized here. Non-value facts make no reference to what matters for subjects, and consequently have a different logical structure. We should, on the view presented here, recognize the existence of such valuation facts as part of our basic description of the natural world.

III. Human Ethical Values

How does this general account of valuation apply to ethical values? The starting point is the insight that ethical values are found in social mattering. In evolutionary terms, our ethical practices and moral sentiments are adaptations to help us survive and reproduce in our social environments. If so, then an individual is good with respect to some social functioning in social environments, and a trait, behavior, action, etc. is good or bad in various ways for specified individuals in specified contexts. What this implies is that social goodness is found in the semantics of valuation: what counts as having ethical value or disvalue depends on how we instantiate the variables. In order to have social value, a trait must function well in social contexts—environments where the social interaction with other humans matters. Politeness, language skills, sense of humor, etc. typically function in this way. We can restrict the instantiation of variables in the schemata to reflect this social mattering:

Simple (Social) Valuation Schema: “w is (socially) good (or bad) with respect to x for y in z.”

Comparative (Social) Valuation Schema: “v is (socially) better (or worse) than w with respect to x for y in z.”

Here v and w are traits (behaviors, tendencies, sentiments, desires, motives, etc.), x is some kind of social mattering, y is a social creature or group of creatures, and z is a social context.

As with valuation in general, it is difficult to say with precision what might matter socially and therefore have social value or disvalue. But there are obvious examples: “Wearing clothes satisfies privacy needs of Southern Baptists in church.” Or: “Lying creates distrust among family members in the family contexts of western cultures.” The valuation schemata do not, however, try to distinguish instances of social functioning that are obviously ethical from those that are not. Certainly we need to be able to make this distinction. The behavior of serial murderers is bad in a way that violations of mere etiquette are not. To do this satisfactorily will require some sort of naturalistic theory of normativity and obligation. I shall not try to say here what this would involve. Nonetheless, we can still see that some ways of functioning have more at stake because of the kind of mattering at issue. For instance, it is more important biologically and socially for a parent to support the functioning of their children with respect to food and shelter, than it is with regard to providing them with fashionable footwear. Consequently, we generally regard the failures of parents to feed their children to have much greater disvalue than the failure to buy them the newest style of shoe. This is a simple and straightforward way of distinguishing the different kinds of social mattering—one kind has implications for survival of offspring, the other does not. Since the survival of our offspring typically matters to us (biologically and psychologically), there is a significance associated with this mattering that is not associated with social functioning that does not have such significant implications. Perhaps this kind of social significance can ground a theory of normativity that can in turn justify the “oughts” and “shoulds” of normative ethics.

But whether or not a naturalistic theory of normativity is possible[13], we can see how our actual judgments about ethical valuation are often relational in this way. We do in fact condemn some behaviors of some people, in some contexts, but not others. Certain kinds of sexual behavior, for instance, are fine for some consenting adults in some contexts, but are morally reprehensible in other contexts—with children, for example, or during church services and at restaurants. What is most important is how this relational approach allows us to make useful and necessary distinctions. First, the various schemata explicitly recognize that there may be conflict in function and value. A trait, behavior, or action may function well in one respect, but not another. Helping Socrates escape from prison may be good in that it saves a virtuous person from an undeserved punishment. But, as Socrates recognized, this action may also have disvalue relative to other functions, in particular, the general respect for law. Second, a trait, behavior or action might be good for some people but bad relative to others. Robin Hood’s redistribution of wealth from the rich to the poor might be good with respect to the wealth of the poor, but it would be bad relative to the rich.[14] Third, as in general evolutionary valuation, there can be different degrees of subjective generality. Something can, in principle, be socially good relative to y’s of different scope, from a single individual to a family, community, nation, or even an entire species.[15] Fourth, social functioning can also have different degrees of generality relative to context. A trait, behavior, or action can be good in a single context or many. Killing other humans is bad in many contexts, but good in some cases of self-defense. Notice also that culture is part of the context, so we can accommodate those cases where cultural relativism is appropriate. But since culture is only part of environmental context, valuation cannot be reduced to a cultural relativism. Fifth, the comparative schema makes it clear that in cases of social mattering (as in biological functioning) often we want to know more than whether a trait, behavior or action is simply good, we want to know if it is better than the alternative. We not only want good functioning, we want the best functioning. This relational approach explicitly recognizes this fact.

Finally, and highly significantly for practical purposes, we can formulate generalizations about social functioning. In similar social contexts, and for similar subjects, similar traits have similar value. Violence within a nuclear family is dysfunctional in similar ways among families in Alabama as well as in Afghanistan. We can therefore formulate many ethical principles that will generally function in the same way—even if there may be subjects or environments that are problematic. Just as we can formulate generalizations about the value of feathers and hearts in various contexts, we can formulate generalizations, and consequently general rules about lying, killing, sexual behavior etc., but without being absolute (non-relational).[16] On the other hand, many values will not be general in this way if functioning depends on the particular facts of the relevant relations. Whatever the case, the degree and nature of value generalizations will depend on the particular ways the schemata are instantiated (the semantics of valuation), and empirical facts about humans and their social environments.

