1.

Different philosophers define relativism in different ways, though the definitions are (so to speak) near relatives. James Rachels, for example, after discussing a number of different theses that are often lumped together under the heading of relativism, states that moral relativism is the view that ‘there is no measure of right and wrong other than the standards of one's society’,Footnote ¹ while Louis Pojman says that the relativist holds that ‘there are no universally valid moral principles, but rather… all moral principles are valid relative to culture or individual choice.’Footnote ² In a similar vein, C.E. Harris takes relativism to be the view that ‘moral beliefs are true only relative to principles held by some group’.Footnote ³ More formally, Bernard Rosen distinguishes direct from indirect relativism, with direct relativism being:

and indirect relativism being:

Focusing on what he calls ‘Protagorean Relativism’, Emmett Barcalow says that, according to relativism, ‘Each society or culture has its own moral code and whatever a society's moral code says is right and wrong is right and wrong for that society.’Footnote ⁵

2.

These sample formulations of relativism could be multiplied by several orders of magnitude without exhausting the definitions found in the literature, and even the few quoted aren't free from difficulty, or equivalent to one other. Rather than stop to survey and critique definitions of relativism, however, I'd like to extract what I take to be the root notion.

As I understand it, relativism is the view that there are two or more equally valid or true but conflicting moral codes. On the one hand, relativism contrasts with universalism, the view that there is a single true or valid moral code, binding on all people at all times. On the other hand, it contrasts with nihilism, the view that there isn't even a single true or valid moral code.

One consequence of this definition is that, contrary to much of the literature, relativism shouldn't be contrasted with absolutism. An absolute moral rule or principle, at least in my parlance and that of many philosophers, is one that holds without exception. Conceptually speaking, an absolute rule (or principle, or value) contrasts with a prima facie rule, a rule that holds not without exception but only other things being equal. The absolute/prima facie distinction functions within normative ethics, and indicates the scope, force, and weight of a rule: it functions as a qualifying distinction, implicit or explicit, about how the rule is to be applied. To illustrate: one question that might be asked about a rule, such as ‘One ought not to lie’, is whether it holds absolutely or only prima facie. As is well known, Kant says that it holds absolutely, and Ross only prima facie. But in saying so, neither has indulged in meta-ethics; both have stayed within the confines of normative ethics, and indicated the force, scope, or weight (the exceptionless or exception-possible character) of the rule. Whether relativism is true, however, is a question about normative ethics, not a question within normative ethics. It's a question of meta-ethics, in other words. Reinforcing the point is not just the traditional classification of relativism under meta-ethics, but the fact that a relativist qua relativist isn't committed to any normative theory, or any specific normative rule or principle at all, and certainly not to whether any moral principles or rules are absolute or prima facie. In effect, he tells us something about the number of equally true or valid moralities or moral codes, but nothing about which one he subscribes to or is bound by, or what its structure or content is. A relativist could thus hold to absolute moral principles, for the valid ethical code he's bound by could include such principles, and he could know as much. In short, then, relativism as such entails no normative principle or concept at all.

Moral relativism is also distinct from moral pluralism. Valuational pluralism contrasts with valuational monism, and the two theories concern whether all the values/principles within a given domain, such as morality, are or can be reduced to one, or whether a multiplicity of independent and irreducible values/principles is needed. Like the absolute/prima facie distinction, the pluralist/monist distinction is a distinction within normative ethics, and independent of the universalist/relativist/nihilist distinction. A pluralist could be a universalist, a relativist, or a nihilist, and the same holds for a monist.

3.

But where, it might be asked, is the relativism in the definition of relativism offered above? Nothing in the definition even implicitly speaks of anything being relative to anything, yet the doctrine is, after all, known as relativism. The definition, this objection concludes, is Hamlet without the Prince of Denmark.