At this point, a caution is in order. No automatic and direct inferences can be made about the past or future selective value of social traits on the basis of any particular valuation. For instance, we cannot automatically infer that a sense of humor has or will be favored by natural selection simply because it is socially good in some particular way, for some person, in some social context. Most obviously this inference fails because a trait might be good in some respect, but bad in some other more important respect. Or a trait may be good only in a narrowly specified environment, but bad in many others. Even more importantly, that a trait is good in some way does not imply that it is better than the alternatives—as selection would require. Finally we need to determine whether a phenotypic trait or behavior has some genetic basis that can be selected for. All these complications suggest that inferring the action of natural selection from any particular case of valuation is highly problematic. Conversely, inferring valuation from the operation of natural selection is also problematic. That a trait functions in particular way in the past does not imply that it continues to function that way. Xenophobia may have served humans well in an ancestral environment where strangers were likely to pose a significant risk. But it may serve us poorly in modern environments where the risk is small.

IV. A Defense of The Relational Conception

The value theory developed here has something in common with other accounts of value. But each of these accounts fails to get valuation quite right, primarily by excluding some kinds of value. This failure is due, at least in part, to the fact that these accounts do not explicitly identify each of the relevant factors through a valuation schema. The schemata presented here are useful in that they force us to recognize the full relational nature of valuation, in all its varieties and relative to all the important variables. This is best understood by contrast with some similar, yet more limited value theories.

I borrowed the term ‘mattering’ from Peter Railton to identify the sorts of things that might have value. He has used this notion to develop a naturalistic and relational account of value, arguing that ethical value originates in mattering.

We begin by thinking of value as essentially subjective, arising from mattering. In any nonmetaphorical sense of mattering, we should be able to fill in the formula “x matters to y for G.“ (Railton: 105)

This account is relational in that mattering is relativized to subjects—the “y” in the schema presented. But as his schema also makes clear, it is not relativized to environment or context. This is an important omission. Just as feathers are good only in some environments, so are many social behaviors and actions. While there is no obvious reason why Railton could not consistently relativize his conception of value to environment or context, he does not explicitly do so.

There are other differences as well. First, Railton’s focus is on human value and he does not seem to be interested in a general, non-anthropocentric theory of value. As far as I can see, there is no reason he could not give a more general theory, but the starting point in human value may have consequences not found in an approach that begins in evolutionary value in general. An anthropocentric theory may come to overemphasize the role of distinctly human traits—conscious belief and reason for instance.[17] Second, even though Railton recognizes that valuation is relational, he still seems intent on conceiving goodness as a property.

I will argue for a form of moral realism that holds that moral judgments can bear truth values in a fundamentally nonepistemic sense of truth; that moral properties are objective, though relational, that moral properties supervene upon natural properties, and may be reducible to them; that moral inquiry is of a piece with empirical inquiry… (Railton:5)

While there is much I agree with here, I find Railton’s claim that there are moral properties that “supervene” on, and may be reducible to, natural properties puzzling at best. Nonetheless, Railton’s commitments here may seem appealing to those who want to save the property conception by appealing to “supervenience.” According to this strategy, the property of goodness supervenes on other properties. What value supervenience involves here is not exactly clear, but it seems to imply that the goodness of a thing is dependent on its natural properties, but is still distinct from (and usually not reducible to) those properties.[18] This is the approach typically advocated by those who think that goodness is a property like color.[19] But since this supervenience relation is usually understood to be among properties and does not include reference to a subject and context, the appeal to supervenience is neither helpful, nor harmless. It is not helpful in that it simply excludes the factors that are often most important in our deliberation about value—the nature of the subject and the context. We simply cannot decide whether something is good or bad until we know for whom it is good or bad, and the relevant context. It is harmful in that it may mislead us about the nature of value, implying that valuation is independent of subject and context.

Railton’s commitments in this passage suggest another way to save the property conception of value—via relational properties. Goodness is then a property of an ordered set of variables: the trait in question, the manner of functioning, the subject and the environment. I have no objection to this, as long as the necessary factors are included in the relation, and the property is understood to be of the ordered variables, not of a single variable—the trait or thing in question. Similarly, we might try to save the property conception of value by conceiving goodness as a causal power. Once again, as long as we conceive this causal power as a relation between the thing that has the causal power, the thing that is affected by it, and the circumstances under which it has the power, then I have no objection.

Another theory of value that has something in common with the approach advocated here, is developed by Philippa Foot, who, like Railton, proposes a naturalistic and relational theory. But unlike Railton, her theory is explicitly general. She argues that value in non-human species is the same sort of thing as value in humans.

…I believe that evaluations of human will and action share a conceptual structure with evaluations of characteristics and operations of other living things, and can only be understood in these terms. I want to show moral evil as ‘a kind of natural defect’. Life will be at the centre of my discussion, and the fact that a human action or disposition is good of its kind will be taken to be simply a fact about a given feature of a certain kind of living thing. To make such a suggestion, as I interpret it, is to contemplate a naturalistic theory of ethics. (Foot:2001:5)

But, as friendly as this view is to my own approach, it diverges at a key point. Foot identifies value only relative to abstracted life cycles of species in abstracted environments, and not in the many other ways something can matter—be good or bad in all sorts of environments. This comes out in how she relates goodness to “Aristotelian necessities.”

…It is necessary for plants to have water, for birds to build nests, for wolves to hunt in packs, and for lionesses to teach their cubs to kill. These ‘Aristotelian necessities’ depend on what the particular species of plant and animals need, on their natural habitat, and the ways of making out that are in their repertoire. These things together determine what it is for members of a particular species to be as they should be, and to do that which they should do. (Foot:2001:15)

While I have no objection to her proposal that we can and should consider goodness relative to species-specific life cycles and in particular “natural habitats”, these are not the only considerations relevant to valuation. Something can matter—have value relative to an individual in an atypical way, (non-species specific) and in an atypical (“non-natural”) environment, and an adequate theory of value must be able to accommodate these facts. If we regard “Aristotelian necessities” as the basis of valuation, then we cannot clearly distinguish the many ways that things matter independently of ‘Aristotelian necessities’ and natural habitats.

At bottom, this problem with Foot’s conception of value seems to originate in her essentialist tendencies to focus on the “teleology of the species” over all the contingencies of existence that are emphasized by evolutionary theory (emphasis added).

We start from the fact that it is the particular life form of a species of plant or animal that determines how an individual plant or animal should be: the Aristotelian categoricals give the ‘how’ of what happens in the life cycle of that species. And all the truths about what this or that characteristic does, what its purpose or point is, and in suitable cases its function, must be related to this life cycle... We could say, therefore, that part of what distinguishes an Aristotelian categorical from a mere statistical proposition about some or most of all members of a kind of living thing is the fact that it relates to the teleology of the species. (Foot:2001:32-3)

It is not insignificant that in her Natural Goodness, she mentions evolution just in two footnotes (Foot:2001:32,40) and only to reject the possibility that evolution can provide insight into values.

A third approach that also has something in common with the theory of value developed here, is presented by William Casebeer in his Natural Ethical Facts: Evolution, Connectionism and Moral Cognition. This is a naturalistic and relational approach to value (albeit without an explicit value schema), but it focuses more on the cognitive aspects of ethics. It has many virtues, but suffers from some of the same limitations of Foot’s approach. In particular, while valuation is relational, it is indexed only to a particular kind of functioning, what Casebeer describes as a “modern history” concept of functioning. He quotes Peter Godfrey-Smith with approval: “functions are dispositions and powers which explain the recent maintenance of a trait in a selective context.” (Casebeer:52) The basic idea is that something is good or bad relative to the recent selective environments that are responsible for its presence.

It is not too difficult to see that this is a limited sense of value. Something might matter in a positive or negative way without being part of a selection based explanation for its presence. In other words, something can be good or bad independently of any recent evolutionary function. Casebeer seems to fall into the same Aristotelian trap that caught Foot, limiting value to a particular kind of “proper” functioning. He virtually admits as much, describing his own view as a “pragmatic neo-Aristotelian virtue theory.” (Casebeer:2) By limiting value in this way, he cannot recognize the full range of respects with which something might matter independently of modern history function. It is not clear, for instance, that the enjoyment I get from solving a philosophical puzzle, or hearing an opera aria sung well, admits a recent evolutionary function. If so, then this kind of mattering could not have value on Casebeer’s approach. Nor can he recognize the full range of environments that may affect mattering. But surely we can say something is good or bad relative to environments that did not affect past selection—in particular, the many social and cultural environments that are of too recent origin to affect our evolutionary development. This is not insignificant. The recent selective environments of humans did not include the cities and towns we now live in. It is not clear how Casebeer’s theory of value can help us understand mattering in a modern city. In the end, we need a theory that can accommodate the full range of mattering, and Casebeer’s cannot.

Furthermore, Casebeer, like Railton, apparently cannot abandon the property conception of value, and its associated supervenience strategy, arguing:

Functional properties are not “strange” or “odd” properties that could not supervene on matter in any comprehensible way. Rather functional properties are interesting and conceptually tractable, and they can serve a useful purpose in scientific theories, particularly in the biological sciences. (Casebeer:53)

But I see no reason why Casebeer cannot think of functions and value in fully relational terms, so it is unfortunate that he ignores his own suspicions about supervenience:

…That mere supervenience relations, though acceptable in a developing science, often are used as an excuse not to explore the phenomena in question in more depth, or, in the worst of cases, merely restate a problematic relation rather than “solving” it. (Casebeer:11)

Casebeer would have been better off recognizing the fully relational nature of function—and more broadly value—and avoid the property conception of value altogether.

There is one last difference between the relational conception and the theories advanced by Railton, Foot and Casebeer. Each of these thinkers has a tendency to postulate the existence of some “the good” for an individual (and a species) as if there is a set of things that constitutes a single, coherent, encompassing good. For Railton it is found in a set of “objective interests” (Railton:2003:11); for Foot it is to be found in the Aristotelian necessities; and for Casebeer it is found in the modern history functions. But once we recognize that goodness is always relative to respects, subjects, and environments, and we recognize there always will be systematic conflict in the various ways of being good, it is clear that there is no “the good” in this sense for any single individual, group or species. There are just many different, independent and often conflicting goods.