True enough, there is nothing relativistic in the definition of relativism. But perhaps paradoxically and certainly counterintuitively, and the label of the theory notwithstanding, the relativistic aspect of relativism needn't and shouldn't show up in the definition of the term itself. It should and does show up in the near vicinity, though, namely, in the supposed ground for there being a number of equally true or valid moral codes. It there is a variety of true but conflicting moralities, that must be because moral truth is grounded relative to something – a culture, a set of agreements, a location, The Empire State Building, a framework – and (a) not every valid code bears the same relation to that something, or (b) every valid code bears the same relation to a different something of that kind. For instance, one possible ground for moral relativism is that the truth of a moral code is a function of distance from the equator: different moral codes are true or binding depending on location relative to zero degrees latitude. One code would be valid and binding from 0 to 500 miles from the equator, another from 500 miles to 1,000 miles, and so on. (This is obviously an example of an (a)-based relativism. Much more common are (b)-based relativisms.) On the view in question, the truth of a moral judgment would vary in much the same way that the truth of a temperature reading does, or an assessment of climate does, namely, as roughly a function of distance from a certain latitude. (Of course, that's not quite the way that temperature or climate varies, but the point is clear enough.) I'm not saying that such a view is true or even remotely plausible, only that it's relativistic. The essence of relativism is that there are a number of equally true or valid moral codes, but there does, after all, have to be a basis, metaphysically speaking, for multiple moral truth. It's there, in the grounding of multiple moral truth, and not in the definition of relativism, that the relativistic element of relativism should come in.

Excluding a relativistic component from the definition of relativism itself, but insisting that relativism entail multiple moral truth, also has the advantage of precluding certain intuitively non-relativistic views from being classified as forms of relativism. The view that moral truth is grounded relative to what God commands, for example, or that moral truth is grounded relative to what the majority of people in the world believe is right or wrong at time t, is intuitively not relativistic. Both views, however, make moral truth relative to something, in fact a subjective something in both cases (assuming that God's commands reflect his mental states). Neither view generates a multiplicity of true moral codes, as relativism, at least intuitively, should.

A final and practical advantage of the definition advanced is that it draws an exclusive and exhaustive distinction of meta-ethical significance, that between universalism, relativism, and nihilism. One, many, or none – that's a simple and important question concerning moral truth. An equally simple and important, but different, question is whether the metaphysical basis of moral truth is or isn't wants, wishes, desires, agreements, cultural traditions, and/or so on. But that I take to be the question not of relativism, but of the subjectivity or objectivity of moral truth – the question of whether the ultimate ground of moral truth is or isn't human mental states.

Note that, even apart from the heterodox claim that nothing relativistic should figure in the definition of relativism, it's a mistake to define relativism (as some of the definitions quoted in section I do) in terms of moral truth being a function of the beliefs of a person or a group, or the standards of a society, or any such subjectively based state of affairs. While any plausible form of relativism would have to ground moral truth in some such way, as the equator example of three paragraphs back shows, relativism could be based on something entirely independent of personal or societal beliefs, agreements, attitudes, standards, or whatever. Distance from the moon would do just as well as anything else, as far as being able to generate multiple moral truth is concerned. Relativism is thus compatible with objectivism, for it's compatible with the view that moral truth, although multiple, is not identical with, a function of, or dependent on any subjective state of affairs at all. That said, I want to stress what was mentioned three sentences back, that any form of relativism that isn't wildly counterintuitive will be subjectivist, just as any plausible form of consequentialism will insist on maximizing the good for a (potentially) very large number of people. ‘Everyone always ought to act so as to maximize Kate Moss's happiness’, while a consequentialist theory, is so implausible as not to be worth considering. So is equator-relativism.

It's subjective relativism – more precisely, subjectively-based relativism – that's center stage, then, and culture-based or society-based forms of subjective relativism more than anything else. Culture-based or society-based relativisms take moral truth to be underpinned by agreements between individuals, shared beliefs, shared normative frameworks, and cultural traditions. But also of some interest is individual-based relativism, relativism which take moral truth to be underpinned by, and relativized to, each individual's beliefs, desires, or normative framework. On individual relativism, each individual has his/her own equally valid or true moral code.