There are some obvious objections to the evolutionary and relational conception of value advocated here. First, many will argue that the relational conception makes morality subjective because valuation is always in relation to subjects and contexts. This conclusion is correct. But I hope it is clear by now that values are also objective in the sense that they are based on facts about mattering. We should also recognize that even though we may want absolute, non-relational values, there are many things we want but cannot have (immortality and less dysfunctional sinuses for instance), and that we may want them does not imply that they exist. Nor is it even clear that we really want non-relational values. Why would we want values that don’t depend on our natures and actual valuation? Ultimately, don’t we want norms and rules that function well, to make us function well, relative to both our natures and the situations in which we find ourselves? Absolute, universal (non-relational) values cannot be responsive to the facts of our existence.

Nonetheless, an implication of this subjectivity is that seemingly monstrous acts can still be good. Torturing children for fun might be good for the torturer with respect to the pleasure it brings. Once again this conclusion is correct. But this is only one small part of the picture. Not only would the torturing likely be bad in many ways for the children tortured, it would also likely be bad for the torturer in many ways and contexts. It might, for instance, have psychological effects that reduce the torturer’s ability to function in human societies. And certainly if other humans knew of the torturing, they would be unlikely to trust and respect the torturer. Furthermore, torturing children would likely be bad for a community in many ways in the patterns of interaction learned by those who were affected. The bottom line is this: once we see the many ways such actions would be bad for the torturer, the children tortured, and others, we can see that this narrow goodness relative to the torturer’s pleasure is trivial.

I hope it is clear by now that this relational account of value it is not relativistic or subjective in the usual sense. First, it is not a cultural relativism nor is it purely subjective. That is to say that things are not good or bad only relative to a culture or individual. Since culture is part of the human social environment, some things will be good or bad relative to culture. Driving on the right side of the street is good with respect to safety and efficiency in the United States, because it conforms with conventions and expectations. But it would be bad on similar grounds in Great Britain. And since valuation is subjective, some things will be good or bad relative to a single individual. Just as a flavor of ice cream may be good with respect to the pleasure it brings, a particular sexual preference or way of life may be good or bad with respect to the pleasure it brings. But many valuations will not be dependent on cultural context or subjective preference in these ways. Necrophilia is bad in many ways relative to the health of those who engage in it—regardless of cultural values or subjective preference. Furthermore, the fact that the ethical facts are relational does not imply they are mere opinion, as many philosophers assume.[20]

Finally, non-naturalists might object that this approach to valuation simply leaves out the non-natural, it does not establish that there are no non-natural valuational properties. This is partly correct. I do not think it is possible to prove that there are no non-natural values, any more than it is possible to prove there are no Platonic forms—even though it is unclear how we could have good empirical grounds for believing in things like Platonic Forms. Nonetheless it is possible to draw out the problems of a theory of non-natural values. One problem for the non-naturalist here is how to answer the following questions: First, why should valuation be different—have a different logical form—for humans than for other creatures? If valuation is not different, then there would need to be non-natural values for all creatures. If it is different, then the non-naturalist has to give an account how there are non-natural values, but only for humans. Either way, the non-naturalist must provide an account of how we come to have and know these values! Whatever the case, my purpose here is not to refute non-naturalist approaches, but to give a naturalistic account that better explains actual valuation.

There are many loose ends here that I cannot address, but that are clearly in need of analysis. We need to say more about how we can distinguish ethical mattering from non-ethical, and how that might make a naturalistic theory of normativity possible. Then there are a series of philosophical questions to be addressed about the nature of value relations: What sorts of things are these value relations? What metaphysical and epistemic issues, questions and problems do they raise?[21] While these questions must ultimately be answered, I cannot do so here.

V. The Other Side of Error Theory

In the introduction, I agreed with the error theory advanced by Michael Ruse. According to Ruse our ethical systems are adaptations to facilitate our social behavior, and there is nothing further to serve as absolute justification for our ethical values—our systems of normative ethics.

What I want to argue is that there are no foundations to normative ethics. If you think that to be true a claim has to refer to some particular thing or things, my claim is that in an important sense, normative ethics is false… the claims of normative ethics are like the rules of a game. In baseball, it is true that after three strikes the batter is out; but this claim does not have any reference or correspondence in absolute reality. (Ruse:1995:248-9)

So while the claim that “killing is wrong” may be true or false by reference to the ethical game (Ruse:1995:271), it is not true or false by reference to some absolute value or set of values. But, according to Ruse, when we say things like “killing is wrong” we mean it in some absolute, non-relational way. This is the error.

To use a useful if ugly word of Mackie (1979), we ‘objectify’ morality. We think that killing is wrong because it seems to us that killing is wrong. Somehow whatever the truth may be, the foundation of morality does seem to be something ‘out there’ binding on us. (Ruse:1995:254)

The explanation for this error is simple. As Ruse puts it: “We need to believe in morality, and so, thanks to our biology, we do believe in morality.” (Ruse:1995:250) The idea is that the belief in the objective foundations of normative ethics makes us more likely to follow the normative rules that we accept. If I were to believe that killing is wrong only relative to some game, rather than absolutely wrong, then I am less likely to take the prohibition of murder seriously. In short, the adaptive benefit of belief in morality is enhanced by the false belief in its absolute nature.

This belief in an absolute moral foundation is not the only error. As should be clear by now, the tendency to think of moral judgment as simple predication is also an error. The claim that “killing is wrong” is not just an error in its assumption of absolute foundation, but also in its assumed logical structure. Moral judgments are better understood relationally: “killing is wrong with respect to x, for y in context z.” This is reinforced by the fact that we recognize that killing is appropriate in many cases of self-defense. Similarly, “lying is wrong” is better formulated in one of the valuation schema I have outlined above: “lying is worse than telling the truth with respect to x for y in z.”

This logical error has implications for attributions of truth value. The cognitivism/non-cognitivism debate has focused on whether moral judgments have truth values in the ordinary sense. If my analysis is correct, then we can see why moral judgments do not have truth value in their uncorrected form, but do in the corrected logical form. This is revealed by the analogy between the predications “x is good” and “x is tall”. When we say of a person that they are tall, we seem to be predicating of him or her, the simple property of “being tall”. But what we are typically saying is that some person is taller than those in a reference class. By “x is tall”, for instance, we might mean “x is taller than the average Alabamian,” “x is taller than the median for women,” or “x is taller than the other 12 year olds.” But it is clear that there is nothing like just being tall. Tall is a comparative predicate. Consequently, the sentence “x is tall” will not have a truth value, because it is ill-formed—it lacks the other comparative term. It is equivalent to the fragment: “x is taller than…”. Here it is obvious that we cannot assign a truth value until we complete the sentence. Similarly, good and bad are relational predicates, and the fragment “w is good…” will not have a truth value until we complete the sentence: “w is good with respect to x, for y in z.” Once we complete the valuational sentence, it can have a truth value, but not before.

An adequate error theory demands an explanation for the error. I can hardly provide one here. Nonetheless, we can see how it might be useful to think of relations as simple predications. If someone is taller than some implied comparative class, we think that person to have some property of being tall. Likewise, if something tastes good, we tend to think that it has some property of “good flavor,” rather than thinking it is related to our tastes in such a way that it is pleasurable. If we find some action morally offensive, we think the action has the property of being morally offensive. This sort of strategy is useful in everyday life, because what we need to know for biological and social functioning is how things are good or bad for ourselves in general ways, and in the contexts we generally find ourselves. A fast (and dirty) way to do this is by projecting value or disvalue onto the things themselves that tend to function in certain ways relative to us.[22] The value of this strategy is obvious. Suppose we are looking for dinner and encounter a patch of “death cap” mushrooms. It may be true that such mushrooms function well in different ways for different creatures, in various contexts, but this is more than we usually need to know. What we need to know is that they would be a bad, deadly, meal for us. We are better off if we focus on that fact, than if we allow ourselves to be distracted by the many different ways “death caps” can function relative to other creatures. In other words, we can and should ignore the irrelevant relational facts. This can be done relatively easy by thinking of the mushrooms as being deadly—having a property of deadliness. This simple judgment is adequate for our purposes, and much easier to apply than the relational judgment in all its complexity. Similarly when confronted with a hungry lion, we are better off thinking that the lion is simply dangerous. The person who pauses to consider all the ways a lion is good or bad relative to various subjects and contexts, will likely be good for the lion relative to taste and nutritive benefits. By contrast, the person who incorrectly believes the simple predication “the lion is dangerous” is more likely to live another day to incorrectly believe other similar, simple predications. Similarly, it is useful to think that someone who poses a threat to me (or a group I belong to), has the property of being “bad” or “evil”. If I think of him as evil, I will have a useful attitude towards him—an attitude that will cause me to avoid him. It will also be useful to believe that someone who behaves in ways that benefit me (or a group I belong to) has the property of being “good.” We also need a fast and easy way to evaluate actions, rules and sentiments relative to human social functioning. One good way to make these quick judgments is through simple predication: certain actions are simply bad, and others are simply good. This works if the relevant actions, rules and sentiments function in sufficiently general ways relative to us and the circumstances we find ourselves.[23] In other words, it is often useful to think that good actions have the property of being good—they have “goodness”, while bad actions have “badness”.

VI. Conclusion

There are two main reasons to adopt the relational conception of value presented here. First it gets valuation right. It corrects and clarifies our conception of value, revealing its deep logical structure. This structure is apparent from the understanding that evolutionary theory gives us of valuation in general. For those who want to postulate another source of value, the burden is on them to show how it might arise.