4.

What of the fact that, on the proposed definition, the equally valid or true moral codes conflict, meaning that at least at some points yield inconsistent judgments. How is that possible? If the relativization is duly spelled out, or even tacitly assumed, conflict (or at least one important kind of conflict) would seem to evaporate: ‘A is right relative to moral code X’ isn't inconsistent with ‘A is wrong relative to moral code Y’, any more that ‘B is legal relative to legal code M’ and ‘B is illegal relative to legal code N’. Once the relevant points of relativization, the relevant (subjectively-based) moral frameworks, are specified, the inconsistency vanishes.

This is a problem with the definition of relativism – and others’ definitions as well as my own – that has sometimes briefly addressed but is surely in need of an extended and convincing response. One response, or so I take it to be, is Mackie's: as normally employed, the concepts of right and wrong are not relativized but contain a claim to objectivity, a claim to obtain independent of human minds. As such, two seemingly conflicting judgments, one to the effect that A is right, the other to the effect that A is wrong, do indeed conflict, since they ascribe incompatible objective properties to one and the same action. Building a claim to objectivity into moral judgments thus saves moral conflict; moral disputants are indeed making inconsistent claims. However, since there are no objective values, at least if values are subjectively based, the view in question entails that all value judgments, including moral ones, are false.Footnote ⁶ That implication, unfortunately, sinks the ship. Universal falsity is highly counterintuitive, plays havoc with moral reasoning, and has numerous other untoward consequences.Footnote ⁷

A less radical view that saves the appearances is partly negative and partly positive. The negative component is that neither a claim to objectivity nor relevant relativization is part, not even a tacit part, of moral judgment as such. In a moral debate, ‘A is right’ and ‘A is wrong’ are to be understood as they stand and as moral disputants themselves understand them: as simple, categorical claims of right and wrong, just as ‘This is red’ and ‘This is not red’ are understood in a color dispute. Relevant relativization makes for truth or falsity, but, like objectivity, is not part of what is claimed. Disputants are claiming that A is right and that A is wrong, and each understands that if he is correct, then the other person is wrong; and vice versa. Within morality, and regardless of morality's metaphysical basis, the claims are logically incompatible, and disputants, in using the appropriate terms, understand them as such, and understand their adversaries to be making incompatible claims. They do so without any commitment respecting, or any view on, either the metaphysical nature of right and wrong (as qualities or relational properties) or the metaphysical grounds of moral truth (as objective or subjective). In fact, except for certain select philosophers, no one who engages in moral dispute ever has a thought on such matters; and even meta-ethicists bracket their meta-ethical views when engaged in normative dispute. Hence disputants, whether they are bound by one and the same moral code or two different codes, take their claims to be conflicting, and they do in fact conflict, even if, if truth be told – assuming subjectively-based relativism to be true – it's possible that they both are true. Dispute requires only a shared network of basic moral concepts and conceptual relations, and at least some agreement on the nature and bounds of moral argument, but not a shared metaphysics, much less a true metaphysics.

Logically speaking, the way to understand relativization is not as part of the content – explicit, implicit, or parenthetical – of moral judgment, but as an index on moral judgment as a whole. ‘This is right’ is strictly speaking ‘This is right’₁, and ‘This is wrong’ is, assuming a different framework, ‘This is wrong’_2. The indices mark different moral frameworks and indicate that proper relativization is to the marked framework. (I take a moral framework to be the more prominent contents and abstract elements of a moral code – that is, a morality – properly interrelated.) For two reasons, moral disputants can, do, and, in most cases, should ignore ‘correct logical form’. First, in the great majority of cases there is either no or little difference in the moral codes and frameworks of disputants. Considerable overlap is common, and enough so, in general, to drop indexed relativization. In fact, whether moral relativism is true or not, social life would be impossible unless there were considerable overlap in the moral codes that people actually employ. That means that in everyday life, we pay no attention to indices, even if moral judgment is indexed and moral truth framework dependent. Second, key moral terms, such a ‘good’, ‘bad’, ‘right’, ‘wrong’, and so on, are univocal. The fact that they retain whatever meaning and import they have across moral codes and frameworks, in a context of ignored relativization, makes possible moral reasoning, moral dispute, moral agreement, and even – something few philosophers are acquainted with – changing one's mind.