Second, this conception is better able to explain human ethical behavior. It reveals what is subjective and what is objective in valuation. And it shows that just because value is relational to subjects, it is not therefore a matter of opinion. There can be facts about valuation, just as there can be facts about mattering. The relational conception of valuation also gives us a framework for understanding the semantics of value—how we instantiate the variables. We can consciously consider the different ways something can be good, for whom it is good and the contexts in which it is good. This highlights what is empirical in evaluation—facts about mattering, subjects and contexts. We cannot determine what is good by a priori philosophizing.

This framework also makes it clear why moral reasoning is often complicated and contentious. There is great complexity in social functioning, and it is often not obvious how a particular trait, behavior or action will matter relative to a single subject or various sets of subjects in a variety of contexts. One source of ethical disagreement is therefore to be found in the fact that there are empirical disputes about well functioning. There is also a political element: many ethical disputes are ultimately over who benefits from the functioning of particular traits, behaviors and actions. That an action or rule benefits me over you is a good reason for me to prefer it, and you to abhor it. Ethical disputes can also be about how the different ways of being good and bad get balanced. A particular action can be good in one respect but bad relative to another. Killing a child murderer may be good relative to the fact that this person is eliminated, but it may be bad in other ways. Since many, if not all, actions will be both good and bad in different respects, we typically must weigh the competing valuations, and we may disagree about how the various options balance out.

We can explain agreement on this account as well. First, as our error theory suggests, the fact that we think of value in terms of simple predication implies that we often simply suppress some of the potential sources of dispute by ignoring the relational structure of valuation, and the factors that would be expected to produce disagreement. Thinking of value in property terms hides the relational aspects of value that are likely to cause disagreement—conflict relative to different respects and subjects. But also, many things are good for groups of subjects, and are therefore good in some way for all or most of those in the group. If truth telling benefits everyone in the group, then all those in the group have a good reason to value truth telling. They will have a reason to agree.

This model also reveals the limits of ethical discourse. On the property conception of value, consensus is achievable in principle by pointing to the particular goodness or badness of a thing or action.[24] Then once we all see that it is good or bad, we can agree about its value—and decide whether it should be pursued. But if value is relative to respects, subjects and contexts, as I argue here, there will be ineliminable conflicts in value judgments. It is inevitable that there will be actions or rules that are good for me and bad for you, and nothing can change that fact. In these cases, we can arrive at a consensus only if one of us is mistaken about the valuational facts.[25]

Notice that this relational conception makes sense of Moore’s “open question” argument. Moore argued that it is always an open question whether or not some natural property like pleasure was good. He concluded that therefore we cannot simply define goodness in terms of natural properties like pleasure. But if value is relational, as I argue here, of course the goodness of any particular property will be an open question. We need to know the respect in which it is supposed to be good, and for whom and under what conditions it is supposed to be good.

There is an additional way this conception of valuation may be useful, although I can only gesture toward it here. If I am right and valuation should be conceived in relational, and not property terms, then those philosophical debates that have been premised on the property conception of value will need to be revisited in light of this fact. As already suggested, we might doubt the value or necessity of value theories that postulate supervenience. But we might also look critically at the debates about intrinsic value. If intrinsic value is conceived in property terms, then we might have something to say here as well. Whatever the case, there are, I believe, significant philosophical consequences for how we conceive value. We need to get our value concepts right.

Richard A. Richards
University of Alabama

While these views, and any errors, are my own, I have benefited greatly from my discussion of these issues with my colleagues Max Hocutt and James Otteson.


Armstrong, D. M., A Theory of Universals, Cambridge Univ. Press, Cambridge, U. K., 1978.

Christman, R. J., Sensory Experience 2nd Ed., Harper & Row Publishers, N. Y., 1979.

Collier, John and Michael Stingl, “Evolutionary Naturalism and the Objectivity of Morality,” Issues in Evolutionary Ethics, Paul Thompson (ed.), State University of New York Press, Albany, N.Y. 1995.

Coren, Stanley, Lawrence M. Ward and James T. Enns, Sensation and Perception, 6th Ed., John Wiley & Sons Inc., Hobokan, N. J., 2004.

Darwall, Stephen, Allan Gibbard, and Peter Railton, ”Toward Fin de siècle Ethics: Some Trends”, Meta-Ethics, Michael Smith (ed.), Dartmouth University Press, 1995.

Darwin, Charles, On the Origin of Species, A Facsimile of the First Edition, Harvard Univ. Press, Cambridge Ma., 1964.

Davis, Steven (ed.), Color Perception: Philosophical, Psychological, Artistic and Computational, Oxford Univ. Press, Oxford, U. K., 2000.

De Marco, Joseph P., Moral Theory, A Contemporary Overview, Jones and Bartlett Publishers, Boston, Ma., 1996.

Foot, Philippa, MoralDilemmas, Clarendon Press, Oxford, U. K., 2002.

Foot, Philippa, Natural Goodness, Clarendon Press, Oxford, U. K., 2001

Hilbert, David, Color and Color Perception: A Study in Anthropocentric Realism, Center for the Study of Language and Information, Stanford, Ca., 1987.

Hilbert, David, and Mark Eli Kalderon, “Color and the Inverted Spectrum,” Color Perception: Philosophical, Psychological, Artistic and Computational Perspectives, Steven Davis (ed.), Oxford Univ. Press, Oxford, U.K., 2000.