This may seem an odd, and quite possibly a weak, defense of the possibility of conflicting judgments, but consider, as a point of comparison, conflicting judgments about color or shape. ‘That's red’, ‘That's not red’, ‘That's circular’, and ‘That's not circular’ can be and are made in bona fide disputes, and do conflict. Only a shared vocabulary and a shared conceptual network are needed to account for the possibility of such disagreement, however, at least as disagreement is normally understood. If Locke is right, colors are subjective (mind-dependent) and relative (relational) qualities, and shapes objective (mind-independent) and non-relative; if Berkeley is right, both are subjective and relative. But no matter, as far as disputes of both sorts are concerned. Metaphysical views and metaphysical truth respecting shapes and colors are no part of and do not inform the nature of the disputes. Metaphysics, at least until the time when the truth is known and accepted by all, is irrelevant, and makes disagreement neither possible nor impossible.

But to return to normative dispute: conflict isn't preserved at the level of first-order normative dispute by occupying a contingent ‘field position’ of obliviousness to the metaphysics of morality, employment of a work-a-day and conceptually simple understanding of right and wrong, and argumentation of a standard normative sort. That field position is, at least in normal circumstances, required. Actual conflict requires commitment to the topic at the level at which the conflict occurs. In the case of standard normative assessment, the topic is action A, and the question is whether A is right or wrong. To ascend to a meta-level is to commit to a different topic, that of rightness and wrongness, and to ask a different question, whether rightness and wrongness is objective or subjective, and non-relational or relational. Perhaps answers to the latter question can and should affect answers to the former question in a fully formed and satisfactory theory of ethics – something which certainly isn't on the horizon – though in fact, at least as far as I've been able to ascertain, not even philosophers confident in their meta-ethics invoke their meta-ethical views in addressing substantive moral issues. But even if metaphysics is relevant to first-order moral disagreement, it would seem that its relevance is minimal. Moral reasoning proceeds quite successfully without it, and has long done so. If metaphysical premises add to normative reasoning in a constructive way, they can't add much without themselves becoming suspect, at least if established forms, patterns, and contents of moral reasoning are essentially sound. In sum, then, for all intents and purposes – all practical intents and purposes, all first-order intents and purposes – meta-ethical issues, including metaphysical ones, are well-nigh necessarily put to the side in normative dispute.

Bolstering this fundamentally negative and deflationary defense of the possibility of conflicting moral judgments is a positive consideration. Even if ethical values are not overriding, a position held by many, they are at least extremely weighty, and supply a rational order for human life in the grand scheme of things, an order that a purely empirical or a normative but non-moral take on life can't supply. It's partly for that reason that moral judgments, or at least those concerning core values, are normally understood by adherents to have universal jurisdiction, and to be of a particularly binding nature (which is not to say that they take them to be objective – as argued above, no stance is taken on that issue). ‘Killing for no reason is immoral’ is of a different force and existential significance than ‘Killing for no reason is illegal’. The fact that moral claims are taken to have no spatial or temporal borders is a reflection of their paramount significance and place in human life; and the fact that disputants can argue using the concepts of right and wrong simpliciter is a reflection of their felt universal jurisdiction.

5.

Another important concept that figures in my definition is that of a moral code. As I use the term, a moral code comprises more than rules and principles, which are all that philosophers typically focus on. A moral code also contains ideals, virtues, goals, aims, and values, and, at least ideally, integrates all such components into a unified, workable whole, a ‘way of life’. Order, structure, coordination, and integration, in other words, are important features of a moral code, as is an explanation and justification of both the content and structure of the code. ‘Stealing is wrong’, for example, might be a rule that figures in a moral code, and its relation to other rules of the code, such as ‘One ought to take care of one's family’, should be explained and justified. To consider just one instance, one question that could be asked about the two is, Which rule takes precedence if the two conflict? So, too, should the circumstances for the application and utilization of a code be made clear. We can ask, for example, Does the second rule require taking care of one's family at all times, in all circumstances? If not, when doesn't it, and why? And regardless of the answers to those questions, What should such care consist in? This conception of a moral code is my own, and it's clearly an idealization of the moral codes we actually work with. Still, it does seem to be the basic notion of a morality that people have in mind when they assert or deny that relativism is true.

To this specification of a moral code, and consequently to my explanation of relativism, I need to add what has become a commonplace in the literature: that, for relativism to be true, the differences between conflicting but equally valid moral codes not be due to mere differences in the objective circumstances or the factual beliefs of adherents of the codes. Nothing technical or theory-laden is meant by ‘objective circumstances’ or ‘factual beliefs’ here. Objective circumstances are just facts about the world. That apples grow on trees is an ‘objective circumstance’. And factual beliefs are empirical beliefs, or, more accurately (so as not to restrict evidence or beg any questions), non-normative beliefs. If one person believes that restaurants ought not to serve fish, while another that there's nothing wrong with serving fish, the difference may reflect a mere difference in factual circumstances: fish might be poisonous in one region but not the other. No fundamental difference between the codes would then be reflected by the conflicting rules. Or fish might be poisonous in neither region, but people living in the first might believe, perhaps for good reason, that they are. Again, no fundamental difference between the codes would be reflected by the different but conflicting rules. Perhaps more to the point, in one group a number of moral rules might be predicated upon belief in God and an afterlife of a certain sort, while in another such beliefs might be wholly absent, and so many of the rules of the first group absent as well. Otherwise, though, the two groups might share the same environmental circumstances and underlying moral principles, values, ideals, and so on, and even the same factual beliefs. If the codes the groups subscribe to are otherwise identical, or even if they would be if both groups lived under the same conditions and held the same non-normative beliefs, then their codes don't count as conflicting, or even, in an important sense, as two different codes at all.Footnote ⁸

An implication of this, and one that universalists are quick to pick up on, is that the moral codes that people actually subscribe to may not be, or simply are not, nearly as diverse as many relativists claim.Footnote ⁹ This is important, universalists think, because relativists invariably cite the wide diversity of moral codes that people subscribe to as evidence for the truth of their theory. Their premise is flawed if universalists are correct on this point, and thus the common relativist argument from the diversity of morals to the truth of relativism is deprived of much of its strength.