Hocutt, Max, Grounded Ethics:The empirical bases of normative judgments, Transaction Publ., New Brunswick, 2000.

Jackson, Frank, “Philosophizing about Color,” Color Perception: Philosophical, Psychological, Artistic and Computational Perspectives, Steven Davis (ed.), Oxford Univ. Press, Oxford, U.K., 2000.

Joyce, R. “Darwinian Ethics and Error,” Biology and Philosophy, 15:713-732, 2000.

Klagge, James, C., “An Alleged Difficulty Concerning Moral Properties,” Meta-Ethics, Michael Smith (ed.), Dartmouth Publishing Company Limited, Brookfield, 1995.

Lahti, David, C., “Parting with Illusions in Evolutionary Ethics,” Biology and Philosophy, 18:639-651, 2003.

Lemos, John, “A Defense of Darwinian Accounts of Morality,” Philosophy of the Social Sciences, Vol. 31, 3:361-385, 2001.

Lemos, Noah M., Intrinsic Value, Cambridge Univ. Press, Cambridge, U. K., 1994

Levine, Michael, W. Fundamentals of Sensation, 3rd Ed., Oxford Univ. Press, Oxford, U. K., 2000.

Lillehammer, Hallvard, “Debunking Morality: Evolutionary Naturalism and Moral Error Theory,” Biology and Philosophy, 18:567-581, 2003.

Livingstone, Margaret, Vision and Art: The Biology of Seeing, Harry N. Abrams, New York, N.Y., 2002.

Maienschein, Jane and Michael Ruse (eds.), Biology and the Foundations of Ethics, Cambridge Univ. Press, Cambridge, U. K., 1999.

Maund, Barry, “Color”, The Stanford Encyclopedia of Philosophy (Fall 2002 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2002/entries/color/>.

Maund, Barry, Colors: Their Nature and Representation, Cambridge Univ. Press, Cambridge, U. K., 1995.

Mayr, Ernst, What Evolution Is, Basic Books, New York, N.Y., 2001.

McDowell, John, “Values and Secondary Properties,” Essays on Moral Realism, Sayre-McCord (ed.), Cornell Univ. Press, Ithica, N. Y., 1988.

McShea, Robert J., and Daniel W. Mc Shea, “Biology and Value Theory,” Jane Maienschein and Michael Ruse (eds.), Biology and the Foundations of Ethics, Cambridge University Press, Cambridge, U.K., 1999.

Moore, G. E., Principia Ethica, Cambridge University Press, Cambridge, U. K., 1978.

Richards, R. J., “A Defense of Evolutionary Ethics,” Issues in Evolutionary Ethics, Paul Thompson (ed.), State University of New York Press, Albany, N.Y. 1995.

Richards, Robert, Darwin and the Emergence of Evolutionary Theories, The Univ. of Chicago Press, Chicago, Il., 1987.

Ruse, Michael, “Evolution and Ethics: The Sociobiological Approach,” Ethical Theory, Louis P. Pojman (ed.), Wadsworth Publishing Company, Belmont, Ca., 1998.

Ruse, Michael, “Evolutionary Ethics: A Phoenix Arisen,” Issues in Evolutionary Ethics, Paul Thompson (ed.), State University of New York Press, Albany, N.Y. 1995.

Ruse, Michael, Evolutionary Naturalism, Routledge, New York, N. Y., 1995.

Ruse, Michael, and Edward O. Wilson, “Moral Philosophy as Applied Science,” Philosophy, 61:173-192, 1986.

Rottschaefer, William A., and David Martinsen, “Really Taking Darwin Seriously: An Alternative to Michael Ruse’s Darwinian Metaethics,” Issues in Evolutionary Ethics, Paul Thompson (ed.), State University of New York Press, Albany, N.Y. 1995.

Savallos, Elias E., and Umit D. Malcin (eds.), Supervenience, Cambridge Univ. Press, Cambridge, U. K., 1995.

Sayre-McCord, Geoffrey (ed.), Essays on Moral Realism, Cornell Univ. Press, Ithaca, N. Y., 1988.

Sober, Elliott, “Prospects for an Evolutionary Ethics,” Ethical Theory, Louis P. Pojman (ed.), Wadsworth Publishing Company, Belmont, Ca., 1998.

Thompson, Evan, Color Vision, Routledge, London, U. K., 1995.

Thompson, Evan, “Comparative Color vision: Quality Space and Visual Ecology,” Color Perception: Philosophical, Psychological, Artistic and Computational Perspectives, Steven Davis (ed.), Oxford Univ. Press, Oxford, U.K., 2000.

Thompson, Paul, ed. Issues in Evolutionary Ethics, State Univ. of New York Press, Albany, N.Y. 1995.

Woolcock, Peter G., “The Case Against Evolutionary Ethics Today,” Jane Maienschein and Michael Ruse (eds.), Biology and the Foundations of Ethics, Cambridge Univ. Press, Cambridge, U. K., 1999.