But to return to the issue of the definition of relativism proper: I hasten to add the material of two paragraphs back, and it is part of the conventional wisdom respecting the specification of relativism, but I have my doubts about whether all differences in objective circumstances or factual beliefs that have an impact upon moral codes should be discounted, artificially nullified, or counterfactually smoothed over before two moral codes are counted as conflicting, or as more than one in number. Taking such a methodological dictum seriously would generate a massive problem of determining the truth values of many, many difficult-to-evaluate counterfactuals regarding what widely diverse and culturally distant people would believe under all sorts of conditions – something that no one, relativist, nihilist, or universalist, has ever done more than gesture at. Even more importantly, if such counterfactual levelling were effected, the probable result would be that the ranks (number) of moral codes would be so drastically reduced, and so much of their contents would be eliminated, and what would remain would be so emaciated, that relativism would lose much of its pre-philosophical, and probably philosophical, interest. Brandt, for example, after waving his hand in the direction of taking different circumstances and non-normative beliefs into consideration, reduces the differences between the ethical code of the Hopi and that of contemporary America to a rule respecting animal suffering.Footnote ¹⁰ The vast complexity of the Hopi way of life is lost, and so is that of contemporary America. Truly different moral codes in effect dissolve in a counterfactual solvent. That's why I think that although some levelling principle is needed, something weaker than unrestricted counterfactual bulldozing is desirable. What that weaker principle should be, I don't know. I suspect that virtually any restricted principle would have a limited usefulness, because groups, cultures, and so on differ so radically from each other in their circumstances and in their empirical, theological, and metaphysical beliefs. That suggests that the matter might be better handled on a case-by-case basis, with principles playing only a limited role. The matter merits discussion. In the meantime, as far as the narrow aim of this paper is concerned, it's possible to minimize the impact of the problem, and operate on the level of ‘informed intuition’ when conceptualizing different and conflicting moral codes. Informed intuition may be intuition, but by being informed with the general sorts of considerations of two paragraphs back, some needed caution will be exercised in counting moral codes as different or not.

6.

On subjective relativism, morality is, at least in substantial part, conventional. The notion of a convention thus needs at least some discussion, even if space limitations permit no more than a brief critical survey.

Drawing upon Hume, David Lewis has developed a game-theoretical conception of convention, according to which a convention is a self-reinforcing solution to a recurrent co-ordination problem faced by two or more conceptually sophisticated agents motivated by self-interest.Footnote ¹¹ In essence, a convention is a somewhat arbitrary behavioral regularity that participants create and find satisfying. A number of conditions respecting a regularity must be met for it to count as a convention, however. They include near uniformity of behavioral response among participants (i.e., conformity to the regularity), near uniformity of preference for that shared response, near uniformity of expectation of conformity, and a number of other conditions, some of which require fairly complicated knowledge among participants. Lewis's theory has been much discussed and criticized on a wide variety of grounds, but this is not the place for a detailed examination. Suffice it to say that, for my part, I think what he has identified is what a convention can give rise to, or what can give rise to a convention, but not what a convention is.Footnote ¹²

Taking a very different tack, Ruth Millikan thinks of a convention as a pattern of behavior that is reproduced within a population.Footnote ¹³ Not every reproduced pattern is a convention, however. For example, if a person sees another person turning on car lights by flipping a switch, reproduces that response, and passes on that pattern of behavior to others, no convention is thereby established. Rather, to count as a convention a pattern has to be reproduced and passed on simply because it sets a precedent. In other words, the pattern is followed, not necessarily for any good reason, but simply because it catches on and spreads. The fact that conventions, such as nodding one's head up and down to indicate agreement, are normally taken to be arbitrary is well captured by this aspect of Millikan's theory – but perhaps too much so. If Lewis's game-theoretic theory of convention over-intellectualizes conventions and demands too much of participants, Millikan's theory seems to require too little, for mindlessly repeated behavior, having no discernible point, would count as conventions. Thumb twiddling that begins with one student and unconsciously spreads across the university would count as a convention instead of, what I take it to be, a mere (and bad) habit that has irrationally caught on.

A view closer to my own has been advanced by Margaret Gilbert.Footnote ¹⁴ Focusing on the notion of a social convention, not a convention simpliciter, Gilbert notes that individual agents can and do frequently ‘join forces’ to achieve a common goal. In some cases when they do so, they form a ‘plural subject’, with individual agents sharing, and jointly knowing they are sharing, beliefs, desires, and goals in commitment with each other, and in commitment to each other. This unique cognitive, conative, valuational, and social bonding makes for a plural subject, a ‘we’ of social action, and undergirds the notion of a social convention. ‘Our everyday concept of a social convention’, she says, ‘is that of an accepted principle of action, a group fiat with respect to how one is to act in certain situations’.Footnote ¹⁵ The fiat in question is not that of any group, however, but that of a plural subject, and a plural subject ‘speaking’ as a plural subject.