[1]. See also Ruse and Wilson, (1986) Ruse (1986a, 1986b) and Joyce (2000).

[2]. Sometimes Rottschaefer and Martinsen describe values as relations, which would seem to be more in conformity with the view that I will be advocating, but they are unclear what the relation is supposed to be, and often they seem to interpret it as a relation between properties—supervenience. Whatever the case, mostly they identify values as properties.

[3]. See Darwall, Gibbard and Railton for a survey of metaethical approaches that indicates the range of thinkers who draw the analogy between color and value, and conceive of values as properties,

[4]. Some see color as purely subjective. In a recent textbook on perception, Michael W. Levine argues: “As real as color seems, it is, in fact, a purely psychological phenomenon. Light rays are not colored; they are radiations of electromagnetic energy of different wavelengths. The attribute called ‘color’ is entirely a fabrication of your visual system. (Levine: 297)

[5]. Thompson explains: “Thus I hold that color, rather than being distal in the physicalist sense, is relational: having a surface color is equivalent to having a SSR that belongs to a certain chromatic equivalence class for a particular perceiver in its photic environment.” (Thompson:2000:179)

[6]. It is beyond the scope of this paper to say how supervenience in all its versions is inadequate. That would be a daunting project. What I hope to do here is make an end run around all the philosophical contortions designed to make supervenience adequate, and show that the relational approach does the job.

[7]. This seems to be the strategy first adopted by Galileo, Boyle and Locke. See Hilbert:4-5.

[8]. We are not, for instance, tempted to start with Platonic, Kantian or religious presuppositions about what value must be like, based on their respective ways of misunderstanding human nature.

[9]. I am following Peter Railton here in his account of ethical value. As he explains, valuation is grounded on “mattering”. “It seems to me that notions like good and bad have a place in the scheme of things only in virtue of facts about what matters, or could matter, to beings for whom it is possible that something could matter.” (Railton:2003:47)

[10]. This is an insight derived from Hocutt, who identifies the good with that which reinforces. (Hocutt:2000:113-134)

[11]. This kind of generality is most obvious in kin selection and coevolution.

[12]. In Richards (1997), I argue that Darwin thought that the flaws produced in domestic breeding could be used as experimental evidence for the functioning of natural selection in eliminating the unfit—the worse forms. Darwin thought we observed this disvalue in the browsing of the highly bred Niata cattle of La Plata. In effect, the value of domestic breeding was partly found in the fact that it tended to produce bad traits in domestic breeds.

[13]. See Hocutt:2000 for one promising approach that is consistent with this evolutionary account of valuation, but is not itself based on evolution.

[14]. Notice this redistribution may be bad in other respects relative to the poor as well—if it promotes dependence or discourages individual initiative.

[15]. How an entire species could be benefited is a bit unclear, but possible in principle nonetheless. What is of consequence here is that there can be different degrees of “egoism”: if an action benefits only me then it is most egoistic, but if it benefits my family, community or species as well, it decreasingly egoistic- even though it still benefits me as part of the group.

[16]. These generalizations can be useful not just in formulating moral rules of thumb, but also in underwriting predictions about social behavior. People tend to act in ways that seem to be good for them in particular ways and contexts.

[17]. Max Hocutt, although he focuses on normativity rather than value, advocates a similarly relational theory of value in his Grounded Ethics (2000) Like Railton’s, his interests are narrower than those here. He focuses on human value from within a behavioral and cultural context. Goodness, for him, is that which reinforces, and is always dependent on “desire”. (Hocutt:6) While his views, like Railton’s, could be made consistent with the more general approach developed here, it is nonetheless, narrower by virtue of a non evolutionary starting point.

[18]. Railton is unusual in postulating a reducible supervenience. Most accounts of supervenience postulate nonreducibility. Savallos and Yalcin claim that the basic notion of supervenience involves covariance, dependence, determination and non-reducibility. After Jaegwon Kim, the standard view is that supervenience is a relation between properties. (Savallos:1995:2)

[19]. Railton argues that goodness is not so much like color, but taste. See his “Red, Bitter Good” in Railton:2003.

[20]. Elliott Sober for instance: “…So called ethical relativists say that murder is right or wrong only because various people in a society have come to believe it is.” (Sober:135)

[21]. David Armstrong raises some of these questions, but recognizes that there has been insufficient philosophical work done on them. (Armstrong:1978:75)

[22]. Similarly, a fast and easy way to make color judgments is by projecting the color onto a thing—thinking of it as a property of that thing.

[23]. If we cannot generalize mattering in this way, then we would be less likely to make these property attributions. In other words, if an action is sometimes good and sometimes bad in common contexts, then we don’t tend to thing of it as either being good or bad.

[24]. This hope does not demand that we be able to recognize or perceive goodness directly. We might be able to do this indirectly by pointing to the properties that the goodness or badness depend on.

[25]. The obvious mistake is to think that the action or rule or thing has some property that makes it good in a non-relational way. We can incorporate this possibility into our error theory: the property conception of value can facilitate social functioning by causing us to believe that something is good in itself and therefore good for everyone.