Neither as demanding as Lewis's highly rationalistic theory nor as lax as Millikan's non-rationalistic one, Gilbert's theory occupies a nice middle ground. Although obviously idealized and focused on a social convention,Footnote ¹⁶ it correctly identifies what I take to be the basic notion of a convention: not a game-theoretic regularity or a precedent-determined repeated pattern of behavior, but a rule, principle, or norm of conduct or understanding ultimately based on choice and decision.Footnote ¹⁷ I would distinguish a convention from both a command and a fiat, however. I take

1. (a) conventions to be inherently normative, and paradigmatically of the form ‘This ought to be done/is good’ or ‘This ought to be understood as…’Footnote ¹⁸ (more on the latter in a moment), but

2. (b) commands to be inherently not normative, and paradigmatically of the form ‘Do this’ or ‘Understand this as…,’ and

3. (c) fiats to straddle the conceptual ground between the two, with one foot on the non-valuational side of the divide and the other on the valuational, and to be paradigmatically of the form ‘This is to be done’ and ‘This is to be understood as…’Footnote ¹⁹

Conceptually speaking, conventions are based on fiats, which are themselves based on commands, stated or tacit. Understood as explanatory underpinning or reconstruction, not as historical progression, commands beget fiats and fiats beget conventions through an uptake of normative force, a ‘taking’ that moves from ‘Do this’ or ‘Understand this as…’ to ‘This is to be done’ or ‘This is to be understood as…’ to ‘This ought to be done’ or ‘This ought to be understood as…’ Not surprisingly, parallel to the ascent to normativity is a grammatical progression from imperatival to quasi-imperatival/quasi-indicative to indicative form. Commands, fiats, and conventions all speak of doing certain things in certain situations or understanding certain things in a certain way, with the latter being naturally expressed as an understanding that. In America there is an operative convention to place the fork on the left side of the plate, but also a convention of understanding that pieces of paper produced and issued in a certain way, and having certain distinct features, are ten-dollar bills. Conventional propositions of a universal sort – for example, ‘Pieces of paper produced and issued are…’ – are the indicative sentence read-out of universal ‘understand that’ rules – basically, what others have called constitutive rules. They are, in a sense, a priori truths created by us, but also a priori truths capable of being destroyed by us, or dying deaths of neglect. Singular conventional propositions – for example, ‘This is a ten dollar bill’ – record conventional facts that are, metaphysically speaking, the result of universal conventional facts (and, derivatively, other singular conventional facts) ‘infiltrating’, ‘interpenetrating’, ‘commingling with’, or – choose your own metaphor – non-conventional reality. They, too, can suffer deliberate annihilation or pass out of existence through abandonment. All of this can also be said of conventions respecting doing certain things. The two sorts of convention, in fact, are not really as distinct as the above may make them appear, for understanding is a (special) sort of doing, and doing – performing conventional actions – requires conventional understanding (except in derivative, degenerate, or ‘courtesy attribution’ cases).

But besides all this, there's the question of, to put it in older terms, the ontological foundation of conventions. Millikan's thinks that they're based on human biology, and apparently as much to be expected as whistling. Hume, Lewis, and Harman found conventions on intelligent self-interest, with intelligence making possible the conceptually sophisticated co-operation needed to serve self-interest. Gilbert, as I read her, essentially follows Aristotle in taking human beings to be by nature social animals, in a relatively strong sense of term. Millikan, Hume, Harman, and Gilbert all invoke aspects of human nature, though different ones, to explain the existence of conventions. I do too.Footnote ²⁰ (In fact, as far as I can tell there is nowhere else to go for an explanation.) On my view, conventions are inherently normative, and are so because there is normative uptake of a conceptual sort from commands to fiats to conventions. For me, then, the question is: What is the basis of this normative uptake, this progressive insinuation of value into the empirical realm? My answer is that normative uptake exists simply because it's natural for us to see and experience the world normatively. Normativity is as natural to us as hunger, self-interest, parental affection, sympathy, seeking the approval of others, and language. It's who we are, and how we take the world, and not reducible to biological function or clever self-interest. Language well illustrates the point. Multiple and deeply conventional, language is acquired at about the same age as walking upright is, and both are to be expected, because both are based on native human potentialities and tendencies. To paraphrase Aristotle, human beings are by nature conventional animals. And, on my view, it is because human beings are by nature conventional animals that they are by nature social animals.Footnote ²¹

7.

Concrete illustration may endow these abstract remarks on the notion of convention with more meaning. Consider spelling, then, which is certainly conventional.

Spelling has evolved over the years. Four hundred years ago, many words could be spelled – meaning correctly spelled – in several different ways. (Shakespeare, in fact, never spelled his own name the same way twice.) The conventions of spelling weren't yet established, or at least not established as hard and fast rules. ‘H-a-r-r-d’ would do for ‘hard’, but so would ‘h-a-r-d-e’ or ‘h-a-r-d’. ‘S-l-e-e-t-h’, however, would not do. Eventually, the conventions of spelling became firmer, narrower, and better established, and now only one spelling is acceptable. But there was never, not at any time over the past four hundred years, a vote or a poll taken on the correct spelling of ‘hard’, nor was there ever even a thought of doing so or of soliciting the opinions of speakers of the English language. The convention evolved slowly, in a very informal manner, and in a way that followed no preordained course. However, although informal, the evolution of the conventions of spelling was guided by and responsive to a number of factors, just as all conventions are. Foremost among them were:

1. (a) historical precedent and continuity with the past – drawing upon the past and not breaking sharply with it, even when modifying prior spelling;

2. (b) consistency;

3. (c) coherence;

4. (d) integration with other conventions of spelling and pronunciation – a system of interrelated and interlocked conventions is needed, not just conventions for single words, and such a system requires a smooth meshing of parts;

5. (e) the actual spelling of participant spellers – the empirical input of users of the system of spelling;

6. (f) the empirical limitations and actual desires of human beings, such as the facts that human beings have a finite memory, a desire for simplicity, and limited time, energy, and interest; and

7. (g) historical contingency, for example, the spread of literacy and development of public and private school systems.

In the background, and guiding the whole informal and (almost) unconscious process was nothing more metaphysically solid than the idea of human convenience: we, a not very clearly defined group of speakers of the English language existing over centuries and living in different locales, need to tack down standardized spelling for ‘hard’ and other words in order to run our affairs in as efficient and trouble-free a way as possible. The conventions of correct spelling didn't evolve completely independently of what speakers of the English language took to be correct spelling, or believed to be correct spelling – I doubt that that would have been possible – but such empirical input is but one factor among many, and a vote on correct spelling or even anything like a survey on it, would have been extremely impractical (in fact, almost impossible to conduct), largely a waste of everyone's time, and unneeded anyway. Norms could be and were established through other mechanisms, and without paying undue attention to what the majority thought. Once established, such norms can easily survive and even prevail over majority opinion to the contrary. ‘Accommodate’ (and its adjectival and other grammatical kin) is one of the most frequently misspelled words in the English language; it's misspelled far more often that it's correctly spelled. If simple majority rule determined correctness of spelling, that would be a contradiction in terms. It isn't precisely because a norm has been established informally, and it determines correct spelling regardless of what the majority thinks.

8.

In sum, that's the somewhat different conception of relativism that I defend. With expected author's pride, I think that it's clearer and more useful than many others on the market, and that it will help to facilitate discussion of arguments for and against relativism, as well as the inevitable objections to both.Footnote ²² But even if I'm wrong about that, it's still important to call attention to the necessity of being clear about exactly what relativism is.Footnote ²³